The shape of the account, before the copy

Most people picturing a large ecommerce Google account picture search ads. The library says otherwise.

Live US ads by format, gymshark.com
FormatAdsShareWhat sits behind it
Video~4,000~80%Demand Gen, YouTube, and video assets inside Performance Max
Image~1,000~20%Shopping and display, largely feed-driven
Text~400~8%The search account most people imagine is the whole thing
Total~5,000100%Advertiser verified as Gymshark USA Inc

Source: Google Ads Transparency Center, US region, format filter applied separately for each row. Counts are Google's own rounding, so the rows do not sum exactly. Checked 20 September 2026.

Four fifths of a mature DTC brand's Google presence is video. That is worth sitting with if you are budgeting for one: the thing you will spend most of your production money on is not the search ad, and the skill the account needs most is not keyword work. It is a creative pipeline.

It also explains a common disappointment. A brand hires a search specialist, gets a well-built search account, and sees very little change in revenue – because the search account was never where the volume was.

Every text ad is defending the brand

All four distinct text creatives we found lead with the brand name, and two of them say “Official” in the headline.

Observed text ads, US
HeadlineWhat it is doing
Gymshark – The Gym Is Everything We Do – We Do GymBrand term, identity angle
Gymshark Official StoreBrand defence, with five sitelinks doing the navigating
Gymshark Official Site – Shop Workout Apparel & GearBrand defence plus shipping and discount terms
Gymshark Women's Gym Clothes – Gym Clothes for WomenSegment headline, same description as the brand ad

Source: Google Ads Transparency Center, advertiser Gymshark USA Inc, verified. Checked 20 September 2026. Last shown dates on these creatives were the same day.

The word “Official” is the tell. It is what a brand writes when somebody else is bidding on its name – resellers, marketplaces, affiliates, competitors. You are not reading an acquisition strategy here. You are reading a toll being paid to keep a click that was already yours.

Whether that toll is worth paying is the oldest argument in paid search, and it is not settled by looking at the ad. It is settled by a brand-term holdout, which almost nobody runs because the downside of being wrong is visible immediately and the upside is invisible.

The segmentation stops at the headline

The women's ad has its own headline and then reuses, word for word, the description from the general brand ad: the same opening line about clothes, accessories and community.

Compare that with a subscription app we took apart earlier, which segmented the destination rather than the headline – marathon searches to a marathon page, half-marathon searches to a half-marathon page. That is segmentation doing work. A segmented headline over a generic description is segmentation that stops at the part the platform makes easy.

We would not call this a mistake without seeing the numbers. Responsive search ads assemble themselves, and a description that wins across segments may genuinely be the best available. But it is the first thing we would test, because it costs nothing to write a second description and the platform will tell you within a fortnight.

The annotations are doing more than the copy

Attached to those four ads: a seller rating with several hundred reviews, a return policy line reading “Most items 30+ days”, an email reply time of one hour, an app-promotion link reading “See results in-app”, free shipping over $75 stated in the description, a student discount, and five sitelinks – Sale & Exclusives, 20% Student Discount, Women's Best Sellers, Gym Clothes, Men's Best Sellers.

None of that is written by a copywriter. Ratings come from a reviews integration, return and delivery lines come from the Merchant Center feed, reply time comes from support data, sitelinks come from the account structure. On a crowded commercial result this material occupies more vertical space than the ad text and answers more of the objections.

02

Meta is running a different argument entirely

Same brand, same week, and almost nothing in common with the search account.

On Meta the price is right there in the headline – “From $24” – and the copy gives a reason to act now. One ad opens by announcing that a collection has restocked. Another sells the best sellers as social proof: the pieces everyone is talking about. A third names a curated edit, the Short List, and positions it as the answer to choosing.

On Google, across four text ads, there is no price at all.

This is not an inconsistency. It is the correct response to what each click means. Someone typing the brand name into Google has already decided; a price in that ad can only cost you margin or talk someone out of it. Someone scrolling Instagram has decided nothing, and the price is the first qualifying question. Any brand running one message across both is making one of those two mistakes.

What differs by channel
Google textMeta
PriceAbsent“From $24” in the headline
Reason to act nowNoneRestock, best sellers, curated edit
Job of the adKeep a click you had already earnedCreate a click that did not exist
ProofSeller rating, return policy, reply timePopularity and product photography
ToneFlat and functionalConversational, emoji, second person

Source: Meta Ad Library and Google Ads Transparency Center, US, active ads. Checked 20 September 2026.

Two pages, not one

Gymshark runs ads from a second page, Gymshark Women, alongside the main one. The women's page ad we read was a restock announcement written in the first person to a reader who is assumed to already know the product line.

A separate page changes the sender, not just the targeting. The ad arrives from something that looks like a community the reader belongs to rather than from a corporate account that also sells to everyone else. It is the same mechanism behind creator-account advertising, applied to a segment instead of a person.

The cost is not trivial and is usually underestimated: two content pipelines, two audiences to build, two sets of social proof to maintain, and a second page that looks abandoned if the programme stops.

Cadence, visible in the dates

Start dates on the active ads cluster in mid-July and across August, with several carrying Meta's “multiple versions” marker and one carrying “10 ads use this creative and text”. Placements run across Facebook, Instagram, Messenger, Threads and Audience Network rather than a chosen subset.

That pattern – batches launched on a schedule, one concept fanned out across many units, all placements on – is what an actively managed social account looks like from outside. The opposite pattern, which we found on a competing app in an earlier teardown, is a single block of copy unchanged for four months.

03

What we would ask if this were our account

None of it is answerable from outside, and all of it is answerable in a fortnight from inside.

  1. What does brand search cost, and what does it return incrementally? Three of four text ads defend the brand term. A geo holdout on brand-only campaigns is uncomfortable and it is the only way to know whether that budget is buying customers or buying back visits.
  2. Does the women's ad get its own description, and does it matter? A second description is twenty minutes of work and the platform returns an answer within a fortnight.
  3. Is four fifths of the budget in video because it performs, or because Performance Max put it there? Those are different situations with the same appearance in the library, and only the asset-group reporting tells them apart.
  4. Does “From $24” set the right anchor? A price-led headline recruits the price-sensitive. For a brand whose margin story depends on full price, that is a trade worth measuring in contribution rather than in conversion rate.

Questions people actually ask

Where do these numbers come from?
Google's Ads Transparency Center and Meta's Ad Library, US region, checked on 20 September 2026. Both are public and both are required by law, so anyone can pull the same figures. The counts are the platforms' own rounding.
Does 5,000 ads mean 5,000 different creatives?
No. Ad libraries count ad units, and a single asset deployed across many units counts many times. That is why the ratio between formats is more informative than the absolute numbers.
Why is there no pricing in the search ads?
Because a search ad against your own brand name does not need to sell. The person typing the brand has already decided; the ad exists so a competitor does not take the click. Price belongs where the ad interrupts rather than answers, which is why it appears on Meta instead.
Is running two Meta pages normal?
It is common in apparel and increasingly elsewhere. A second page lets one audience see a feed that is entirely about them, and it makes the sender identity match the segment. The cost is real: two content pipelines, two audiences to warm, two sets of social proof.
Do you know whether any of it works?
No. Spend, revenue and targeting are private, and nothing in a public library tells you whether an ad was profitable. This reads decisions, not outcomes – and where we infer a reason, we say we are inferring.

This, on your account, with the numbers attached

The free audit is this analysis run on your own advertising – except we can see spend, conversion data and the feed underneath, instead of inferring from what the libraries publish.