In short
This is a pattern library of ad creative angles for eight industries: SaaS, B2B, ecommerce, DTC, fintech, healthcare, education and real estate. Each page names eight or nine recurring creative patterns, explains the hook structure behind each one, and says why it works against the specific objection buyers in that category actually have. It also covers the mistakes that waste budget in that vertical and what to test first. There are no screenshots of other companies’ ads and no performance numbers attached to them, because we cannot verify either.
Pick your vertical
Why a pattern beats a swipe file
A screenshot of a well-known brand’s ad is an appealing thing to collect and a difficult thing to use. You do not know what it was optimizing for, what it was tested against, which audience saw it, or whether it was any good. Most of the time it was chosen for the collection because the brand is famous, which is the one variable you cannot copy. The pattern underneath is the transferable part: an ad that leads with the objection instead of the benefit works for a reason that has nothing to do with who ran it.
So each page here is written as a teardown in the abstract. Eight or nine named patterns, what each one is structurally, why it holds up against that industry’s buying cycle and its particular objection, and an illustrative hook we wrote ourselves. The illustration is labelled as an illustration. We are not going to put a made-up percentage next to a company name and call it a case study.
How to use these pages
- Find the objection first. Every vertical here opens with what actually stops the purchase, and it is rarely a missing feature. In B2B it is career risk. In ecommerce it is returns and shipping. In fintech it is custody. Choose the pattern that removes yours.
- Rebuild the hook, not the ad. Take the structure, put your own specifics in it, and keep everything after the first line constant so you can read the result.
- Run one variable at a time. Five simultaneous changes produce a winner you cannot explain and therefore cannot repeat.
- Judge it on the event that matters. Cost per click will pick a different winner than cost per qualified lead, and in most of these verticals the two disagree.
Where compliance is the creative brief
Four of these verticals are constrained by rules rather than by taste, and the constraints arrive before the first draft rather than at review. Fintech ads carry disclosure obligations and credit products sit in a restricted targeting category. Healthcare ads may not assert knowledge of the reader’s medical condition, which invalidates the category’s most common opening line. Real estate ads fall under Fair Housing rules and lose most of the targeting toolkit. Education ads cannot promise the outcome they are selling. In all four, the constraint is genuinely useful: it forces the qualifying into the copy, which is where it produces better ads anyway.
The part after the creative
Writing the ad is the half everyone enjoys. The other half is deciding, every week, which creative to keep, which to pause, and where the budget goes next, across accounts that each report their own contribution generously. Muze connects Meta Ads, Google Ads, Amazon Ads and Shopify through one OAuth connection into ChatGPT, Claude or any MCP client, so those questions get answered in one place rather than in four tabs that disagree.
It acts as well as reads: pausing the loser, moving the budget, building the campaign. Every write previews first and waits for explicit confirmation, new campaigns are created paused, and Muze never takes a percentage of ad spend. If you want the vocabulary nailed down first, the paid media glossary defines the metrics these pages assume, and the MCP page explains how the connection works.
Frequently asked questions
- What is an ad pattern library?
- It is a catalogue of creative angles rather than a gallery of finished ads. Each entry names the pattern, explains the structure of the hook, and says why that structure works against the specific objection a buyer in that industry has. The point is that a pattern transfers between products, where a screenshot of somebody else's ad does not.
- Why are there no real ad screenshots on these pages?
- Because we do not have verified performance data for other companies' campaigns, and publishing an ad next to a result we cannot source would be an invention. Plenty of sites do it anyway. We would rather publish the structural layer, which is the part that actually transfers to your account.
- How do you pick which ad pattern to use?
- Start from the objection, not the format. Work out the single reason a qualified buyer in your category says no, then choose the pattern that removes it. Format follows from that. Most weak creative is a good format applied to the wrong objection.
- How many ad variations should you test at once?
- Test distinct angles rather than variations of one line. Three to five genuinely different angles will teach you more than twenty rewrites of a single headline, because rewrites usually land inside the statistical noise. Once an angle wins, iterate inside it.
- Do these patterns work on Meta, Google and Amazon equally?
- The angles transfer. The formats do not. Search advertising meets people who already named the problem, so the hook has less work to do and the copy is tighter. Social advertising has to establish the problem before it introduces the product. Retail media sits at the point of purchase inside somebody else's storefront. Same pattern, three different builds.
- What is the fastest way to improve ad creative?
- Change the first line and the first frame, and hold everything else constant. It is the cheapest variable to change, it is the only part most of your audience sees, and it produces the largest difference between two ads that are otherwise identical.
Patterns are the easy half
Ask which creative is actually carrying the account, from ChatGPT or Claude, across Meta, Google, Amazon and Shopify. Free tier is 25 read-only tool calls a month, and nothing changes until you confirm it.
See how Muze works