Ads that write themselves, prove themselves, and cut themselves.
Most small businesses run ads by hand: a few creatives, a monthly check-in, and no honest link between spend and revenue. The Customer Factory runs that loop every day. It mines real customer pain points, writes and renders ad variants in your brand style, waits for your approve or reject in Slack, publishes to Meta under hard spend caps, ties every dollar to sign-ups and revenue, and kills losers and scales winners on a schedule — while pulling fresh angles from competitors' live ads each week.
- 2
- of our own products run on it today
- Daily
- research → creative → approve → publish → optimize loop
- < 1 day
- to onboard a new brand onto the factory
What you get
Outcomes, not features
Ten new Facebook and Instagram ad variations generated every morning in your brand style, waiting in Slack for a 15-minute approve or reject.
Ads that are not converting get paused automatically after a three-day window; winners get more budget, never exceeding your daily cap.
One dashboard tying ad spend to sign-ups and actual revenue — Stripe, your CRM, or your booking system — not just clicks.
Fresh ad angles every week pulled from competitors' live ads and the YouTube channels your customers already watch.
A Monday growth brief and ready-to-post founder social drafts in your voice.
Is this for you?
Who this is for
This is an engagement, not a $100-a-month tool. It makes sense when the alternative on the table is hiring people to run paid acquisition — and paying for their tools and the media on top.
A good fit when
- You already spend on paid ads, or are ready to fund a media budget the factory can learn from — roughly $1,500 a month and up is where daily optimization starts to pay for itself.
- You have an offer people can buy, trial or sign up for online, and a page to send them to (or we build one first).
- Someone on your side can approve creative for 10–15 minutes a day. That is the whole job.
- You want to own the system, the data and every ad at the end — not rent a relationship.
Not yet, if
- No clear offer or landing page yet. Start with a Mini-Site or a custom app; the factory needs somewhere to send people.
- No media budget. The engine is only as good as the results it gets to learn from.
- You want a self-serve subscription. There isn't one; every factory is built to spec.
Not sure which side you are on? That is what the discovery call is for.
What you are choosing between
Your alternative is hiring humans at full rate, maybe augmented by AI — or a system that is already proving itself and works while you sleep.
| Option | What you get | What it costs |
|---|---|---|
| Hire a marketing manager | One person, business hours, plus a designer, tools and the ad budget on top. | $166,790 median salary; ~30% more in benefits |
| Hire a marketing specialist | Runs the ads by hand; a handful of creatives a month; learns when they have time. | $78,760 median salary; ~30% more in benefits |
| Performance agency | Monthly reports, platform metrics, 4–8 weeks of onboarding; the relationship is what you keep. | Typically a retainer plus a share of spend |
| Customer Factory | Researches, writes, publishes, measures and optimizes every day. Ten minutes of your time. Works while you sleep. | Custom build you own; optional support retainer; media paid straight to Meta |
Source: U.S. Bureau of Labor Statistics, Occupational Employment and Wages, May 2025 (marketing managers; market research analysts and marketing specialists). Benefits share: BLS Employer Costs for Employee Compensation, 2026 (~31.5% of total compensation).
Results so far
Running today on two of our own products
- 249
- ads written, rendered and quality-checked
- 165
- pause and budget decisions made on its own
- 17.9%
- of clicks became leads on the first brand, against a 5.4% category average
- $9.71
- per signup on the best ad; about $21 average across 39 signups
- $0.24
- per click on the second brand, across 1,359 clicks
- 1 day
- from onboarding the second brand to live ads
Not proven yet: return on paid revenue. What it did prove within two weeks was that the first brand's gap was product onboarding, not the ads — the kind of truth a clicks-and-CPC report never tells you.
Our BigQuery warehouse, 22–23 September 2026. Three weeks of data, roughly $1,200 of media across both brands, both accounts still in Meta's learning phase. Directional.
Against industry benchmarks
Our numbers next to the published averages
Same engine, different bottlenecks — each one found in weeks because every result is tracked to the funnel, not the click.
Brand one vs. employment lead ads
| Metric | Ours | Benchmark |
|---|---|---|
| Cost per click | $1.96 | $0.73 |
| Click to lead | 17.9% | 5.4% |
| Cost per lead | $10.99 | $12.30 |
Expensive clicks, strong page: leads still beat the category. Lever: creative.
Source: WordStream / LocaliQ Facebook Ads Benchmarks 2026, Leads objective, Career & Employment.
Brand two vs. finance traffic ads
| Metric | Ours | Benchmark |
|---|---|---|
| Cost per click | $0.25 | $0.86 |
| Click-through rate | 1.27% | 1.46% |
| Click to lead | 0.7% | 8.5% |
Very cheap clicks, weak page. Lever: landing page and offer — which is exactly what the funnel data flagged.
Source: WordStream / LocaliQ Facebook Ads Benchmarks 2026, Traffic objective, Finance & Insurance.
Creative volume vs. $1.3B of Meta creative data
| Metric | Ours | Benchmark |
|---|---|---|
| New ads launched per week | 14–45 | 2–3 (small accounts); ~5 (top quarter) |
| Spend behind the top 10% of ads | 58–60% | 23% (small); 53–57% (large, staffed) |
| Best format's share of spend ÷ share of ads | 3.3× | 2.1× (the study's top format) |
Winners are about 5% of ads everywhere. Accounts that launch more creative find more of them. Volume is the whole game, and the factory has volume.
Source: Motion, Creative Benchmarks 2026 — 578,750 ads, 6,015 accounts, $1.29B Meta spend, Sep 2025–Jan 2026. Ours: both brands combined, 23 Sep 2026.
How it runs
The loop, step by step
This is the actual pipeline we run for ourselves, not a diagram. A human approves at the points that matter; the machine does the rest.
Built to grow our own products. It runs the paid acquisition for oneresume.ai and optionscout.ai every day.
- 01
Research
Mines forums, reviews and competitor ad libraries for the exact language your customers use to describe the problem, and keeps an evolving bank of angles.
- 02
Creative
Writes copy variants against your brand guide, runs a direct-response judge tournament to keep the top half, renders images, and vision-checks each one for brand compliance.
- 03
Approve
Posts ad previews to a Slack channel with Approve and Reject buttons. Nothing goes live without a human click.
- 04
Publish
Creates the approved ads in Meta paused-first, under a daily spend cap and a maximum budget change per day, with a one-command kill switch.
- 05
Measure
Syncs Meta, analytics and your revenue source nightly into a warehouse, so every ad is judged on funnel and revenue, not vanity metrics.
- 06
Optimize
A rules engine kills the bottom performers after a three-day window, rebalances budgets, distills what worked into the next day's prompts, and posts a digest.
- 07
Entropy
Weekly injection of competitor ads and industry video transcripts so the creative never goes stale.
Once it's running
A day in the life
Your part is ten minutes over coffee. Every rejection becomes a written rule the generator must follow, so approvals get faster over time.
| When | What happens |
|---|---|
| 4:00 am | Yesterday's ad spend, signups and payments land in your warehouse |
| 5:00 am | Every ad is ranked; losers paused, winners funded; a digest is posted |
| 6:00 am | New ads written, designed, judged and image-checked |
| 6:30 am | Ad previews land in Slack for approval |
| Your coffee | Ten to fifteen minutes of Approve / Reject |
| 4:00 pm | Approved ads go live on Meta |
| Mondays | A plain-English memo: what's converting, what only looks good |
Compare
Why this, and not an agency or a hire
| Performance agency | In-house team | Customer Factory | |
|---|---|---|---|
| New ads each week | A handful a month | One designer's output | 14–45 launched, ~50 made |
| Ads judged on | Clicks, platform ROAS | Same platform numbers | Leads, trials, paid customers |
| Learns and acts | Monthly report | When someone has time | Every morning; rules persist |
| First ads live | 4–8 weeks onboarding | 2–3 months to hire | ~3 weeks after accounts |
| Your time | Weekly calls and briefs | Managing people | 10–15 minutes a day |
| Yours after a year | A relationship | A person who may leave | System, data, every ad |
Cost is deliberately not a row. The argument is the six above; the quote comes from a discovery call.
Guardrails
You stay in control
You approve every ad
Nothing runs until you tap Approve, for as long as you want it that way.
Hard spend cap
Checked in code before every budget change. The engine cannot raise spend past it.
Claims rules in code
Banned claims and off-brand words are blocked before you ever see the ad.
One-command stop
A kill switch pauses every ad at once.
Account-safe API use
It only writes to Meta; reports are pulled once a day. No scraping, no risky automation.
Special-category ready
We have run Employment and strict financial-claims accounts already; Housing is checked in discovery.
Timeline
What you hold at the end of each phase
Deliverables you can point at, not activities. The measurement chain is proven with a test signup before a dollar of media is spent.
Typical, not fixed. The phases are fixed; the calendar is not. Our fastest build went from onboarding to live ads in a day, because the brand was already on our stack. A new cloud project, a tracking change in a codebase we have not seen, or Meta's business verification can each add one to three weeks. Every scope is set integration by integration on the discovery call.
- Typically 1 week
01 · Blueprint
Readiness report on ads account, pixel, analytics and billing; personas, claims policy and the free hook; first ten angles; fixed scope.
You
Ad account access, one interview
- About 1 week
02 · Foundation
Your cloud project: warehouse, creative storage, scheduled jobs, secrets, alerts; Slack app live; one job verified end to end.
You
Cloud project, billing, keys
- 1–2 weeks, stack-dependent
03 · Tracking
Pixel and server events, ad id carried through signup and checkout; a test signup visible in the warehouse; brand kit and angle bank built.
You
Merge one small change
- About 1 week
04 · Pre-launch
First campaign live at a small cap; first ad batches in Slack; morning digest and per-ad scoreboard running.
You
Approve ads daily
- 2–3 weeks
05 · Launch
Main campaign live; optimizer proposals reviewed together, then execute mode; a written read of which personas and formats convert.
You
Approve; weekly review
- About 1 week
06 · Handover
Runbook, secrets inventory, key rotation, access review, recorded training. Optional support retainer starts.
You
Sign off
Proof
Built on our own products
We don't sell a factory we haven't run. Every one of these grows something in our own portfolio first.
Said plainly
Three weeks and about $1,200 of media is not a case study. It is enough to show the loop runs, judges ads on funnel events instead of clicks, and finds the real bottleneck fast. Paid return is the number we are still earning.
What you'll need
- A Meta Business Manager (Facebook and Instagram ads) we can connect to
- Your analytics and a revenue source — Stripe, your CRM, or a booking tool
- A brand guide, or an hour with us to write one
- A Slack channel and one person who will approve creative each morning
Don't have all of it? That's normal. Most of the list gets set up during the build.
How we work
How an engagement works
Custom pricing, scoped on a discovery call. You own what we build. Optional support retainer after launch.
- 01
Discovery call
Thirty minutes. You describe the job the factory has to do; we tell you honestly whether one of ours fits, needs adapting, or should be built from scratch.
- 02
Scope
A written scope with the systems we connect to, the approvals you keep, and a fixed quote. Custom pricing — every factory is built to spec.
- 03
Build
We build on the same stack and playbooks that run our own products, in your accounts, with you signing off on each piece before it goes live.
- 04
You own it
The code, the accounts, the data and the prompts are yours at handoff. No platform lock-in, no subscription to us to keep the lights on.
- 05
Support (optional)
A monthly support and tuning retainer once launched — monitoring, model upgrades, new angles — for as long as it earns its keep. Cancel anytime.
Want a Customer Factory of your own?
Thirty minutes to find out whether it fits, what it connects to, and what it takes to build.