The gap between the advertisers making money and the ones burning it is not budget. It is not targeting. It is not the agency they hired.
Brands testing 20 or more new ads per month see 65% higher return on ad spend than brands testing fewer than 10, according to Common Thread Collective's data across more than 170 ecommerce brands.
Twenty. Every month. Which sounds impossible if you think an ad is something you commission, and completely achievable once you build it as a production line.
So how does a small business actually test twenty new ads a month? You stop making ads one at a time. You write your angles first, shoot a full quarter of creative in one structured day, and release five genuinely different ads a week from templates.
Here is the factory.
First, define what counts
Twenty new ads does not mean twenty exports. Modern ranking systems cluster near identical creative together, so twenty crops of the same photo get treated as roughly one idea. If you want the background on why, Meta's Andromeda update is the reason a pile of lookalike variants now buys you almost nothing.
A genuinely new ad differs on at least one of four dimensions.
- Hook. The first three seconds and the first line of copy.
- Format. Static, carousel, talking head, screen recording, customer selfie, before and after, motion graphic.
- Presenter and setting. Who is on camera and where they are standing.
- Angle. The actual argument. Speed, price, risk reversal, proof, or the cost of getting it wrong.
Change one of those and you have a new ad. Change the font and you have a duplicate.
Twenty variations of one idea is one audition in twenty costumes. Twenty ideas is twenty auditions. The system can tell the difference even when you cannot.
Of the four, angle does the heaviest lifting. A new hook on an old argument is a small test. A new argument is a big one. When you are short on time, spend it on angles.
How to write fifteen angles in one sitting
Everything downstream depends on this page of notes, so do it properly. Block out ninety minutes, close the laptop lid on everything else, and work through it in order.
- List every reason people buy. Not what you think they should care about. What they actually said when they chose you. Dig through reviews, sales call notes and old emails. Write each reason as one plain sentence.
- List every objection. Too expensive. Not sure it works for me. Tried something like this before. Bad timing. Each objection is an angle, because an ad that answers it head on speaks to everyone quietly holding it.
- List every proof you can film. A customer who will talk. A before and after. A number from your own records. A process that looks impressive up close. Proof angles are often the strongest and the most overlooked.
- List the cost of doing nothing. What happens to the customer who never fixes the problem? This is the angle most businesses are too polite to run, and it often wins.
- Cut and merge. Cross out duplicates. Merge anything that makes the same argument. What is left is your angle bank.
If you finish with fewer than fifteen, go back to step one and ask your front desk or sales team. The people who answer the phone hear objections the owner never does.
The one day shoot that produces a quarter
This is the part that makes the maths work. You do not produce twenty ads a month. You produce sixty ads a quarter in a single structured day, then release them weekly. We break the same idea down in detail in one video shoot, 90 days of content.
Before the day. Write your angles first, not your shot list. Sit down and list every distinct reason someone buys from you, every objection that stops them, and every proof you can put on camera. Aim for fifteen. Most businesses stall at six and that stalling is the actual problem.
Book real people. Two customers, one team member, yourself. Real customers are the highest performing asset you can film and the hardest to organise, so book them first.
The day itself. Phone on a tripod, clip microphone, natural light or one cheap panel. Three locations, on site, indoors, and somewhere with life in the background.
Shoot each angle four ways. Straight to camera. Over the work. As a customer answer. As a quick demonstration. Four hooks times four formats across three settings gives you far more distinct combinations than you will use.
Capture the boring footage. Hands working, tools, the van arriving, the mess before, the finish after. This is what makes every other clip credible and it costs nothing to gather while you are there.
Run it like a set, not a hangout. Print a running order with times against each angle. Give customers their slot and a rough question list so they are in and out in thirty minutes. Check audio on playback after the first take in every location, because bad sound is the one thing you cannot fix in the edit.
A worked example: a plumber with one phone
Say you run a three-van plumbing business in Brisbane spending $4,000 a month on Meta ads. You currently run four ads, all polished, all saying some version of "fast, reliable, local."
You sit down and write your angle bank. Reasons people buy: you turn up in the window you promise, you give a fixed price before starting, you clean up after. Objections: plumbers are expensive, they never show up, they upsell. Proof: a customer who had a burst pipe on a Sunday, the before and after of a bathroom rough-in, the tidy van. Cost of doing nothing: a slow leak behind a wall turning into a mould and flooring job. That is already more than a dozen distinct arguments.
On shoot day you film yourself straight to camera on the fixed price angle, a customer telling the Sunday story, your apprentice walking through the van, and hands-only footage of a leak repair. Every angle gets four treatments. By 3pm you have enough raw material for three months.
Each week five new ads go live. After a few weeks, one stands out: the customer telling the Sunday story, filmed on their own front step. So you ask what made it work. Was it the real customer, the emergency angle, or the outdoor setting? Next week you run five more emergency stories with different presenters and settings. That single discovery now feeds the whole next quarter.
The weekly release rhythm
Do not dump sixty ads into the account at once. Two reasons.
The system needs enough spend per creative to learn anything, and flooding it splits budget into slices too thin to read. And ad fatigue windows are now around two to four weeks, so a steady stream matched to that cycle keeps performance alive where a single dump peaks and then collapses together. We cover that clock in why your winning ad now dies in 14 days.
Release five new distinct ads per week. Kill anything that has clearly lost after enough spend to judge. Keep a rolling set of proven, testing and queued.
At any moment you want roughly fifteen to twenty genuinely distinct live ads. That is the shape that consistently performs, not four hundred variants of eleven ideas.
Editing without an editor
The bottleneck is usually post production, and it should not be.
Build three templates once. A talking head with captions. A before and after with a text overlay. A customer clip with your branding at the end. Then every new video is an assembly job rather than a design job.
Captions on everything, because a lot of viewing happens with the sound off. Vertical as the master format. Keep the first three seconds clean of logos, since a brand mark in the opening frame is a reliable way to get scrolled past.
Batch the editing the same way you batched the shooting. One session, one template, twelve exports.
Reading the results properly
With twenty new ads a month you will produce a lot of losers. That is the point. You are buying information, and most of it is confirmation that an idea does not land.
What you are hunting is the outlier. In many accounts one ad performs dramatically better than the rest, and it is almost never the one anyone predicted in the meeting.
When you find it, do not just scale it. Interrogate it. What was the hook, the angle, the person, the format. Then produce five more ads that share that single winning attribute and vary everything else.
To make that possible, name every ad by its attributes, not by the date. Something like angle, format, presenter, setting. When the results come in you can sort by any column and see patterns instead of guessing. Judge on the number that pays the bills, cost per purchase or cost per qualified lead, not clicks or thumb-stop rate on their own. A cheap click on the wrong promise is still the wrong promise.
That is how a creative programme compounds rather than restarting from zero every quarter. It is also why your ROAS problem is usually a creative problem and not a targeting one.
The mistake that wastes a shoot day
One warning, because it happens constantly. Businesses book the day, film for six hours, and come away with footage they cannot use.
The cause is always the same. They shot assets without writing angles first, so they have forty clips that all make the same argument from slightly different distances.
Write the fifteen angles before you touch a camera. Print them. Tick each one off on the day. If you leave with fewer than twelve genuinely different arguments captured, the day was expensive footage rather than a quarter of advertising.
The honest cost
One shoot day per quarter. One editing session per week, ninety minutes. One review session per week, thirty minutes.
That is roughly two days a quarter plus two hours a week, and it replaces the retainer most businesses pay for four ads a month that all look the same.
The businesses winning on paid social right now are not the ones with the biggest budgets. They are the ones who turned creative into a routine instead of an event.
Your first week
You do not need the whole machine running by Friday. You need the foundations.
- Monday. Write the angle bank. Fifteen arguments, one line each.
- Tuesday. Call two happy customers and book them for a shoot date within the month.
- Wednesday. Build your three edit templates.
- Thursday. Set up a naming convention and a simple sheet for proven, testing and queued.
- Friday. Lock the shoot day in the calendar.
If you would rather have the factory built for you, that is exactly what our advertising team does: angle strategy, shoot days, templates and a weekly testing rhythm, handed over so it keeps running. Either way, put the shoot day in the calendar. That is the whole first step.
Frequently asked questions
How many new ads should I test per month?
Aim for twenty or more genuinely new ads a month if your budget can give each one enough spend to judge. Common Thread Collective reports that brands testing at that volume see 65% higher return on ad spend than those testing fewer than ten. If your budget is small, test fewer ads but keep every one distinct in hook, format, presenter or angle.
What makes an ad genuinely new rather than a variation?
A genuinely new ad changes at least one of four things: the hook, the format, the presenter and setting, or the underlying angle. Swapping a font, colour or crop does not count, because ad ranking systems group near identical creative together and treat it as one idea. The angle, meaning the actual argument, is the most valuable dimension to change.
Can I make twenty ads a month without a video editor?
Yes. Build three reusable templates once, a captioned talking head, a before and after with text overlay, and a customer clip with branding at the end. Every new video then becomes an assembly job. Batch the work into one weekly session of about ninety minutes, export vertical as the master format, and caption everything.
How long should I run a new ad before turning it off?
Run it until it has had enough spend to judge against your real goal, such as cost per purchase or qualified lead, then kill it if it has clearly lost. Do not decide on a day or two of data. With fatigue windows now around two to four weeks, even winners need replacing regularly, so keep new ads queued.