Industries / Technology / Social media
Advertising
This is the work of selling space in the feed to businesses and choosing which person sees which ad, which pays for most social media.
5
Milestones achieved, from 2012 to 2025.
3
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
2012
Ads move to the phone
Facebook's filing for its stock market listing, dated 1 February 2012, warned: "We do not currently directly generate any meaningful revenue from the use of Facebook mobile products." In its results for the fourth quarter of 2012, mobile was about 23% of its advertising revenue, up from about 14% in the third quarter, and daily users on phones outnumbered those on the web for the first time.
LessonWhen use moved to a small screen with room for little but the feed, the ads moved into the feed with it.
Credit
Facebook
2013
Ads for people like customers
On 19 March 2013 Facebook announced Lookalike Audiences. An advertiser first uploads its own customer list, which Facebook matches to accounts (a "custom audience", launched the previous autumn); Facebook then finds the 1% of people in a chosen country most similar to those customers, or the top 5% for wider reach.
Credit
Facebook
2013
Friends' faces in ads
Facebook's Sponsored Stories turned things people did, such as liking a brand or checking in at a store, into paid ads shown to their friends with their names and photos. A class action for about 150 million members, filed in April 2011, said users were not properly told and could not opt out. On 26 August 2013 the US District Court for the Northern District of California approved a US$20 million settlement, US$15 to each member who claimed, and changes to Facebook's terms.
LessonUsing people as endorsers in ads needs their informed permission, even when what they did was already visible to friends.
Credit
Facebook; US District Court for the Northern District of California
2023
Courts and regulators curb profiling
On 4 July 2023 the Court of Justice of the European Union ruled in Meta Platforms v Bundeskartellamt (case C-252/21) that the personalized advertising that finances Facebook cannot justify, as a "legitimate interest", Meta's processing of users' data from on and off Facebook without their consent, and that a competition authority may find breaches of the EU's data protection law. On 17 July 2023 Norway's data protection authority banned Meta's behavioral advertising to users in Norway, ads based on tracking and profiling them, for three months from 4 August, on pain of fines of up to NOK 1 million a day.
Credit
Court of Justice of the European Union; Datatilsynet, Norway
2025
Pay or consent fined
From November 2023 Meta asked people in the EU either to consent to personalized ads or to pay €9.99 a month on the web, or €12.99 on iOS and Android, for Facebook and Instagram without ads. On 23 April 2025 the European Commission fined Meta €200 million under the Digital Markets Act, the EU law for the largest online "gatekeepers", finding that the two-way choice gave no equivalent option using less personal data. The fine covered March to November 2024; in November 2024 Meta had cut the price to €5.99 on the web and added a free option with "less personalized ads".
LessonA paid way out of tracking did not count as a free choice; the regulator wanted a free option that uses less data.
Credit
European Commission; Meta
Watch for
Bigger ad models, more sales
Meta says GEM, which it calls the largest foundation model for recommendation in the industry, raised ad conversions, actions such as purchases that advertisers pay to get, by 5% on Instagram and 3% on Facebook Feed in the second quarter of 2025, and is four times as efficient at improving ad performance for a given amount of data and computing as its original ranking models.
Achieved whenadvertisers' own holdout tests or an independent measurement firm report a comparable lift.
Claimed by
Meta, in its engineering blog, November 2025
Fewer ads, little extra time
Economists George Beknazar-Yuzbashev, Rafael Jiménez-Durán, Andrey Simonov and Mateusz Stalinski ran a field experiment with 1,810 US Facebook users who installed a browser extension. Hiding 82% of ads raised time on the platform by only 6%; replacing targeted ads with untargeted ones sharply cut ad clicks and long-run use; and two-thirds of the variation in ad load, the share of the feed given to ads, was between users rather than over time.
Achieved whenthe paper is published after peer review or the result is repeated on another platform.
Claimed by
the authors, in a University of Warwick CAGE working paper, February 2026
Less personalized ads comply
On 8 December 2025 the European Commission acknowledged Meta's undertaking to let EU users choose between sharing all their data for fully personalized ads and sharing less for more limited personalization, and said it would then seek evidence on the model's "impact and uptake". Meta rolled the choice out in January 2026. On 17 March 2026 BEUC, the European consumer organization, said the model still breaches the Digital Markets Act and the GDPR and that its wording steers users toward full personalization.
Achieved whenthe Commission closes its review finding compliance, or a court rules on the model.
Claimed by
Meta, December 2025
Open
01
Count ads in the feed
A public count, repeated every quarter, of how many items in the main feed of each large platform are ads: Facebook, Instagram, TikTok, X, Snapchat, Pinterest and Reddit, in at least three countries. It is made only on fresh accounts that the testers set up for the test, and it records nothing but each item's position, whether it is an ad and the ad's format; no post, name or handle is kept.
The briefCredit · unclaimed02
Record the ad-choice screens
A monthly, filmed record of the screens that new accounts see about ads on each large platform in two EU countries and the UK: which options are offered (personalized ads, less personalized ads, a paid version without ads), the prices, how many taps each option takes, which one is highlighted, the wording, and whether the choice still holds 30 days later. Each change is dated and published.
The briefCredit · unclaimed
Open 01 · the brief
Count ads in the feed
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public count, repeated every quarter, of how many items in the main feed of each large platform are ads: Facebook, Instagram, TikTok, X, Snapchat, Pinterest and Reddit, in at least three countries. It is made only on fresh accounts that the testers set up for the test, and it records nothing but each item's position, whether it is an ad and the ad's format; no post, name or handle is kept.
Why it is still open
The only recent cross-platform count found is WhistleOut's, made by its staff scrolling 100 posts per platform on their own accounts (updated November 2025), which put Pinterest at one ad in every two posts and X at one in six. The 2026 Warwick field experiment measured Facebook alone and found that ad load varies far more from one user to another than over time, so one person's feed says little.
Built with what exists
Phones or Android emulators with screen recording; a vision-language model such as Claude Fable 5.2 (see the Note "Three working devices") to mark each item in the recording as ad or not ad, after which the recording is deleted; a written protocol for setting up accounts and scrolling.
First step
Set up three fresh accounts on one platform, scroll 300 items on each, tag them by hand and by model, and compare the two.
Done looks like
Seven platforms, three countries, ten fresh accounts per platform and country, 300 feed items per account, repeated each quarter for a year, with published ad shares and 95% confidence intervals and a check of the model's tags against hand tags on a sample.
Open 02 · the brief
Record the ad-choice screens
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A monthly, filmed record of the screens that new accounts see about ads on each large platform in two EU countries and the UK: which options are offered (personalized ads, less personalized ads, a paid version without ads), the prices, how many taps each option takes, which one is highlighted, the wording, and whether the choice still holds 30 days later. Each change is dated and published.
Why it is still open
The European Commission has said it will gather evidence on how Meta's January 2026 choice works in practice, and BEUC published a one-off legal analysis of Meta's version in March 2026. No repeated record across platforms turned up, so changes to these screens go unrecorded between complaints.
Built with what exists
Fresh test accounts on phones, screen recording, a vision-language model to transcribe each screen, and Open Terms Archive, an open-source tool that tracks changes to the written terms of online services.
First step
Film the full ad-choice flow that a new Facebook account sees in one EU country, and count the taps to each option.
Done looks like
Eight platforms in three countries, filmed every month for six months, with tap counts, prices and wording in a public table and every change flagged within a week of first being seen.