Blog Playbooks · Who Keeps Score · Part 9
Three companies take most of the ad money. Here is how to test your dependence.
eMarketer forecasts that Google, Meta and Amazon will take 62.3% of worldwide digital ad spending in 2026. Part 9 of Who Keeps Score: what the filings show, what one iPhone setting did to Meta, and how a local business can measure its own reliance.
Say you own a seven-location fitness brand. Most months, almost every new lead you pay for passes through one of three companies. Someone searches on Google, scrolls Instagram, or watches a video on a Google or Meta property. Each time, you pay for the contact, and the same company tells you what it was worth.
This part is about that concentration. We buy on these platforms every week, and they work. Still, the businesses we work with should know how much of their growth depends on three companies, and should test that dependence on purpose instead of finding out by accident.
Where the money goes
The two biggest players publish their numbers to the SEC. Alphabet's annual report for 2025 shows Google advertising revenue of $294.69 billion, out of total revenues of $402.84 billion. Meta's annual report shows advertising revenue of $196.18 billion, out of total revenue of $200.97 billion. Meta reports that ad impressions rose 12% for the year and the average price per ad rose 9%.
Source: Alphabet 10-K for 2025 and Meta 10-K for 2025
For Meta, ads are nearly the whole business: about 98% of revenue, by our arithmetic on its own figures. For Alphabet, ads are about 73%.
Forecasts look ahead. eMarketer projected in April 2026 that Meta will pass Google in worldwide net ad revenue this year for the first time, with $243.46 billion against Google's $239.54 billion. Add Amazon, and the three together are forecast to take 62.3% of worldwide digital ad spending in 2026.
Source: eMarketer forecast, April 2026, as reported by Marketing Dive and MediaPost
Those are forecasts, and the figures are eMarketer's. They still describe something a gym owner can feel: when the auction gets more expensive, there are few other places with comparable reach.
When the owner of the phone changed a setting
The clearest lesson about dependence comes from Meta itself. In April 2021 Apple shipped App Tracking Transparency with iOS 14.5. On Meta's earnings call on February 2, 2022, CFO Dave Wehner put the 2022 hit at "on the order of $10 billion," as reported by Fortune. It was a forecast made early in the year.
Meta still describes the effect in its latest annual report, in its own words.
"In 2021, Apple made certain changes to its products and data use policies in connection with changes to its iOS operating system that reduce our and other iOS developers' ability to target and measure advertising."
The same filing says the change "has negatively impacted, and we expect will continue to negatively impact, the size of the budgets marketers are willing to commit to us and other advertising platforms."
One of the largest ad businesses in the world depends on a phone it does not make, and a single setting change by the phone's maker cost it an estimated $10 billion in one year. A local studio depends on Meta in much the same way Meta depends on Apple, and the owner of the channel sets the rules for everyone downstream.
Even Amazon stepped back
Amazon is the third company on that list, and it is also a huge advertiser on Google. In July 2025 it stopped bidding in Google Shopping ads. MediaPost, citing client data from the agency Tinuiti, reported that between July 21 and 23, 2025, Amazon's average Shopping impression share among Tinuiti's clients "fell to zero" from 60% in the US, 55% in the UK and 38% in Germany. The article carried no statement from Amazon, and we know of none explaining why.
We will not guess at the reason, and Amazon has not given one. What the data shows is simpler: a very large advertiser stopped buying a big paid channel, on its own schedule, which most small advertisers never try.
A contact you rent
A lead from a Google or Meta ad. You pay per click or per thousand views. The price is set in an auction, the rules can change, and you reach the same person again by paying again.
A contact you own
A member on your email list, your SMS list, your app or your referral program. You reach them as often as they will let you, at close to no cost per message, and no auction stands between you.
Test your dependence with a defined pause
Owned channels are the one place where you do not pay for every contact. That makes them worth building, and it makes one experiment worth running before you need it: a short, defined pause of one paid channel, measured with your own numbers.
- Pick one channel and one market
For example, Google Search ads for two of your seven locations. Keep everything else running, so the test isolates one variable.
- Fix the window before you start
One or two full weeks, chosen ahead of time, away from holidays and launches. Do not end early because the first days look good or bad.
- Record a baseline
Leads, booked intros and joins for those locations over the same weeks last month, plus direct, referral and organic visits.
- Pause and count in your own system
Count leads and joins from your booking system, not the platform dashboard. Note what direct, referral and member-referral traffic does.
- Compare against the locations that kept running
If the paused locations lose as many joins as the platform claimed it was driving, the channel was doing the work. If they lose fewer, some of those conversions would have come anyway.
- Decide with your numbers, then turn it back on
Restart the channel and set next quarter's budget with what you learned.
The businesses that come out of this test well are usually the ones with a member list they actually use. If a pause shows that most of your joins still arrive through referrals and your own list, you have more pricing power with every platform than you thought. If it shows the opposite, you know where to build next. We covered the data side of that in Stop paying Meta to find your own customers, and one way to put a ceiling on a Google channel in You can finally cap Google's display spend.
In Part 8 we looked at who counts the results. Next, in the final part, we show how we keep our own score: How we keep our own score.
Sources
- Alphabet Inc., Form 10-K for the year ended December 31, 2025, filed February 5, 2026. sec.gov
- Alphabet Inc., "Alphabet Announces Fourth Quarter and Fiscal Year 2025 Results," February 4, 2026. abc.xyz
- Meta Platforms, Form 10-K for the year ended December 31, 2025. sec.gov
- Marketing Dive, Sara Karlovitch, "Meta to surpass Google in digital ad revenue for first time: Emarketer," April 14, 2026. marketingdive.com
- MediaPost, Laurie Sullivan, "Meta Set To Surpass Google In 2026 Ad Revenue," April 14, 2026. mediapost.com
- eMarketer, "Meta to surpass Google in digital ad revenues for first time ever," April 13, 2026 (press release; it blocked our fetch, so figures are as reported above). emarketer.com
- Fortune, Jacob Carpenter, on Meta's fourth-quarter 2021 results and Apple's privacy changes, February 3, 2022. fortune.com
- MediaPost, Laurie Sullivan, "Amazon Turns Off Google Shopping Ads," July 27, 2025. mediapost.com
