A new era of advertising networks is dawning, and even in-car banner advertising is becoming a reality. Recently, BMW placed a promotion for the movie “Spider-Man” on the dashboards of its cars. When the engine started, drivers saw a banner, clicking on which activated a 19-second animation with music and a light show. The reaction of owners was mixed.
Many did not appreciate the advertising content in an already paid-for car, especially considering last year’s statement by a top BMW executive that the car’s interior is a “private space” not intended for advertising. However, from the perspective of a digital arbitrageur, this only confirms the theory: “everything around us is an advertising network.”
Why are cars turning into advertising platforms?
The logic is simple: any company that controls enough attention has a basis for a media business. Primary data about users (what they watched, searched for, bought, where they traveled) increases the value of this attention, making it targeted. The car is a natural extension of this trend:
Modern cars are equipped with connected digital interfaces.
Screens are getting bigger, and software is updated remotely.
Automakers have direct and long-term relationships with customers.
The potential grows when you consider that autonomous driving will free up even more driver attention. This is mass reach for pennies, if approached with proper analytics.
Transparent statistics and ROI: key questions
BMW and Sony probably know how many cars received the “Spider-Man” banner and how many drivers clicked on it. These metrics describe engagement, but they don’t answer the main question: did this activation lead to movie ticket purchases? Did it create additional value, and how does its contribution relate to the rest of the media campaign?
“Until the industry can answer these questions with the same rigor as in other media plans, ‘Spider-Man’ on the dashboard remains an experiment. As soon as it can, the car will become another advertising network.”
Context is critically important. Advertising in a parked car and a message interrupting navigation will elicit different reactions. BMW limited full animation only to parked cars, which shows an understanding of the importance of context. This is not just an unskippable banner, it’s a new ecosystem.
Alternative to influencers: programmatic video banners
Forget about bloggers and their inflated prices. Here we are talking about CPM cheaper than 50 rubles and pennies for a million impressions. The ability to buy impressions without creating a video, by placing a banner without filming on hundreds of videos – this is real math. This is programmatic video banners in Shorts, Reels, TikTok.
Testing hypotheses for 1000 rubles: how does it work?
For an arbitrageur, this is a goldmine. We can test hypotheses for 1000 rubles, getting transparent statistics. This is an ideal option for:
Promoting mobile applications
Advertising crypto projects
Banners for infobusiness
We are talking about a native banner in Shorts, which does not annoy, but organically fits into the content. This is the lowest CPM in Shorts you can find, and mass reach for pennies.
Frequently Asked Questions
What does “everything around us is an advertising network” mean?
This is the concept that any platform or device capable of attracting and retaining user attention can be monetized through advertising placement. From smartphones to cars and smart TVs.
How to get CPM cheaper than 50 rubles in short videos?
To do this, you need to use platforms that offer programmatic placement of banner advertising in Shorts, Reels, or TikTok, where the cost per impression (CPM) for bulk purchases can be significantly lower than traditional channels.
Is it possible to place banner advertising without creating a video?
Yes, it is possible. This refers to overlay advertising or native banners that appear over existing video content, without requiring the advertiser to create their own video.
Conclusion
The world of advertising is changing, and banner advertising in short videos, and now in cars, opens up new horizons for those who are willing to think outside the box. While BMW is experimenting, we can already order banner advertising and get cheap impressions in Shorts, Reels, and TikTok. Don’t wait for the market to settle – start testing new channels right now and grab your millions.
Interested in how to create AI-generated video that looks cinematic, not cheap? Do you want to have a repeatable workflow for turning a simple concept into polished AI video content using tools like Seedance? In this article, you will learn a step-by-step process for creating professional-quality AI video, from concept development and building key visuals to generating clips with Seedance and assembling them into a finished product.
The biggest misconception about AI video is that it’s easy. Entering a prompt into a video model yields a result, but its quality is entirely dependent on the creator’s ability to communicate with the tool. Ross Symons notes that most people who dismiss AI video as “low quality” are seeing the result of vague prompts, not the limits of the technology itself.
Understanding Models: LLMs vs. Diffusion Models
The first thing to understand is the difference between how Large Language Models (LLMs) and diffusion models process instructions. LLMs, such as ChatGPT, Claude, or Gemini, understand conversational intent. A diffusion model, the underlying technology behind most AI image and video generators, including Midjourney and Seedance, does not.
Diffusion models extract visual keywords from a prompt and ignore conversational “fluff.”
The phrase “create me an image of a cat walking on a beach wearing a cowboy hat” works in Midjourney, but the model doesn’t parse that sentence the way a chatbot would. It looks for visual keywords: cat, beach, cowboy hat.
Prompting strategies that work well in chat do not directly translate to image and video generation. Each model has its own prompting structure, and learning that structure is what separates a polished result from an ordinary one. Ross notes that certain tools excel at specific tasks.
“Midjourney, for example, remains the preferred tool among creatives for art direction, ideation, and creative exploration,” — Ross Symons.
ChatGPT’s image generation has the advantage of a reasoning layer that interprets intent, but many professionals still prefer the aesthetic output of a specialized image model.
For those who don’t know how to write structured prompts for a tool like Midjourney, there’s a practical life hack. Instead of learning the syntax, tell ChatGPT what you want the desired image to look like and ask it to write a prompt in Midjourney format. ChatGPT converts conversational descriptions into a structured, keyword-based format that diffusion models respond to, leading to significantly better results.
Pro Tip:Image generation is the foundation for AI video. Mastering the creation and refinement of static images first leads to a much higher success rate when transitioning to video, because the same principles of prompting, composition, and visual direction apply.
1. Developing the AI Video Concept
The concept is the underlying idea of what the video is trying to convey. It doesn’t have to be complex or deeply artistic. It can be as simple as “I want to show my product in an unexpected environment” or “I want to explain this technique with a visual metaphor.” The point is to have an intention before touching a tool.
Ross illustrates this with a project he originally created as a stop-motion animation many years ago. The concept: a can of Red Bull sits on a table. A piece of paper slides, folds into an origami bull, bumps into the can, opens it, drinks the liquid, grows wings, and flies away. The whole work is a play on the words “Red Bull gives you wings.” This concept has transitioned across different mediums. He recently recreated it using Seedance, feeding the sequence into an AI video model with a couple of image references, and the concept held up because the idea was strong regardless of the production tool.
Ross recommends using LLMs to further develop the concept. Even a basic idea can be expanded by asking ChatGPT to help extrapolate the narrative, suggest visual sequences, or variations. The goal is to have a clear understanding of the story before moving on to visual development.
2. Creating Key Visual Elements
Once the concept exists, the next step is to develop the key visual elements that will anchor the video. Ross divides this into three components:
Subject (or hero): This is the focus of the story. It can be a product, a person, or any object that drives the narrative. For the perfume ad Ross created, the hero was the bottle itself. He generated a product image mockup with Midjourney, establishing the look of the bottle before doing anything else. Creating product mockups with ChatGPT, Midjourney, or Gemini is a valid starting point when professional product photography is not available.
Environment: This is the setting where the story takes place. Ross built a jungle environment in Midjourney for a perfume ad. An environment can be created in any image generation tool or sourced from reference images on Pinterest, Shutterstock, or a personal photo library. It’s important to be specific about concrete visual qualities. Instead of telling the model to make something “cool,” Ross recommends describing what is appealing about the reference: the time of day, how the light hits surfaces, color temperature, and depth of field. The more specific the direction, the better the result.
Secondary Character or Element: This is anything else that appears in the scene to support the narrative. In the perfume example, Ross added a black panther entering the frame and eventually looking at the camera, then leaping toward it. This secondary element created tension and movement in what would otherwise be a static product shot.
Pro Tip: When using a personal photo as a character, isolate the subject on a solid background. Do not use a photo with a busy environment or other people. Provide multiple photos taken from different angles, all showing the subject in the same clothing. This gives the model clear data to work with when placing the character in new environments, poses, and scenarios.
Using Camera Angles for Visual Storytelling
Camera perspective is one of the quickest ways to elevate AI-generated visuals beyond the flat, centered composition most models produce. A few basic cinematography principles go a long way.
Low angles make a character powerful and dominant.
High angles can make a character appear submissive or vulnerable.
Close-ups create intensity.
Wide, pulled-back shots with empty space around the subject suggest isolation or vulnerability.
A practical tip for those without a cinematography background: use ChatGPT to analyze a film still. Upload an image and ask the model to explain what creates a certain emotional effect. Is it the camera angle? Dramatic lighting? Soft bokeh? Contrast? ChatGPT will break it down using the correct cinematic terminology, which can then be directly used in image and video prompts.
Ross tested this approach by asking ChatGPT to generate “Guy Ritchie’s six most commonly used shots” in a single image. The result included close-ups, high-angle shots, and several techniques Ross hadn’t encountered before, each usable as a reference for future prompts. Creators can reference specific directors or films without knowing the technical vocabulary. Phrases like “make this character look like Aladdin in this scene” or “show me a Guy Ritchie-style shot with this character in this environment” yield results that differ from the generic, centered output most people receive.
3. Creating an AI Video Storyboard Using Keyframes
A keyframe in the context of AI video is an image used as a reference point for the model. A storyboard is a sequence of 6-12 keyframes that outline the main moments of the video, establishing the angle, lighting, and composition for each moment.
The storyboard serves the creator more than the model. It provides structure, prevents overloading a single clip with too much action, and establishes a clear sequence of events before video generation begins. Ross emphasizes that the storyboard doesn’t need to be professional. It’s a planning tool, and rough images are perfectly adequate if they convey how each moment should look and feel.
There are two main approaches to using keyframes with AI video models:
Start Frame + End Frame + Prompt: This method provides the model with an image to start the video, an image for its end, and a text prompt describing what happens in between. For example, the start frame is an empty surface. The end frame shows a Red Bull can centered in the shot. The prompt reads: “A Red Bull can slides in from the right at a slow speed and stops directly in the center.” The combination of visual anchors and text instruction gives the model clear constraints.
Start Frame + Prompt Only: This method uses only a starting image and allows the prompt to guide the action without a fixed endpoint. It offers more creative freedom. For a 10-second clip, the start frame could be an empty surface, and the prompt could read: “A Red Bull can slides in from the left, a paper airplane flies in from above, lands next to the can, unfolds into a flat sheet of paper.” The model fills in the motion and timing.
Pro Tip: To create a longer video, stitch clips together by extracting the last frame from the previous clip and using it as the start frame for the next. This creates visual continuity across a multi-clip sequence without having to regenerate consistent elements from scratch.
Common Mistakes When Creating Keyframe Sequences
The most common mistake Ross sees is people cramming too much action into a short clip. A 5-second clip can only accommodate one or two actions. Writing a long paragraph describing a sliding can, flying paper, a bull forming, and an explosion, all within five seconds, overloads the model. The result is distorted limbs, warped perspectives, and visual artifacts. The solution is to match the complexity of the prompt to the duration of the clip.
Ross also distinguishes between using images as keyframes and using them as references. Keyframes require precise reproduction at specific points in the video, often leading to awkward transitions and unnatural camera movements as the model is forced to move between two rigid endpoints. References give the model an idea of the desired look without requiring pixel-perfect accuracy, and they tend to produce smoother, more natural results. Ross recommends using references instead of keyframes in most cases.
4. Video Generation with Seedance and Assembling the Final Product
Seedance, developed by ByteDance (the company behind TikTok), stands out for its adherence to prompts and the accuracy of reference images. Ross describes it as almost flawless in both areas, which is a significant step forward compared to video models even two years ago.
Seedance is not a standalone application. It is a model accessed through aggregator platforms. Ross recommends Luma AI (lumalabs.ai) as the preferred platform, along with Flora, Figma Weave, Artlist, Imagine Art, Open Art, and Krea. These platforms provide access to Seedance alongside other video models such as Kling 3.0, Veo 3, and others, all connected via API.
Google Veo 3 (available via Gemini) offers free video generation and was considered cutting-edge before the advent of Seedance and Kling 3.0. Ross notes that while Veo is useful, skills learned on one model do not automatically transfer to another. Each video model has its own prompting structure. Practice with Veo builds general familiarity with AI video, but getting the best results on Seedance requires learning how Seedance specifically responds to prompts, timing instructions, and reference images.
How to Effectively Use Prompts in Seedance
Seedance supports time-segmented prompting, which allows creators to specify what happens at certain intervals within a single clip. For a 15-second clip, the prompt might be structured as: “Between 0 and 4 seconds, this happens. Between 4 and 8 seconds, this happens. Between 8 and 12 seconds, this happens.” The model adheres to these timing instructions, allowing for the choreography of multi-step sequences in a single generation rather than stitching together multiple short clips.
Seedance Cost and Scaling Workaround
Seedance offers a choice of durations from 5 to 30 seconds per clip. Ross advises beginners to start with shorter, low-resolution clips to get familiar with the workflow. Generating a 30-second high-resolution clip without experience often leads to wasted money and unsatisfactory results. A 30-second Seedance clip at 720p resolution costs approximately $14. Increasing the resolution to 1080p roughly doubles the cost to $28-32. These prices are significantly higher than earlier AI video models, where 5-second clips cost less than $0.20, but the quality difference is also significant.
A cost-effective alternative is to generate at a lower resolution and then upscale. Ross generated a 480p clip for approximately $6, then upscaled it twice using the platform’s built-in tools to near-4K quality for an additional $3. The total cost of $9 yielded quality comparable to a native high-resolution render, at about a third of the price. The two leading upscaling tools are Topaz Labs and Magnific, both of which are integrated into most aggregator platforms.
Conclusion
Mastering AI video isn’t just about mastering tools; it’s about developing a director’s mindset. By applying principles of concept, visual development, and storyboarding, you can create high-quality AI video content that looks professional and engaging. Start with a clear idea, use LLMs to develop it, create strong visuals, and meticulously plan your sequences. Then you can scale your video productions to levels previously impossible, and now possible.
Frequently Asked Questions
What is Seedance and how does it work?
Seedance is an AI model for video generation developed by ByteDance. It allows you to create videos based on text prompts and reference images, while ensuring high accuracy and quality. It is accessed through aggregator platforms such as Luma AI.
How much does it cost to create a video using Seedance?
The cost of a 30-second clip at 720p resolution is about $14, and at 1080p it’s $28-32. However, you can save money by
Generative AI promises a revolution, but the real cost of content production often remains hidden. We calculated how much it costs to create a 25-second commercial using neural networks, including script, images, video, voiceover, and music. Spoiler: the 25-second video cost 2,126 ₽, or 85 ₽ per second. The lion’s share of the budget, almost 90%, was “eaten up” by video generation.
How much does AI advertising cost: a 25-second video for 2,126 ₽
In a world where technology is developing at an incredible pace, artificial intelligence is becoming an integral part of many industries, including content creation. But what is the real price of this innovation? We conducted an experiment to find out how much it costs to create a 25-second commercial entirely with neural networks. The results may surprise you, because the cost of unique content for one video and profitable CPM for advertising materials are becoming key factors.
The experiment was conducted through BotHub using the following models:
We advertised a herbarium, without commercial implications. Important: only the marginal cost was taken into account — the cost of successful generations included in the final video. Actual costs may vary depending on prompting skills and perfectionism level, increasing the number of regenerations and, accordingly, the budget.
Methodology for calculating the cost of AI production
For an objective assessment, it is important to distinguish between two types of costs:
Cost of result generation: direct computational costs for content creation.
Cost of infrastructure access: costs for subscriptions or minimum credit packages for using services.
In the case of a single video, the second amount can significantly exceed the first. Of course, open-source models can be deployed locally, but this requires a powerful PC, which is not always available.
«Don’t replace professionals with browser tabs. AI content has nuances: rights, licenses, character stability, result control, and brand requirements. AI slop is easy to get, and you’ll get flamed for it».
We started by creating a concept. GPT-6 Astra acted as a creative producer and scriptwriter. The prompt was formulated so that the model immediately took into account the limitations of AI generation and formed prompts for each element.
Prompt to GPT-6 Astra:
«Role: You are an experienced creative producer and scriptwriter specializing in creating advertisements using neural networks. Task: Develop a detailed script for a 25-second commercial for a Herbarium (an album with a collection of dried plants/herbs). Important condition: All content (video, photo, audio) will be AI-generated. Therefore, your task is not just to provide a script with a description of the picture, but a script supplemented with prompts for each element. Atmosphere: warm, autumn, cozy».
The script development and adjustments took ₽102.09.
Step 2: Generating images for the ad
Nano Banana 2 was used to create eight base images that set the composition, atmosphere, and style. The total cost of this stage was ₽128.29, which is an average of ₽16 per image.
The images served as an intermediate state: they fixed the spatial structure of the frame, which simplifies the task for the video model. This allows for consistency between scenes, which is critical for a commercial.
Image-to-video: precise control over the result
Unlike text-to-video, where the model simultaneously interprets the description, chooses the composition, and adds movement, image-to-video fixes most variables. This gives greater control over the result and ensures stylistic unity across all frames, which reduces the cost of uniqueizing one video and provides a favorable CPM for advertising materials.
Step 3: Video generation: the most expensive part of AI advertising
Converting static images into video scenes is the most resource-intensive stage. Seven video scenes were generated using Kling v3 Video, preserving the original composition and adding natural dynamics.
The cost of one video scene was about ₽268, and the total cost of video generation was ₽1,878.17. This is 88% of the total budget. The high cost is due to the need to ensure temporal consistency: objects, camera, lighting, and scene geometry must be stable throughout the video.
Step 4: Ad voiceover with ElevenLabs: cheap and tasteful
ElevenLabs Flash v3 was used for the voiceover. This stage turned out to be one of the cheapest — only 6.04 ₽. TTS technologies significantly simplify the process compared to traditional voice actor recording, providing fast generation and low costs. This is an excellent option for cheap hypothesis testing and prototypes.
Step 5: Music for advertising in Suno: budget and limitations
The musical accompaniment was created using Suno v5.5. The model allows describing the mood and genre, rather than specific notes, which is convenient for advertising tasks. The cost of music generation was 11.82 ₽.
Although AI music saves budget, full control over the result is not guaranteed. However, for small projects, it is an effective alternative to stock tracks or ordering original music.
Total cost and comparison with direct subscriptions
The total cost of a 25-second advertising video created with AI was 2,126.41 ₽.
Tool
Cost
Script and prompts (GPT-6 Astra)
102.09 ₽
Base images (Nano Banana 2)
128.29 ₽
Video generation (Kling v3 Video)
1,878.17 ₽
Voiceover (ElevenLabs Flash v3)
6.04 ₽
Music (Suno v5.5)
11.82 ₽
TOTAL
2,126.41 ₽
If access to each service were paid for separately, minimum monthly subscriptions would be around 3,591 ₽ (at an exchange rate of $1 = 86.4 ₽):
ChatGPT (Plus): $20 (~1,729 ₽)
Nano Banana 2: ~$0.54 (~47 ₽)
Kling v3 Video: $6.99 (~604 ₽)
ElevenLabs Flash: $6 (~519 ₽)
Suno v5.5: $8 (~692 ₽)
These 3,591 ₽ are more of a starting capital than the cost of a single video, as tariffs usually include many more credits than required for one project. This highlights the benefit of using aggregators like BotHub for one-off tasks or cheap hypothesis testing.
Conclusions and prospects for AI production
AI production is not yet a seamless process. Each model has its own features and data formats. Cohesion between stages is ensured by an external workflow: prompts, references, manual editing. This distinguishes the demonstration of individual model capabilities from a full-fledged production system, where all elements must work in unison. The price per finished clip with AI can be significantly lower than traditional production, but requires competent pipeline management.
Overall, AI tools significantly lower the barrier to entry for video content creation, offering a budget-friendly alternative to seeding and traditional production studios. However, obtaining a high-quality result requires skills and an understanding of each model’s limitations. This makes CPM profitable for testing various advertising creatives.
Frequently asked questions
What CPM should be considered profitable for AI advertising?
A profitable CPM depends on the niche and campaign goals. If the cost of producing a video is low, as in our case, then even with an average CPM, the overall return on investment (ROI) can be high. The main advantage is the ability to quickly and cheaply test hypotheses.
What is cheaper: a banner, a blogger, or an AI video?
A direct comparison is incorrect, as these are different formats with different goals. However, AI video offers cheap hypothesis testing and can be significantly cheaper than producing a video with a blogger or even some types of banners, especially if complex animations are required. The cost of creating one AI video at 2,126 ₽ makes it competitive for mass content production.
Is it worth investing in AI for content production?
Yes, it is. AI significantly reduces the time and cost of production, allowing for quick creation and testing of advertising materials. This is especially relevant for small and medium-sized businesses, as well as for arbitragers who value the cash and unit economics of each creative. However, it is necessary to consider that the price vs. result strongly depends on the skills of working with prompts and post-processing.
How much does it cost to cut 1 hour of stream using AI?
The cost of cutting an hour of stream using AI will depend on the services used and their rates for video processing. If it’s about creating short clips, it can be significantly cheaper than manual work and comparable to the price of generating one video, as the main costs will be for transcription and content analysis, not for generating video from scratch.
What is the price per minute of AI video with an avatar?
The price per minute of AI video with an avatar varies depending on the platform and the complexity of the avatar (2D, 3D, customization). In our experiment, 25 seconds of video cost 1,878 ₽, which is equivalent to approximately 4,500 ₽ per minute if only the video generation cost is extrapolated. However, avatar platforms have their own rates, which can be both higher and lower, depending on the features and quality provided.
AI advertising opens up new horizons for content creation, making it more accessible and faster. If you want to delve deeper into the world of AI and learn how it can transform your business, visit our blog. Discover new opportunities and start creating unique content today!
Many companies face a problem: their video marketing strategy looks brilliant on paper, but in practice, it turns out to be unfeasible due to limited resources. This article is not just a list of tools, but a practical guide for scaling video production, based on real experience. We will show you how to create a strategy that your team, whether an internal department or an external partner, can effectively implement, avoiding burnout and overload. The key is to align ambitions with actual production capacity, as this is most often the bottleneck.
A video marketing strategy is a document that clearly defines target audiences, funnel stages, formats, distribution channels, budget, realistic production frequency, and success metrics. It requires making concrete decisions. In contrast, a video plan is simply a list of videos (“three product reviews,” “two webinars,” “four founder interviews”). The plan tells you what to shoot, but doesn’t explain why these formats and channels were chosen, or how success will be measured.
The difference between a strategy and a plan is the difference between a collection of disparate videos and a cohesive effort that leads to measurable results in the sales funnel. A successful video strategy is always synchronized with the company’s overall marketing strategy. It ensures that each video actively contributes to achieving global marketing goals, rather than existing in isolation. Such a document, approved by marketing and sales management, is your strategy.
Why Video is the Best Marketing Tool in 2026?
The statement that video attracts more attention than other formats was relevant back in 2018. Today, video is becoming the primary marketing format due to three key changes:
Loss of third-party signals has reduced the accuracy of programmatic advertising.
AI-generated content has flooded search results. Google reviews and ChatGPT answers reduce the amount of textual information, decreasing clicks. Video, especially with recognizable faces and voices, is much harder to condense into a few sentences.
Short video platforms increase session watch time. According to HubSpot’s 2024 State of Marketing report, 21% of marketers believe that short videos yield the highest ROI. A SundaySky 2025 buyer survey showed that 83% of consumers want more video content from brands.
Wyzowl has been recording an increase in the use of video in marketing annually since 2017. It is important to rely on verified data. For example, it can be argued that video on product pages significantly increases the intent to purchase, especially in e-commerce for products requiring serious consideration.
Also, video in email increases CTR, and mobile devices generate more than 50% of video views (Statista). This data can be confidently used in presentations.
Video Content at Each Stage of the Sales Funnel
While the concept of a sales funnel may seem trite, it is the only way to link video production to actual revenue. Structuring a video campaign across the entire sales funnel helps guide a potential customer from initial brand awareness to purchase.
1. Awareness (Top of Funnel)
At this stage, many brands focus too much on aesthetics and not enough on quantity. The goal is to ensure brand visibility in social media feeds.
Formats: short social media videos (15-45 seconds), influencer interviews (1-3 minutes), animated explainers (e.g., “what is product-led growth”), behind-the-scenes footage.
The audience already knows what product category they are looking for, but has not yet chosen a specific solution.
Formats: product reviews (90 seconds – 3 minutes), shortened webinar recordings (5-8 minutes), comparison videos (“X vs. Y in 2026”), demo segments for specific use cases.
Channels: paid YouTube advertising (pre-roll), private landing pages, long YouTube videos, sponsored content on LinkedIn.
Metrics: video completion rate, conversion via forms, contribution to MQL.
These educational videos and detailed product demonstrations help buyers understand technical differences and accelerate purchase intent.
3. Conversion (Bottom of Funnel)
The audience has several solution options and is choosing the most suitable provider.
Channels: sales team library (in proposals, on client pages), LinkedIn and YouTube retargeting, paid placements for search queries.
Metrics: pipeline contribution, sales cycle length for engaged accounts, deal size for engaged accounts.
4. Retention (post-purchase)
The audience is already your customer.
Formats: series of educational videos for onboarding, new release announcements, libraries of educational materials, update videos from leadership.
Channels: in-product prompts and modals, customer email newsletters, LMS or knowledge base, communication with the customer success department.
Metrics: feature adoption rate, reduction in support tickets, contribution to NRR (net retention revenue).
5. Advocacy (champion engagement)
Customers are ready to recommend you.
Formats: presentations at customer summits, joint videos with partners, collections of user testimonials, panel discussions with executives.
Channels: customer events, partner channels, Slack and Discord communities.
Metrics: referral pipeline volume from advocates, loyalty program participation, attendance at customer summits.
Five Steps to an Effective Video Marketing Strategy
We use this process with all our clients, and it takes 2-3 weeks depending on the number of participants.
Step 1: Define Target Audience Segments
Identify 2-4 segments based on buying behavior. For example, “RevOps leaders in mid-sized SaaS companies evaluating attribution platforms” is better than just “B2B marketers.” For each segment, record:
Where do they spend their time (publications, podcasts, communities)?
What are their main work stresses this quarter?
What content have they engaged with previously?
What decision criteria are most important to them?
“Don’t skip interviews. Ten 30-minute calls with customers will provide more useful information than any $40,000 market research.”
Step 2: Define the Goal and Metrics for Each Segment
For each segment, choose one primary goal and 1-2 metrics. Don’t measure too much. Demand-stage video is evaluated by completion rate and form conversion. Conversion-stage video is evaluated by its impact on pipeline and sales cycle. Retention video is evaluated by feature adoption rate.
Be specific: “video on the /pricing/ page increases demo requests from 2.1% to 2.6%” is a good metric, unlike “video increases MQL.”
Step 3: Match Segments and Goals with Video Formats
Based on the previous sections, create a list of formats: “for the RevOps director segment, we release one 5-minute customer testimonial video monthly, four 60-second versions of this video for LinkedIn, and one 3-minute feature overview quarterly.” This list will form the basis for production briefs.
Step 4: Choose Distribution Channels
Many strategies lose credibility by listing all channels as “omnichannel.” Limit yourself to 2-3 channels per segment and commit to publishing content consistently, at least once a week. Example: “4 posts per week on LinkedIn (organic style), 2 long videos per month on YouTube, an always-on budget of $8000/month for paid retargeting on LinkedIn, quarterly updates to the video library for the sales department.”
If you cannot ensure this level of consistency, start with fewer channels.
Step 5: Define the Measurement and Review Process
Decide in advance: monthly performance review, quarterly strategy revision, annual strategy update. Create a unified dashboard (Looker, Google Analytics 4, HubSpot) that includes all metrics from step 2, as well as performance metrics (number of videos released vs. planned, average publication time, video cost).
If you only release 60% of the planned volume, the strategy is not working, regardless of engagement metrics.
Matching Strategy to Editor Bandwidth: A Realistic View
This aspect is often overlooked. One in-house video editor, working at a steady pace, can produce approximately 80-120 minutes of finished video per month. This translates to 8-12 ten-minute videos, 20 five-minute videos, or hundreds of 30-second clips (given ready source material).
The volume varies depending on editing complexity, use of B-roll, motion graphics, and the number of iterations. Two editors do not double the production volume but increase it by approximately 1.7 times due to review cycles, asset management, and the time a senior editor spends reviewing a junior editor’s work.
Typical Production Capacity Scenarios:
Scenario A: 1 in-house editor.
Cost: $7,000–$8,000/month.
Output: 8–10 finished videos/month.
Support: one funnel stage (usually awareness/demand generation) with minimal coverage for one additional stage. Insufficient for a full-cycle strategy.
Scenario B: 1 in-house editor + partial production partner.
Cost: $10,000–$14,000/month.
Output: 15–25 finished pieces/month.
Support: two funnel stages with periodic support for others.
Scenario C: 2 full-time editors + production partner (for short-form and motion graphics).
If your strategy requires Scenario C output and your budget aligns with Scenario A, the strategy is doomed to fail. You must either reduce the scope of the strategy or increase production capacity. There are no compromises here that lead to success.
Video Tools: Not a Panacea
Many strategies mention video tools, mistakenly believing that low prices create more power. This is not the case. Using WeVideo for $4.99/month will not increase content volume; it will simply make the “keyboard” for your editor more accessible.
Wistia (from $99/month), Vimeo (from $12/month), InVideo AI (from $40/month), Loom (from $12.50/month for asynchronous messages) – all these are useful tools, but they should be chosen after solving the production capacity problem, not before.
The Downside of Video Marketing: What is Often Left Unsaid
While the benefits of video are widely publicized, the drawbacks are often ignored. Here are the real challenges we face daily:
High production costs. A blog post takes 4-6 hours of an author’s work. A 5-minute customer testimonial can require 25-40 hours of work from a producer, cameraman, and editor, as well as the availability of the shooting location. The cost difference is significant and does not decrease with cheaper tools.
Complexities with rights and talent management. Processes for managing music rights, image usage permissions, and any regulatory requirements (e.g., claim substantiation) require documentation. When shooting customer testimonials, always get permission before filming, correctly identify the subject’s employer, and ensure their compliance team has approved the script. We have seen two perfectly shot testimonials not be published for nine months due to legal approvals of statistics mentioned on camera.
Platform policy changes reduce ROI. Advertising rules and organic distribution mechanics on TikTok, Meta, YouTube, and LinkedIn change quarterly. A specification for a short ad developed in 2024 may become invalid by 2026. Plan for regular specification audits twice a year; instead of embedding numbers in your playbook, link to official platform policy pages.
Measuring video effectiveness is harder than it seems. Views can be artificially inflated. Autoplay affects watch completion rates. The definition of a “view” on YouTube differs significantly from other platforms.
Frequently Asked Questions
What is a video marketing strategy and how does it differ from a video plan?
A video marketing strategy is a comprehensive document that defines the target audience, funnel stages, formats, channels, budget, production frequency, and success metrics. It explains why and how videos will be created. A video plan is a simple list of videos to be produced, without justification or metrics.
What three key changes make video the primary marketing format today?
Firstly, the loss of third-party signals has reduced the accuracy of programmatic advertising. Secondly, AI-generated content has flooded search results, making text less valuable. Thirdly, short-form video platforms significantly increase watch time per session.
How do I determine if my budget can support my chosen video marketing strategy?
Assess your current production capacity (number of editors, their workload) and align it with the required content volume. One in-house editor
Short videos have long ceased to be just a format for bloggers. Today, earning from short videos has become a reality, as income can be generated from creating and distributing vertical content by completing advertising tasks for brands and platforms. But how to start if there is no experience? How to transition from simply watching videos to real earnings? This is precisely why we created the VibeVO creator school, and today we are ready to share the impressive results of our graduates.
VibeVO School Statistics: From First Steps to Real Income
We launched the VibeVO creator school as an educational platform for those who want to master working with short videos and apply the acquired knowledge to real offers. Now we have enough data to analyze the journey of our participants after starting their training.
Learning Funnel: Who Makes It to the End?
As of today, the school’s funnel looks like this:
752 people started training.
310 people completed at least one lesson.
84 people fully completed the school.
This means that hundreds of participants have already taken the first practical step, and 84 people have successfully completed the entire program. For us, it is especially important not only the number of lessons watched but also how participants transition from learning to real content work. And here the numbers turned out to be even more interesting.
Impressive Results: School Participants Earned Over 21,500 USDT
Creators who completed at least one school lesson have earned an impressive amount on VibeVO over time:
21,506.36 USDT.
This amount includes accruals for completed tasks and content created on the platform. It is important to note that this refers to the earnings of these users on VibeVO throughout their activity, and not a guaranteed result of completing the school. The training itself does not automatically generate views and money. After completing the course, it is still necessary to choose offers, create content, publish it, and test various approaches.
However, the VibeVO school helps overcome the most difficult stage — transitioning from “I want to try” to concrete actions and starting your journey in the world of short video creation.
Almost Every Second Graduate Has Already Earned on the Platform
The most telling figure is the results of those who completed the training. Out of 84 school graduates, 40 have already earned on VibeVO. This accounts for 47.6% of graduates — almost every second one.
Behind each such figure is a creator who did not limit themselves to watching lessons: they chose an offer, created content, published it, and started getting results. This is the main goal of our school.
We do not aim to create training for the sake of training itself. Our goal is to give a person enough knowledge so that after the course they can confidently open the list of active offers and understand:
What content can be created.
How to approach creating short videos.
How to start publishing.
How to test various formats.
How to turn views into real earnings.
Key to Success: Start Publishing
The difference between a person who is “interested in earning from short videos” and a true creator often lies in just one action — the first publication. It is not necessary to immediately create a perfect video or gather millions of views. It is much more important to go through the entire cycle at least once: choose an offer, prepare content, publish it, and analyze the result.
After this, it becomes much easier to analyze statistics, adjust approaches, and scale what works. The results of VibeVO school graduates clearly demonstrate that this path is quite real and accessible to everyone.
Join Successful VibeVO Creators!
If you once started training at VibeVO school but stopped after the first lessons — now is the time to return and continue. And if you have never tried to earn from short content, start with the first lesson, understand the mechanics, and immediately apply the acquired knowledge to active VibeVO offers.
752 people have already taken the first step. 84 have completed the training. And almost every second graduate has already earned on the platform. You could be next! 🚀 Don’t miss your chance to start earning from creating short videos today.
In the world of mass video posting, where automatic short video publishing plays a key role, understanding brand context becomes critically important. The word “date” can have many meanings: a romantic dinner, a sweet fruit, or a specific day of the year. Without additional information, it’s impossible to determine the true meaning. Similarly, for effective interaction with artificial intelligence tools and their seamless functioning, context plays a crucial role.
If marketers’ tools are not trained with previous results and strategy in mind, they will not be able to optimize future outcomes. To fully unlock the benefits of AI, it must be trained based on the specific brand’s context and the knowledge of each member of the marketing team, as noted by Sandeep Menon, CEO and co-founder of marketing orchestration platform Auxia.
Auxia: Solving the Problem of Context and Intelligence in AI
Auxia was developed to create the very “layer of context and intelligence” that AI needs to thrive. The platform aims to solve three key marketing challenges:
The “tribal” nature of marketing: Each company has its own unique internal approach that is difficult to transfer to other brands or industries. For example, Google’s marketing lifecycle is completely different from Uber’s strategy.
Tool fragmentation: Marketers use many different tools across various platforms. “They all do wonderful things,” says Menon, “but they don’t talk to each other.” This creates a problem of a lack of unified context and the need to connect the entire campaign workflow.
Multi-user experience: Marketing is a “multi-user” experience, so any AI tool built for this industry must be accessible to all teams.
How Auxia Solves These Challenges
Auxia offers an end-to-end agency marketing platform that includes two products:
Auxia Decisioning: A cross-channel tool that creates a unique profile for each customer to determine which ads they are most likely to respond to.
Auxia Agent Studio: A new offering launched in August, consisting of various agents that can measure campaign performance and execute changes. Marketers are not satisfied with a tool that stops at the analysis stage, leaving the question: “What to do next?”
Auxia connects to the client’s existing content repository and data from external platforms such as Figma and Salesforce. This allows Auxia agents to suggest next steps, such as which assets might work better for a specific audience or which creative needs to be updated.
“Our agents are trained on a company’s own data, and one company’s data is never used to make decisions for another. This approach immediately solves the ‘tribal’ nature of marketing, ensuring full consideration of brand context and data privacy,” emphasizes Sandeep Menon.
If an employee moves to another team or leaves the company, their knowledge is already embedded in the agent set, preventing workflow gaps. Agent Studio, which includes specialized agents with roles from real-time campaign optimization to bid tracking, also unifies the marketer’s toolkit in one place.
Instead of tracking dozens of different agents or tools, Agent Studio serves as a central “control panel” that determines which ancillary agent is best suited for a task. This provides a unified dashboard for TikTok, YouTube, Instagram, and other platforms, optimizing video cross-posting.
First-Order Data-Driven Optimization
Auxia agents test and analyze performance at the user level, including their past behavior and engagement. Much of Auxia’s optimization occurs on owned and client-managed channels (rather than large third-party ad platforms like Google or Meta).
This provides Auxia with extensive first-order data directly from clients. Access to this real-time data allows Auxia to match the right ads with the right users. For example, Auxia can target readers of a client publisher with ads related to recently read articles.
Other on-site behavior is also tracked, such as if a reader came to the site through another ad, and this information is used to determine what content they like. This is critical for short video distribution and mass uploading with proxies and anti-detection.
Transforming Marketing Processes
The industry is in the early stages of AI development. Menon compares it to the advent of electricity in the late 19th and early 20th centuries: when electric motors were introduced in factories, they initially simply replaced steam engines without changing the factory structure.
True performance improvement only happens when the process itself changes, not just when old technology is replaced with new. “Our goal now is to drive that transformation,” Menon states.
For Auxia, this “transformation” means partnering with Chief Marketing Officers (CMOs) and helping them restructure their teams. People should focus on strategic decisions, while machines should handle more routine, repetitive work. Utilizing AI to its full potential is “not a technology problem,” he adds. “It’s an organizational problem.” This allows teams to effectively set up an auto-posting schedule and perform publishing to 5 platforms without manual labor.
Frequently Asked Questions
How to choose a mass-posting service for videos?
When choosing a mass-posting service for videos, it is important to consider several factors: integration capabilities with various platforms (TikTok, YouTube, Instagram), the availability of features for automatic short video publishing, support for scheduling across 10+ accounts, as well as the presence of analytical tools and options for mass uploading with proxies and anti-detect. Make sure the service provides a unified dashboard and supports mass-posting via API.
How much does a monthly mass-posting package cost?
The cost of a monthly mass-posting package can vary greatly depending on the functionality, number of accounts, and volume of publications. Some services offer plans that include 500 publications, while others offer an unlimited number. It is important to compare offers from different platforms to find the optimal balance of price and quality for your needs, for example, if you need to publish 100 videos a day.
Can I publish videos to 5 platforms simultaneously?
Yes, modern mass-posting services allow you to publish videos to 5 platforms and more simultaneously. This is achieved through cross-posting and integration with the APIs of various social networks. Platforms like Auxia provide a unified dashboard, which significantly simplifies the process and saves time, allowing for seamless publishing to 5 platforms.
How to ensure seamless mass-posting without manual labor?
To ensure seamless mass-posting without manual labor, it is necessary to use automated platforms with auto-scheduling and content rotation features. Such systems allow you to plan publications in advance, manage content through a unified dashboard, and automatically distribute videos to all selected accounts. This allows you to perform posting to 10 accounts and more while you focus on other tasks.
Conclusion
Effective Mass video posting requires not only technological solutions but also a deep understanding of the brand’s context. Platforms like Auxia offer innovative approaches, combining AI with unique company data to optimize marketing campaigns. Investing in such tools not only saves time but also significantly increases the effectiveness of your content strategy. Start using automatic short video publishing today to ensure maximum scheduled reach and achieve outstanding results.
In a world where AI systems generate new queries at the speed of light, old hacking tricks of relevance are getting a second life. Greg Jarboe, a veteran SEO, recounts how back in 2003 he manually implemented what is now called a “fan query,” using the “Russian nesting dolls” principle. This method allows your content to be found for queries that don’t even exist yet, and is key to query repurposing in modern AI search.
Matryoshka SEO: How it worked before AI
Long before the terms “fan query” or “GEO” became common, Jarboe optimized press releases by embedding short phrases within longer ones. For example, if the target phrase was “flights to Philadelphia,” he would look for a four-word version, such as “cheap flights to Philadelphia,” and build the release around it.
This allowed for coverage of both short and long queries, making the content visible to a wide audience. This approach ensured maximum reach, even when search engines were less developed.
15% unknown queries: a Google constant
“I thought at some point most queries would have been done; people just ask the same thing over and over. But when we re-run those metrics, it’s always about 15%,” — John Mueller, Search Central Live NYC, March 2025.
Google first mentioned that 15% of daily queries are new in 2019 when introducing BERT. This figure remains stable despite the development of large language models. Multiply 15% by billions of Google queries, and you get hundreds of millions of unique formulations every day.
This is a huge layer of traffic that most SEO specialists underestimate. Understanding this phenomenon is critically important for effective query repurposing.
News as a query generator
Most of these new queries are not random. They are born from breaking news, new products, changed policies, or phrases coined by journalists. Press releases published on the day of an event are ideal for intercepting this traffic.
If the wording in the release accurately matches what a user types into Google, the content can rank for a query that did not exist at the time of writing. This demonstrates the power of timely content in capturing “new” search demand.
AI and reverse matryoshka: from broad to narrow
M.J. Cachon’s research showed that ChatGPT generates sub-queries that are also not random. Initially, the model uses general, conversational language, then narrows the search with the site: operator and begins to check exact quotes.
The use of quotes increases 25-fold between the first and last search. This is the “matryoshka” principle, but in reverse: AI starts broadly and narrows down to the exact phrase for verification. This AI approach to search opens up new opportunities for query repurposing.
Query Length: The New Reality of AI Search
According to Google data from May 2026, the average length of an AI query in the US is three times that of a traditional search query. Kachon’s data shows that one branded ChatGPT query branches into sub-queries, averaging seven words each.
Length is not a side effect of AI search, but a new norm that the industry has been moving towards for years. This means that content must be optimized for longer and more detailed queries to match AI behavior.
Practical Steps: How to Deceive Algorithms and Scale Content
To effectively use this knowledge for query repurposing, three key habits must be developed:
Look for nested phrases, not just keywords. For every three-word keyword phrase, find two or three four- and five-word phrases that naturally contain it. Build your introductory paragraphs or H2 headings around the longer version.
Publish at the speed of news. 15% of new queries are related to recent events. If your organization has a channel for same-day publication (press release, blog, rapid response page), this is your best chance to rank before competitors even know the phrase exists.
Write the verbatim answer as a quotable sentence. Kachon’s data shows that AI systems are increasingly verifying claims by looking for an exact quote from your content. Rewrite the sentence that answers the question so that it can be taken verbatim and still make sense.
These methods do not replace fundamental SEO principles but complement them, focusing on the long tail of queries that were previously underestimated. They allow for effective adaptation to changing AI search algorithms.
Frequently Asked Questions
What is a “query fan-out” in the context of AI?
“Query fan-out” is the process by which an initial user phrase or query branches into many more specific sub-queries generated by an AI system. This allows the AI to delve deeper into a topic and find more relevant information, narrowing the search from general to specific.
How is the “Russian matryoshka” principle related to SEO?
The “Russian nesting doll” principle in SEO means embedding short, high-frequency keywords into longer, low-frequency queries. For example, the phrase “airline tickets” can be nested into “cheap airline tickets to Philadelphia.” This allows content to be found for queries of various lengths, increasing its visibility and audience reach.
Why are 15% of Google queries always new?
These 15% of queries are always new because they are directly related to real-world developments: news, new product launches, changes in legislation, cultural phenomena. People search for information about what has just happened or appeared, generating unique formulations that did not exist before. This confirms the importance of timely content publication.
How do AI systems use quotes to verify information?
M.J. Kachon’s research showed that AI systems, such as ChatGPT, after a broad search, begin to verify information using exact quotes. They look for specific sentences in the content to ensure that the source actually states what the AI interprets. This emphasizes the need to formulate answers as precisely and quotably as possible.
Conclusion
In the era of AI dominance, understanding how systems generate and process queries becomes critically important. By applying the “matryoshka doll” principles, optimizing content for long and nested phrases, and publishing it promptly, you can outpace competitors and capture traffic that hasn’t even been conceived in users’ minds yet.
Start applying these hacking techniques today so that your content doesn’t just exist, but scales and tricks the matrix of search algorithms. Don’t miss the opportunity to become a leader in your niche by using advanced strategies of query repurposing!
As a professional producer, I know that stream highlights are not just about trimming; they are an entire assembly line for creating highlights that captivate audiences and hold their attention. A recent Twitch report titled “State of Gaming 2026” vividly demonstrates the colossal interest in gaming content: from January 1 to September 1 of this year, platform users watched 8.6 billion hours of gaming streams. This is a huge volume of “raw material” that needs to be skillfully packaged to bring maximum benefit.
Viewing Leaders and New Trends
The Twitch report clearly shows that despite the dominance of the “Just Chatting” category, gaming streamers continue to thrive. Among them, Jynxzi stands out, having accumulated an impressive 78 million hours of viewership playing popular games such as Rocket League, Rainbow Six Siege, and League of Legends. This is a prime example of how high-quality gaming content, skillfully cut and presented to the audience, can bring a streamer immense popularity and recognition.
King of Esports and Global Phenomena
League of Legends: This game confidently holds its position as the leader in viewership in the US, UK, and Latin America, confirming its status as a global phenomenon.
Global Impact: Twitch specifically highlights LoL stars such as Noway4u_Sir from Germany and Kamet0 from France, whose broadcasts gather millions of viewers.
Such data confirms that video podcast editing or gaming stream editing with an emphasis on esports moments is a true goldmine for attracting new audiences and effectively monetizing streams through highlights.
From “Cozy” Games to Horror Hits
The report also identified new niches that are actively gaining momentum and attracting more viewers:
“Cozy” games: Projects with cute characters and a calm, relaxing atmosphere, such as Pokopia and Slime Rancher, garnered an impressive 142 million hours of viewership since the beginning of 2026.
Horror games: The genre shows stable growth of 6% year-over-year, reaching 287 million hours of viewership. Games like Dead by Daylight and Phasmophobia confidently lead in this category, attracting thrill-seekers.
“Since the earliest days, Twitch has been a place where people come together around a shared love of games. 8.6 billion hours of gaming content watched in just eight months is a powerful testament to the strength of the Twitch gaming community,” said Mike Minton, Twitch’s Chief Product Officer.
The Future of Gaming: GTA VI and Limitless Possibilities
Mike Minton also mentioned the highly anticipated GTA VI, emphasizing its enormous potential for creating diverse content – from “cozy” adventures to thrilling horror scenarios. This confirms that cropping 9:16 from horizontal for TikTok and other vertical platforms will become even more relevant, as players will create an endless number of unique and viral moments.
As a person who edits streams, I see huge potential for growth in this. If you want to order 20 clips from a stream or even bulk editing of 100+ clips, now is the perfect time. It’s important not only to professionally cut highlights but also to add subtitles for highlights to reach the widest possible audience and make content accessible to everyone. Learn how we edit highlights to make your content shine.
Frequently Asked Questions
How long does it take to edit a 2-hour stream?
With sufficient experience and the necessary tools, editing without loss of quality a 2-hour stream into 5-10 clips can take from 30 minutes to 1 hour. The key to hyper-productivity is a clear understanding of what moments to cut: emotional peaks, funny phrases, important game events, and unexpected twists.
Are subtitles necessary for highlights?
Yes, subtitles for highlights are critically important in the modern world. They not only improve content accessibility for people with hearing impairments but also significantly increase audience engagement (many watch videos without sound). In addition, subtitles contribute to better SEO optimization. Today, there are convenient services, such as Opus Clip alternative, that allow you to automate this process, saving your time and resources.
How much does it cost to edit 1 hour of stream?
The price per 1 hour of stream for editing can vary greatly depending on many factors: content complexity, the number of required final clips, the need to add special effects, graphics, and subtitles. Typically, the cost is estimated either per hour of source material or per number of finished clips. Get a personalized cost estimate to accurately plan your budget.
Conclusion
The world of gaming on Twitch lives, evolves, and constantly offers new opportunities. From “cozy” games to thrilling horrors and dynamic esports – there’s plenty of content for every taste. If you are a streamer, then turning a long video into highlights is your direct path to audience growth, increased recognition, and effective monetization of your creativity. Don’t miss the opportunity to package your “raw material” into high-quality, engaging highlights that will work for you. Contact us to start creating your best clips today and take your content to the next level!
Forget about bloggers and multi-million budgets! A new player is entering the market, promising mass reach for pennies. The startup Telly, which gives away free smart TVs in exchange for data and ad display, is opening up programmatic banner placement on its home screens. This is real math for those looking for banner advertising in short videos and other short video formats.
In a world where every click and view costs more and more, finding an effective yet budget-friendly channel for promotion becomes a real challenge. But what if there’s a way to get huge reach for pennies, using an innovative approach to advertising? Introducing Telly – a startup that could change your perception of advertising campaign costs and offer CPM cheaper than 50 rubles.
A New Era of Overlay Advertising: Banners on the Second Screen
Telly’s uniqueness lies in its dual-screen TVs. Below the main display, where the user watches content, there is an additional, narrower screen. This is where a constant stream of news and, most importantly for us, programmatic video banners are broadcast. This is an undisguised banner that is always in view.
Technology: Telly uses Magnite’s SpringServe ad server.
Demand sources: The Trade Desk, Teads, and other major platforms.
Advantage: The banner is constantly displayed without interrupting the main content.
Why is this important for an arbitrageur?
Mike Sheehan, Telly’s Chief Revenue Officer and former CEO of SpotX, emphasizes: for home screens to become an effective advertising channel, they need scale. And programmatic placement is precisely what provides this scale, allowing you to buy reach without creating a video.
Telly actively collaborates with IAB Tech Lab to create standardized specifications for advertising on home screens. This will unify the process and attract more investment in this segment.
Data is Everything: Precise Targeting of the Audience
When a user receives a Telly TV, they fill out an extensive questionnaire. This questionnaire includes data on household income, number of residents, brand preferences, insurance companies, and even car makes. Sheehan notes:
“The questionnaire goes very deep into demographics, lifestyle, and brand loyalty.”
Combined with ACR (Automatic Content Recognition) data, which shows exactly what the user is watching, this provides unprecedented targeting.
Imagine being able to show insurance ads only to households actively looking for new insurance. This is not just crypto project advertising or mobile app promotion, it’s a native banner in Shorts that hits the target precisely. Hypothesis testing for 1000 rubles becomes a reality when you know who you’re showing ads to.
Long Banners: A New Format for Engagement
In addition to standard programmatic banners, Telly offers longer ad placements – up to five minutes – through private marketplace (PMP) deals with Magnite. Sheehan claims that this duration, combined with the non-intrusive nature of the placement, significantly increases engagement.
For example, one streaming service (name undisclosed) achieved five times more app opens using a five-minute banner compared to a two-minute one. Mike Trion, Head of CTV and Video Strategy at PMG, confirms the effectiveness:
“The home screen has always been a powerful entry point for engaging viewers. Telly extends this opportunity by providing brands with a persistent presence that drives engagement.”
This opens new horizons for brand integration without filming and banners for infobusiness, where not only reach but also depth of interaction is important.
Conclusions and Call to Action
Telly offers a unique opportunity for arbitrageurs and marketers to get cheap reach and mass reach for pennies. The ability to auto-place on hundreds of videos (or rather, on hundreds of thousands of home screens) with deep targeting and low CPM is exactly what’s needed in 2024.
If you are looking for a way to place a banner in TikTok, advertise in Reels, or banner in YouTube Shorts, but are tired of high prices and opacity, then Telly could be your goldmine. Don’t pay an influencer when there are transparent statistics and the ability to order banner advertising at real prices.
It’s time to rethink your strategy and try this channel. Stay tuned for updates, look for PMP deal opportunities, and test hypotheses. These pennies for a million can bring you real profit.
Frequently Asked Questions
What is Telly and how is it related to banner advertising?
Telly is a startup that distributes smart TVs for free. The unique feature of these TVs is a second screen that constantly displays advertisements, including programmatic banners. This allows brands to place banner advertising in video formats without creating the video content itself.
What CPM can be expected from advertising on Telly?
While exact CPM figures have not yet been disclosed, Telly’s approach to mass free device distribution and programmatic placement suggests the potential for extremely low CPM, possibly CPM cheaper than 50 rubles, making it an attractive alternative to traditional advertising platforms.
Is it possible to place ads without creating a video?
Yes, this is one of Telly’s key advantages. You can buy reach without creating a video, using static or animated banners that appear on the TV’s secondary screen. This is an ideal option for a banner without filming.
How accurate is the targeting offered by Telly?
Telly collects extensive user data through detailed questionnaires upon registration, and also uses ACR (automatic content recognition) data. This allows for deep targeting by demographics, interests, behavior, and even brand loyalty, making a banner for infobusiness and other niches as effective as possible.
How to start advertising on Telly?
To start collaborating with Telly, you need to work with their programmatic placement partners, such as Magnite, The Trade Desk, or Teads. It is recommended to contact representatives of these platforms for detailed information on how to order banner advertising on Telly.
In 2026, speech synthesis technologies reached an incredible level, transforming the complex process of audio content creation into a matter of minutes. Now, it’s enough to insert text, choose a model and voice, and your file is ready. However, practice shows that the first result is often far from ideal. The problem is not in the quality of the neural network, but in the nuances of text preparation that most users ignore. Let’s figure out how to avoid mistakes, choose the right model, and scale AI production to create the perfect text voiceover.
Text Preparation: The Secret to a Perfect AI Voice
A neural network reads symbols, not meanings that a human constructs from context. To avoid errors, the text must be adapted.
Dealing with Homographs and Stresses
The Russian language is rich in words that are spelled the same but have different meanings depending on the stress (e.g., “за́мок” (castle) and “замо́к” (lock)). The model chooses a variant based on statistics, which often leads to errors. There are two solutions:
Rephrasing: Change the sentence to eliminate ambiguity. For example, “Замок заело” (The lock got stuck) can be rephrased as “Дверной замок заело” (The door lock got stuck).
Placing stresses: Use special characters. In Yandex, this is a “+” before the stressed vowel (зам+ок), in most other services – Unicode U+0301 immediately after the stressed vowel (замо́к). This method is more reliable but makes the text less readable.
Numbers, Abbreviations, and Latin Script
Numbers: For correct pronunciation of numbers, especially with cases and units of measurement, it is better to write them out in words. For example, “к 15 марта” (by March 15) will become “к пятнадцатому марта” (by the fifteenth of March), and “1 250 000 ₽” will become “один миллион двести пятьдесят тысяч рублей” (one million two hundred fifty thousand rubles).
Abbreviations: Models handle abbreviations read letter by letter (НДС, МФЦ) or as words (ГОСТ, вуз) well. However, mixed constructions, like “ГОСТ Р 34.10-2012”, are often pronounced character by character. In such cases, it is better to write them out in words or move them off-screen.
Latin script: English words and abbreviations (API, OK) within Russian text may be read incorrectly. The solution is transliteration: “эй-пи-ай вернул двести о-кей” (API returned two hundred OK).
Letter “ё”: The absence of “ё” can change the meaning of a word (“все” (all) and “всё” (everything)). In literary texts and proper names, place “ё” manually.
The choice of model depends on the task. Modern neural networks offer a wide range of functionalities, from basic narrator voice to emotional rendering.
OpenAI TTS: A Reliable Workhorse
The tts-1 and tts-1-hd models offer a smooth, narrator-like reading. They are ideal for educational videos, audio versions of articles, and interface prompts. The request limit is 4,096 characters. These models are available in BotHub: 1,767.86 ₽ and 4,278.21 ₽ per million tokens, respectively.
ElevenLabs: Leader in Expressiveness
ElevenLabs v3 offers maximum expressiveness in Russian, with support for audio tags, pauses, and tempo changes. Multilingual v2 is more predictable for long texts. Flash v2.5 and Turbo v2.5 are fast, inexpensive, and process up to 40,000 characters at a time. However, the more expressive the model, the smaller the context (v3 has only 5,000 characters), which requires dividing the text into smaller fragments. All these models are also available in BotHub.
Google Text-to-Speech: For Large-Scale Projects
Google offers classic voices (Standard, WaveNet, Neural2, Chirp 3 HD) with per-character billing and Gemini TTS with per-token billing and a “Style instructions” feature for delivery control. Over 380 voices in 75+ languages are available. This service is suitable for those who already have projects in Google Cloud and need a large volume. The entry barrier is high: an account and a foreign card are required.
Yandex SpeechKit: For the Russian Market
Integrated into Yandex AI Studio since 2026. The cost starts from 1,342 ₽ per million characters. All calculations are made in rubles and include VAT, which is convenient for Russian legal entities. Ideal for voice robots, answering machines, and IVR in the Russian-speaking segment. The Brand Voice service is available for creating a unique company voice.
Local Solutions: Confidentiality and Scalability
For projects with high confidentiality requirements or a zero budget for volume, synthesis can be run locally. Examples include Kokoro 82M (without Russian language) and Fish Audio S2 Pro (80 languages, commercial use is paid). This approach is suitable for closed circuits, working with personal data, and large volumes if you are willing to manage the infrastructure.
Testing and Optimization: Deepfake-Ethical and Efficient
Voice selection is not just about timbre. Run your own paragraph through several voices to understand how they handle your content.
Voice Cloning: Ethical Aspects
The Voice Clone feature in ElevenLabs allows you to create a copy of a voice from a short recording. Technically, it’s simple, but legally complex: cloning someone else’s voice without written consent is prohibited. It is important to adhere to deepfake ethics.
Iteration Process: From First Pass to Final Render
First Pass: Record a short text fragment (paragraph). Listen carefully to how numbers, names, abbreviations, and sentence boundaries are pronounced. Correct errors in the text, not in the settings.
Second Pass: Record the full text. Evaluate the transitions between fragments and the overall pace. Models with a larger context reduce the number of transitions but may be less expressive.
Final Render: Choose the desired format. WAV or FLAC for further editing without loss of quality, MP3 or AAC for podcasts.
“Scaling was impossible — now it’s possible. Subscription-based neuro-production opens new horizons for content creators.”
Comparative Analysis: Cost of 1 Minute of AI Video and Tariffication Nuances
To evaluate the effectiveness of AI versus real UGC and understand how much one AI creative costs, it’s important to understand the tariffs.
Test Run: 289 Characters
Stress-test text: “On September 8, 2026, OOO “Yolochka” signed a contract for 1,250,000 ₽ with VAT. The lock in the warehouse jammed, and a 16th-century lock has nothing to do with it. Deadlines have already been missed, and this is an expensive lesson. According to GOST R 34.10-2012, the signature is correct, the API returned 200 OK, and the ElevenLabs v3 model read it without a hitch. Or not?”
OpenAI tts-1: 23 seconds of audio, 0.76851 ₽ charged (45,200 characters per hour, 120 ₽/hour).
OpenAI tts-1-hd: 23 seconds of audio, 1.85998 ₽ charged (45,200 characters per hour, 291 ₽/hour).
ElevenLabs v3: 34 seconds of audio, 7.68625 ₽ charged (30,600 characters per hour, 814 ₽/hour).
Conclusions on Pricing
Duration: ElevenLabs v3 reads slower, which increases the audio duration and, consequently, the cost when billed by time.
Tokens vs. Characters: Token-based billing in Russian differs significantly from character-based billing. 289 characters can turn into 435 tokens, increasing the actual cost. Always check the actual charges in the interface.
Frequently Asked Questions
How to choose the best neural network for video generation?
The choice depends on the specific task: for a narrator’s voice, OpenAI TTS is suitable; for emotional voiceovers, ElevenLabs; for large volumes within the Russian context, Yandex SpeechKit. Consider character limits, cost, and the ability to control intonations.
Can AI voice be used in commercial projects?
Yes, most services offer commercial licenses. However, when cloning a voice, always obtain written consent from the copyright holder to avoid legal issues.
How much does mass video generation by a neural network cost?
The cost depends on the chosen model, the volume of text, and the duration of the final audio. For example, for a 10-hour audiobook, the price can range from 1,200 ₽ (tts-1) to 8,100 ₽ (ElevenLabs v3). For AI UGC for a crypto project or an AI video factory for e-commerce, it is important to consider the scale and optimize the process.
Where to order AI content production?
Services like BotHub offer access to various speech synthesis models, including OpenAI and ElevenLabs, with payment in rubles and trial access. This is a convenient solution for getting started with AI avatars with subtitles, AI voiceovers, and AI clips with face swapping.
Conclusion: The Future of AI Content is Here
AI video generation and audio are not just a trend, but a new reality. With the development of technology, we gain access to tools that allow us to create mass video generation by neural networks and an AI conveyor of 500 videos per day. The key to success is understanding the nuances of text preparation and choosing the right tools. Start experimenting with BotHub today to evaluate the capabilities of neuro-production and scale your content!
Prices as of September 8, 2026:
Model or Service
Characters per request
Free
Payment in Rubles (approximate)
tts-1 in BotHub
4 096
trial access
1 767.86 ₽ per 1 million tokens (2.66 ₽ per 1000 characters)
tts-1-hd in BotHub
4 096
trial access
4 278.21 ₽ per 1 million tokens (6.43 ₽ per 1000 characters)
Eleven v3 in BotHub
5 000
trial access
26.60 ₽ per 1 000 characters (by measurement)
Eleven Multilingual v2 in BotHub
10 000
trial access
17.53 ₽ per 1 million tokens
Eleven Flash v2.5, Turbo v2.5 in BotHub
40 000
trial access
8.76 ₽ per 1 million tokens
OpenAI directly
4 096
no
$15 per 1 million characters, HD $30
ElevenLabs directly
depends on model
10,000 credits/month (without commercial license)
from $6 per month
Google Standard, WaveNet
depends on model
up to 4 million characters per month
$4 per 1 million characters
Google Chirp 3 HD
depends on model
up to 1 million characters per month
$30 per 1 million characters
Google Gemini 3.1 Flash TTS
depends on model
32,000 tokens
$1 per 1 million text tokens + $20 per 1 million audio tokens
Yandex SpeechKit, API v1
not specified
according to AI Studio terms
1 342 ₽ per 1 million characters including VAT
Yandex SpeechKit, API v3
not specified
according to AI Studio terms
0.1626 ₽ per request (long requests are counted proportionally)
