Teardown · Cross-section
Seven advertisers, one day, one table
We have taken these accounts apart one at a time. Put side by side and re-counted on the same day, they answer a question people ask constantly and almost never get a number for: how much advertising does a company like mine actually run?
- Checked
- 20 September 2026
- Advertisers
- 7
- Range
- 40× between top and bottom
- Source
- Google ATC, US
| Advertiser | Category | Total | Text | Video |
|---|---|---|---|---|
| monday.com | B2B software | ~8,000 | ~5,000 | ~3,000 |
| Gymshark | Ecommerce & DTC | ~5,000 | ~400 | ~4,000 |
| Wise | Fintech | ~2,000 | ~2,000 | ~600 |
| Headspace | Subscription app | ~400 | ~400 | 8 |
| Notion | B2B software | ~200 | ~200 | 57 |
| Runna | Subscription app | ~200 | ~200 | 10 |
| Duolingo | Subscription app | ~200 | 66 | 10 |
Source: Google Ads Transparency Center, US region, filtered by domain then by format. All seven re-pulled on 20 September 2026 so the figures are comparable with each other. Google rounds above roughly a hundred, so columns do not sum.
The individual teardowns are linked at the bottom. What follows is only what the table says.
The range is the first finding
Between the largest and smallest here there is a factor of forty. All seven are well known, well funded and competently run. None of them is doing it wrong.
Which means the question people usually ask – "how much should a company like us be running?" – has no answer at the level of the category. It has an answer at the level of your own demand, your own margins and your own capacity to produce creative, and that answer is arithmetic rather than a benchmark. We work through the arithmetic on when not to hire a PPC agency.
The second finding kills the first tidy conclusion
Read the top two rows and an obvious rule appears: B2B buys search text, DTC buys video. Read the whole table and it falls apart.
Notion and monday.com are both B2B software, and they differ by a factor of forty. Duolingo and Headspace are both subscription apps, and Headspace runs six times the search text. Category is not doing the work here. Strategy is.
What actually separates them is whether the buyer arrives with a question already formed. monday.com sells to someone typing "project tracker", so it buys search at volume and splits one platform into four products by the job each searcher described. Gymshark sells to someone who was not looking for clothes at all, so four fifths of its Google presence is video. Duolingo is the sharpest case: it barely advertises the famous free app and spends half its search budget on a $70 exam, because an exam has a deadline and a language app does not.
02
What these numbers are not
Stated plainly, because a table this quotable will be quoted badly otherwise.
They are ad units, not ideas. Responsive formats assemble many units from a few assets. Five thousand ads can be a few dozen headlines deployed across a wide account, and often is.
They are not spend. Nothing in any public library reports budget. A thousand cheap ads and a hundred expensive ones look the same here and are not the same business decision.
They are not performance. No library publishes conversion data. Any claim on this page about why a decision was made is an inference, and is labelled as one.
They are not complete. US only, Google only, one domain each. Apple Search Ads, app networks, affiliate, programmatic and every other market are outside it.
They are rounded hard. Google gives "~200" and "~2K", so a row whose parts exceed its total is a rounding artefact rather than an error. We have left the numbers as the source reports them rather than tidying them into something that looks more precise than it is.
03
The seven, individually
- monday.com – B2B software
- Gymshark – Ecommerce & DTC
- Wise – Fintech
- Headspace – Subscription app
- Notion – B2B software
- Runna – Subscription app
- Duolingo – Subscription app
Each one goes through the copy, the destinations where they are visible, the offer structure and what we would ask if it were our account. None of them is a client and none of them knows we did this.
Questions people actually ask
- Why do the columns not add up?
- Because Google rounds. Above roughly a hundred it reports "~200", "~2K", "~8K" and nothing finer, so a total and its parts can each be rounded in different directions. Wise shows ~2,000 total, ~2,000 text and ~600 video, which cannot all be exact. Treat every figure as an order of magnitude, not a measurement.
- Does a bigger number mean a bigger budget?
- Loosely, and not reliably. Ad libraries count ad units, and responsive formats multiply: a handful of headlines across many ad groups produces a large count without a large idea behind it. A high number means the account is broad; it says nothing about spend per ad or about results.
- Is this all of their advertising?
- No. It is US ads pointing at one domain, in Google's library. App install campaigns on other networks, Apple Search Ads, affiliate and programmatic buys, and everything outside the United States are not here, and most of it is not public anywhere.
- So what should I take from it?
- Two things. That the range between well-run accounts in the same category is enormous, so "what do companies like us spend" is a weaker question than it sounds. And that the format split follows the buyer's intent rather than the industry label.
- Can I reproduce this?
- Yes, in about ten minutes. Open Google's Ads Transparency Center, search a domain, read the count, then apply the format filter and read it again. Everything on this page came from that and nothing else.
Your account, with the numbers we cannot see here
The free audit is this analysis run on your own advertising – except spend, conversion data and the destinations are all visible, so the conclusions are measured rather than inferred.