We read 119 ads that competitors are running on Facebook and Instagram right now, worked out whether the numbers add up, and wrote a plan. This page is the short version.
Competitor ads read
Companies advertising
Client slots to fill
Suggested first spend
Everything else on this page is detail. If you read only one section, read this one.
Eighteen competitors offer a free audit and some version of "you don't pay if it doesn't work." So saying it won't make us stand out. Being unusually specific will.
The two longest-running ads have been going two years. Both are quiet, simple, and make no big claims. Every loud ad with huge numbers is only a few months old, so nobody knows yet if they work.
One competitor published their own test results in an ad. Plain text won about 12% of the time. Expensive polished video won 6%. Animation won 3%. The better it looked, the worse it did.
Not a feeling. The best one we found: "You shipped 50 products to influencers. 7 posted. You paid for all 50." A brand owner can check that against their own business in five seconds.
Not one of the eight best ads showed a dashboard. Every figure sat in the written text. That matters because we can't show Meta results — and it turns out the best advertisers aren't showing screenshots anyway.
Three prices in about ninety-five ads. Everyone else hides behind a call. Our published pricing is unusual enough to be an angle on its own.
Yes, with room to spare. Here is the arithmetic in full.
What one client is worth, at $2,000 a month for six months
All three slots together
What we can afford to pay for one enquiry
What we should actually spend first
A healthy business spends around 30% of what a client is worth to win them, which puts our ceiling at $315 per enquiry. Published 2026 figures for business-to-business Facebook advertising put the real cost at $15 to $80, and $40 or more for professional services like ours. So we have room — but two warnings. Those figures are for forms filled in inside Facebook; sending people to our own page costs three to five times more per enquiry, which lands much closer to the ceiling. And costs rise through the year: an average enquiry cost $36 in the first quarter and $46 by the fourth.
The full budget is justified, but there is no reason to commit it before we know our real cost per enquiry. Three thousand pounds' worth of spend buys that answer. If the number comes back where we expect, the rest is available and earned.
All of these are live right now. Each card opens the real ad on Facebook, where you can see the picture or video, the full text and where it sends people.
Three groups: A is our actual competition, agencies selling to real brands. B sells "passive income stores" to beginners and only clutters the same keywords. C is software. Sorted by how long each ad has been running, because an ad nobody has switched off is probably making money.
Everything above comes from ads still running. That is like judging an industry by the companies that survived it. Facebook also keeps ads after they are switched off, so we went and counted the failures too.
| Company | Still running | Switched off | Share killed |
|---|---|---|---|
| Chase Chappell | 163 | 215 | 57% |
| Young With Solutions | 25 | 24 | 49% |
| Neil Bannister | 19 | 4 | 17% |
| OmniFunnel Marketing | 46 | 0 | 0% |
| Lure Agency | 5 | 0 | 0% |
OmniFunnel has 46 ads and has never switched one off. Every single one was created on 8 August. We read their volume as a sign they had found something that works. It was a launch, not a track record.
Their writing is still worth borrowing. But there is no evidence yet that any of it works, and we should stop treating it as proven.
Chase Chappell has made 378 ads and switched off 215 of them. That is what a real testing operation looks like — make a lot, keep the few that work, kill the rest without sentiment.
He is the one to learn from. The others are mostly still guessing, the same as we are.
The two companies with a real history both killed roughly half of everything they made. So when two of our three ads fail, that is not a bad campaign. That is the normal rate, and the plan already budgets for it.
One company launched two ads on 9 August and had killed both by the 10th. A day was enough for them to know.
An ad being switched off is not proof it failed — a campaign can end for a dozen innocent reasons. And a company with no dead ads may simply be new rather than careless. What the kill rate does show reliably is who has been testing long enough to have learned anything, which is a much better guide to who is worth copying than counting how many ads someone is running today.
We have been cold-emailing this exact audience for weeks. Those subject lines are hooks that were tested on real TikTok Shop brands, and the replies say which ones landed. Nobody else has this.
On 7 and 8 August we sent about eighty-five emails with subject lines like "$448K GMV in 90 Days (Here's How)" and "$534K GMV Case Study".
Two automatic out-of-office replies came back. No human replies. No calls booked.
Every one of them opened with a compliment, then "quick context on us", then three of our results, then a bulleted list of what we would handle. The reader's own problem never appears anywhere in the email.
On 5 August one email went out with the subject "Double Gold Zero TikTok Sales". He booked a call. It is in the calendar for 18 August.
You've got Double Gold medals and a killer recipe blog but a Habanero Margarita basically sells itself on TikTok, and you're not on the platform capturing those buyers. Award winning product, no seller engine.The whole opening, verbatim
Four lines. A contradiction he recognises about his own business, then what we do, then the question. No case studies. No numbers about us.
The ads that work open on a broken number the reader can check about themselves. Our one email that worked opened on a contradiction the reader could check about himself. The eighty-five that failed led with proof — and proof is what you reach for when you have not got a diagnosis. Treat "Award winning product, no seller engine" as a template, not a one-off: name something true and good about them, then name the gap.
Eighty-five emails in about two hours from one domain is a textbook spam pattern, so some of that silence may be filtering rather than bad writing. Two auto-replies prove at least some arrived, but not how many. It has also only been three days. And the winning email is a single case, not a test. It points somewhere; it does not prove anything on its own.
The cold email claims "1,000+ vetted creator affiliates". The website says "300+ vetted US creators on file". The email asks for a 15-minute walkthrough; the site sells a 30-minute teardown. Pick one of each before any ad points anywhere.
These are a competitor's own published test results. It is the clearest evidence we found, and it points the opposite way to instinct.
How often each type of ad beat the others in their testing.
No photography, no studio, no animation. Dark background, one bold line, three short lines under it. But make plenty: Meta's own figures say advertisers rotating thirty or more creatives per campaign get up to 43% cheaper enquiries than those refreshing once a month. Three ads is where we start, not where we stop. Video comes second, on whichever message wins, filmed on a phone.
You shipped 50 samples to creators. Seven posted. You paid for all fifty.Never goes out of date
Your TikTok agency reports on TikTok. Your Amazon agency reports on Amazon. Neither can tell you if Tuesday's video moved Friday's Amazon rank.Already on our site, currently buried
A creator says yes on Monday. The sample arrives a fortnight later. They have moved on, and you paid for the box.Leads with the 48-hour guarantee
Award winning product, no seller engine.The line that booked us a call. Name something true and good about them, then name the gap.
Agency relationships now last under two years on average, down from more than four in 2018, and retention is at a decade low. The reasons people give are worth stealing as copy: the agency is spread too thin to learn the category, nobody can show what the money did, and the price outgrew the value. Our six-accounts-per-strategist cap, our dashboards and our $999 entry answer all three — we have simply never said so in those words.
"For US brands doing $25K+ a month on TikTok Shop." Six of ten competitors do this and not one of them can actually enforce it, because they have no information until the sales call. We can, because the application asks for a revenue range before anyone reaches the calendar. That is a real advantage and nobody else in this group has built it.
The goal is three clients. That is a small target, and it makes most normal advertising instincts wrong.
Not five campaigns. Splitting a small budget into pieces means no piece ever gets enough data to learn from. Use the Advantage+ Leads campaign type — Facebook built it for advertisers who cannot produce fifty conversions a week, which describes us exactly. Then put plenty of ads inside it and let Facebook move money to whichever works.
About $2,100, inside the $3,000. Anything faster and we outrun what we can learn. Anything slower and Facebook never settles.
Facebook has no way of knowing who runs a TikTok Shop doing $50,000 a month, so there is nothing precise to aim at. We let the first line of the ad do the filtering instead. We cannot use our contact lists, because our privacy policy promises we don't share data for advertising.
Three clients is too few to tell a good ad from a lucky one. Enquiries arrive in useful numbers, so that is what we measure and what we tell Facebook to optimise for.
Every application tells us the brand's monthly revenue before they reach the calendar. So we can see which ad brings $100K-a-month brands and which brings hobbyists, within about a week. An ad producing cheap enquiries from tiny businesses is worse than one producing expensive enquiries from real ones.
Success looks like switching this off early. Advertising capacity we cannot serve would damage the thing the capacity limit exists to protect.
If it spends $250 and brings nothing. Or if it brings five enquiries and none are the right size of business.
Once a good enquiry costs under $150 and we have at least ten of them. Then increase by no more than 20% every third day — faster and Facebook starts learning again from scratch.
The website side is further ahead than the advertising side. Everything now waits on one thing.
Our website says in three places that we have no Facebook advertising results to show. On a sales call that reads as honesty and it works in our favour. In an advert it gives a stranger a reason to leave before they have met us. The main site should keep it. The advertising landing page should not repeat it.
Eleven keyword searches of Facebook's public Ad Library, read twice — once through a browser to capture the wording, and once through Facebook's own data connection to confirm the ad IDs and start dates. United States, live ads only.
We also opened the destination pages of the top ads to see how their enquiry forms work, and looked at the actual pictures and videos rather than just the text.
Facebook does not publish how much anyone spends or how well their ads perform. So nothing here is a performance figure, and how long an ad has run is the only clue we have. The Ad Library also shows the biggest spenders first, so this is the top of the list rather than all of it. Every claim a company makes about itself is their claim, not a verified fact.
We also could not surface much genuine discussion between working advertisers — searches for it mostly returned marketing articles written to be found. So the practitioner view here is thinner than the rest.
Because platform advice dates quickly, we verified the mechanics rather than trusting memory. Confirmed: Facebook removed 7-day and 28-day view attribution on 12 January 2026, so 7-day click and 1-day view is now the only option. Andromeda, its ad-selection engine, judges creative before audience, which is why creative variety beats clever targeting. A 2–5% location surcharge started on 1 July. And ads containing a realistic AI-generated person are now detected and labelled automatically — which applies to our own AI content.