Short answer: video ads work — but most people lose money because the creative is wrong, not because the platform is broken. This page shows you the documented numbers, the case studies with links to the source, the exact research method we run for every client, and an honest calculator so you can judge the math for your business before you spend a dollar.
Read this first: every benchmark below is published third-party data, cited and linked. The case studies are named, documented businesses — not our clients. We produce creative; we do not run your ad account or promise results. Anyone who promises a revenue number is selling you something.
The headline stats
$1.92
median cost per lead across industries (WordStream, 2024)
6–8
ad variants Meta recommends per campaign for stable delivery (Meta)
$6.1B
projected global AI-influencer market by 2026 (industry estimate)
01 · The benchmarks
Before anyone talks about "life-changing money", you need the boring numbers. These are median figures published by ad platforms and independent benchmark reports. They are the floor you're playing against — video creative is how you beat them.
$1.92
Across all industries in WordStream's 2024 Facebook Ads benchmark. Campaigns at the 25th percentile average $1.13; the 75th percentile sits at $9.20 — a 8× spread driven almost entirely by creative and offer.
WordStream benchmark report$0.83
Median Facebook CPC. Video ads typically earn the lowest CPC of any format because the first frame does the qualifying work before anyone taps it.
WordStream benchmark report9.21%
Average landing-page conversion rate from Facebook traffic in WordStream's cross-industry dataset — the step most advertisers neglect while blaming the ad.
WordStream benchmark report56%
Share of ad performance variance that Meta's own marketing science attributes to creative. Targeting and bidding are largely automated now — the video is what's left for you to control.
Meta for Business research3–5×
Typical engagement lift for short-form vertical video versus static image creative on Reels and Feed placements — the single clearest reason ad budgets shift to video every year.
Statista video ad outlook6–8
Number of distinct creatives Meta recommends per ad set to exit the learning phase and find a winner. One ad, one angle, one lucky launch is why most first campaigns fail.
Meta Ads GuideFigures reflect published reports at the time of writing and move month to month by industry, country and offer type. Treat them as starting benchmarks, not predictions for your account.
02 · Documented case studies
These are not testimonials we wrote. They are businesses that have published their numbers, been written up by the trade press, or filed them publicly. Read the mechanics of how each one worked, then check the source yourself.
Fashion-forward AI characters began appearing on brand Instagram and TikTok accounts, doing what a paid human model does — wearing the product, in vertical video, on a posting schedule — for a fraction of a shoot budget. The pattern was suspicious because it worked: no location, no model day rate, no reshoot when the colourway dropped.
What made it work: the character was treated exactly like a real talent account — consistent face, consistent wardrobe lane, consistent posting cadence — and every clip was cut for Reels/TikTok first, not cropped from a horizontal video.
Business Insider reportShoot cost eliminated
$0 model + studio
Ongoing output
Posting cadence held
Margin moved on the cost side first. Revenue followed because the account never went quiet.
A beauty brand founder publicly announced an AI-created virtual influencer as a standing "employee" creating content for the brand. The point she made is the one most business owners miss: the asset isn't the video, it's the fact that the same face can appear for years, in every country, at every launch, with no scheduling conflict.
What made it work: the character was named, given a job description and a personality — so the audience formed a relationship with a consistent identity instead of a rotating stock model.
Vogue Business coverageCharacter reuse
Unlimited, worldwide
Posting reliability
Every launch
This is the honest version of "passive income": you build the asset once, then keep using it.
She has no physical body and has held brand partnerships with fashion and beauty houses for close to a decade. Whatever you think of the aesthetic, the business conclusion is unambiguous: an AI-created character can build and hold a genuinely large, engaged audience — which is the only asset an influencer offer actually monetises.
What made it work: a defined style, a storyline and relentless consistency of the face across thousands of posts. It was never about one great video.
Documented profile & partnershipsAudience built
Multiple millions
Commercial model
Brand partnerships
Sponsorship revenue scales with attention, not with headcount.
For every case study above there are thousands of accounts that posted three AI clips, got 200 views and quit. Reporting on the virtual-influencer space has documented both the headline wins and the crowded, low-quality middle. That is the honest picture, and it is why we put the method before the output.
The difference between the two groups is almost never the AI model. It is: did the character have a defined positioning, were there enough variations to test, and was the offer clear in the first two seconds. Those three things are exactly what our research phase exists to fix.
WIRED on the virtual influencer economyNone of the businesses above are our clients, and we do not claim credit for their results. They are included because their numbers are public and verifiable. Individual results vary with offer, budget, market and execution.
03 · The niche leaders, and what they earn
Not projections. These are the categories where AI-created characters have documented audiences and documented revenue models — and what the earnings actually look like at each level. If your business sits in one of these categories, this is the most useful page on our site.
Fitness is the strongest niche for AI character content because the buyer intent is already proven: people pay monthly for coaching, programmes and supplement routines. A character can demo a workout, hold a product, and post daily — the exact output a coaching business needs and never has the time to produce.
Why it converts: transformation content is inherently visual and repeatable. The same character across months of posts builds the "I've watched her progress" effect that a one-off ad can never create.
Typical revenue model
Coaching subscriptions · programmes · supplements · affiliate
Earnings range observed
$2k – $40k+ /mo
Wide range because it maps to audience size and offer price, not to the creative itself.
Beauty brands were the first to hire AI characters outright, because the content requirement is enormous and constant: new shade, new routine, new season, every fortnight. A single consistent face can carry a product line for years without a shoot budget.
Why it converts: beauty is close-up, product-in-hand content. That is exactly what AI production is strongest at, and what brands pay retainers for.
Typical revenue model
Brand partnerships · sponsored posts · affiliate · own product line
Per brand deal, documented
$1k – $10k+
Per campaign at established audience sizes. Volume, not price, drives the total.
This is the category that produced the most famous AI influencers. Character-led fashion accounts have held long-term deals with luxury and high-street houses for years, and the reason is purely operational: try-on content at scale, per colourway, per market, without a model booking.
Why it converts: garments sell on how they look on a person. A character solves the single most expensive part of that equation.
Typical revenue model
Long-term brand contracts · lookbooks · drops · licensing
Cost side, per shoot saved
$1k – $15k
What a model + studio + location day typically costs. That saving repeats every drop.
Here the maths is not about followers at all — it is about cost per lead versus customer value. A roofing, HVAC, dental, gym or salon business spending on vertical video ads is measuring everything in booked jobs. This is where a business owner should look first, because you do not need an audience to start.
Why it converts: the ad does one job — problem, fix, call to action. No character-building required, and the return is measurable in weeks rather than months.
Typical revenue model
Booked jobs · enquiries · redemptions · direct sales
One customer is worth
$150 – $8k+
Depending on the trade. One added job often covers the whole package.
The household-name AI influencers sit here, and they earn the most per deal — but they also took years to build. This niche rewards patience and a genuinely distinctive character. It is the highest ceiling and the longest runway.
Be honest with yourself: if you need revenue in the next 60 days, start with ads or a fitness/beauty character instead of chasing entertainment fame.
Typical revenue model
Sponsorships · music releases · merch · licensing
Time to real money
12–36 mo
Highest earnings per deal, but the slowest to get there. Not a shortcut niche.
What we will not do: pretend those ranges are a promise to you.
Earnings in every niche above track audience size, offer price and how consistently the account posts — none of which a package can hand you. Anyone showing you a guaranteed income figure for an AI influencer is selling you something other than reality. What a package actually buys is the asset: the character, the production quality, and the volume of content you need to test.
What we will do is tell you, before you pay, which niche your business actually belongs in — and if the honest answer is "run ads, not an influencer", that is what we will say.
Figures above are ranges compiled from publicly reported brand deals, published package pricing across the virtual-influencer market, and documented agency cost benchmarks. They are illustrative of what is achievable in each category, not a forecast for your business. Results vary with offer, budget, market and execution.
03 · Our research method
Every Meta ad that has ever run — on Facebook, Instagram, Messenger and Threads — is publicly searchable in the Meta Ad Library. That database is the most honest creative brief in existence, because brands do not keep paying for ads that lose money.
Before we write a script for your AI influencer or your ad, we go live in your niche and reverse-engineer what is already spending.
Why long-running ads are the signal
An advertiser running the same concept for 60, 90, 180+ days is voting with their own money every single day. A brand new ad might be a test. An ad that has survived a quarter is a winner that has been optimised — and its structure is copyable.
Filters: your country, "Active ads", and a keyword a customer would use — "roof replacement", "meal prep", "lash extensions". Industry-level keywords return noise. Service-level keywords return your competitors' actual campaigns.
We pull the ads that started running months ago and are still active — the "Start date" filter plus the ad's own live-since date. Anything past 60–90 days goes on the shortlist. This is the closest thing to public proof-of-profit that exists.
We transcribe the hook, then measure it: on-screen text, first spoken line, visual movement, and whether the product appears before second three. Survivors almost always pass the scroll-stop test inside two seconds.
We tag each surviving ad with the persuasion angle it runs: problem–agitate–solve, before/after, price anchor, social proof, myth-busting, testimonial or curiosity. Three or more long-running ads on the same angle tells you what the market responds to.
The ad is only half the machine. We open each winner's destination and note the offer, the price framing, the form length and the call-to-action wording. A great video pointed at a weak page burns money.
You receive creative built on the proven angles — not one lucky video. That is deliberate: Meta needs 6–8 variants to exit the learning phase and find your winner, so we hand you a testable set instead of a single gamble.
Want to try it yourself? Open the library, search your service, and sort by start date.
Open Meta Ad Library04 · See it for your business
Pick the type of service or product you sell, set your budget, and see a realistic lead and revenue range based on published benchmarks for that industry — not a fantasy slide. Every figure updates live.
For home services, use average job value or lifetime value.
Est. cost / lead
$—
Leads / month
—
New customers
—
Revenue
—
Return on ad spend
—
A ROAS above 1.0 means the spend came back as revenue; the break-even line depends entirely on your margin.
These are modelled ranges from published industry lead-cost benchmarks, adjusted for creative quality. They are not a forecast, quote or guarantee for your account.
05 · The niche reality check
"How much do they make?" is the right question — most people answer it with a screenshot and no context. Here is the honest version: typical public lead-cost benchmarks per industry, the realistic monthly upside once a character has momentum, and the time it actually takes to get there.
| Niche | Typical lead cost | Value per customer | Realistic monthly upside | Time to traction |
|---|---|---|---|---|
|
Home services roofing, HVAC, remodel |
$60–$180 | $1,500–$12,000 | 5–20 extra jobs | 3–6 weeks |
|
Beauty & aesthetics salon, injectables, lash |
$15–$60 | $120–$900 | 40–200 bookings | 2–4 weeks |
|
Fitness & coaching gyms, PT, online programs |
$12–$45 | $150–$2,400 | 30–150 sign-ups | 2–5 weeks |
|
Legal & professional injury, family, accounting |
$80–$350 | $1,000–$15,000 | 3–15 cases | 4–10 weeks |
|
Restaurant & food cafe, takeaway, catering |
$8–$35 | $30–$400 | Orders & covers up 10–40% | 1–3 weeks |
|
E-commerce product skincare, apparel, gadget |
$10–$40 | $35–$220 | 1.5–4× ROAS | 2–6 weeks |
|
Real estate agent, developer, lettings |
$40–$200 | $3,000–$20,000+ | 1–6 listings / deals | 4–8 weeks |
|
B2B / SaaS agencies, software, services |
$50–$250 | $500–$25,000 | 3–20 qualified calls | 4–12 weeks |
The 60-day truth
Most accounts that succeed see their first profitable stretch between weeks 3 and 8. Anything promising day-two revenue is selling, not informing.
Volume is the variable
One character posting twice a week underperforms. The accounts that win publish consistently and test multiple hooks per month.
What we control
We control production quality, research and delivery speed. Your offer, budget and follow-up control the outcome.
Ranges are compiled from published lead-cost benchmarks and typical industry deal values. Low-end estimates assume cheap leads with weak close rates; high-end assumes good creative, strong follow-up and repeat customers. Your results depend on your own numbers.
06 · Included with every order
Whether you want a custom character, want to become the AI influencer yourself, or just need one ad that converts — our team analyses the option with the best odds, then builds it. Here is exactly what happens after you submit your brief.
Age, look, tone, wardrobe lane and niche — chosen from what already performs in your category, not from a template folder. Want to be the AI influencer yourself? We build the character from your reference photos with your real facial structure.
The long-running ads in your service category are pulled, tagged by angle, and turned into the creative brief — so your first video starts from a proven hook instead of a guess.
First two seconds designed to stop the scroll, the offer stated in plain language, and a single call to action on the final frame. No vague brand films.
Native captions, clean cuts, correct vertical framing, branding placed where platform UI won't cover it, and delivery in the aspect ratios you actually need for Reels, TikTok and Shorts.
Angles rotate across your package so you can test properly — the way Meta's own guidance recommends. That's how you find the winner instead of guessing at it.
We stay on it until the character reads right and the video is genuinely postable. Your package includes defined revision rounds — no hidden negotiation to fix something that isn't good enough.
Not sure which route fits you — character, your own face, or a one-off ad? Send your service and target market and we'll tell you which has the best odds before you spend anything.
07 · Why you can trust this page
You've seen the "5.0 stars from 400 happy customers" badges on sites that launched last month. We'd rather earn the purchase with things you can actually verify.
Sources are linked
Every number on this page points to a public report, an official platform document or documented press coverage. If a claim can't be sourced, it isn't here.
Work shown, not claimed
Our portfolio videos play on the page. You can judge the production quality yourself rather than taking a written description's word for it.
Watch the workFixed prices, published
Every package price and deliverable count is public before you commit. No quote gate, no sales call required to find out what things cost.
See all pricesClear revisions, in writing
Revision rounds and delivery windows are part of the package terms and the Terms of Service — not something you discover after paying.
Read the termsA real human replies
Questions go to a monitored inbox and a contact form that reaches an actual person — not a chatbot loop pretending to be support.
Contact usNo invented results
We don't publish revenue figures we can't source, and we don't call agency concept work a client success story. That restraint is the proof.
Common questionsVertical-first delivery
9:16 for Reels, TikTok, Shorts
Commercial usage included
Run the creative as your own
24-hour start
Production begins next business day
Secure checkout
Card payment via Square
08 · The questions buyers actually ask
If one of these is the reason you haven't bought yet, it's answered below.
Send us your service and target market. We'll research what's already turning into leads in your category, come back with the character or ad direction with the best odds, and produce it ready to run — so your first campaign starts from proven angles instead of guesswork.
Benchmark data referenced on this page is published by WordStream, Meta, Statista and cited trade press, and reflects conditions at the time of writing. Case studies describe publicly documented third-party businesses and are not our clients or our results. Nothing here is a guarantee of income, leads or sales.