BRENNEN

Industries / Technology / Software and SaaS

Ad and marketing tech

Ad and marketing tech is the software that picks which ad appears on a web page, app or screen, runs the auction that prices it, checks that a real person could have seen it, measures what it did, and sends automated marketing messages to customers.

1995

Dave Zinman and Jason Strober, business students at Stanford, put an ad server into production in June 1995: software that stores banner ads on one computer, picks which one a web page shows and counts each showing. Their firm, Focalink Media Services, launched the service that July, with Saturn, Prudential and Microsoft as its first customers. Zinman calls it the first ad server placed into production. A 2018 VideoWeek history credits Tom Shields and John Danner of NetGravity with the first publisher-side ad server, which went live on Time Warner's website around Thanksgiving 1995. Both accounts come from the founders.

go2market.blogspot.comvideoweek.com

36

Milestones achieved across 6 domains.

12

To watch for: announced, not yet confirmed outside.

12

Open deliverables. Nobody has shipped them.

In one minute

Where it started

In 1995 websites placed banner ads by hand as image files, and clicks were the only measure. Small firms such as Focalink and NetGravity wrote ad servers to rotate and count ads, and from 1997 DoubleClick rented ad serving to websites from its own data centres. Cookies, small files a browser keeps for a website, let these firms recognise the same browser across many sites, and targeting was built on them, although the first cookie standard, in 1997, said browsers should block such cross-site cookies by default.

Where it stands

Display ads on the open web are sold in automated auctions that run while a page loads, and regulators in France in 2021 and in the EU and the US in 2025 found that Google, which ran the main publisher ad server and an exchange that bid in its auctions, favoured its own exchange. Safari and Firefox block much cross-site tracking by default, while Chrome kept third-party cookies and in October 2025 retired most of the replacements it had built. Fraud has moved from rented servers into millions of infected TV boxes in homes, and AI agents that browse like people are a new kind of traffic that measurement firms have only begun to label.

What is open

Many checks that the trade leaves to private vendors, or to the firms being checked, have no public and repeatable version: how long header-bidding auctions take, whether an ad server's final auction picks the right winner, whether ad filters catch AI agents, whether vendors' cookies last only as long as they declare, and whether labels on AI-made ads survive the trip to the screen. Each can be built with open-source ad servers, automated browsers, made-up data and test accounts, without following any real person.

Domains

AchievedWatch forOpen

Open in ad and marketing tech

3 of 12 open deliverables

Open · Ad auctions

A public header bidding record

A monthly public dataset, made by an automated browser that visits the 10,000 most-visited websites, that records for each header-bidding auction which exchanges were asked, the time limit the site set, how long each bid took, and how long the page waited before asking its ad server for an ad.

Built with what exists

Playwright (Apache 2.0 licence) driving Chromium; Prebid.js's own event log, read in the page with pbjs.getEvents() (auction start and end, bid responses, timeouts); the Tranco research list of top sites; Python for analysis. The browser carries no login or cookies, names itself as a research bot, and stops each page before the call to the ad server, so no ad is shown and no advertiser pays. Both precautions matter: in March 2025 AdExchanger saw log files of ads served to the HTTP Archive's crawler even though it declares itself as a bot.

Credit · unclaimed

Open · Targeting and measurement

Count agent visits in reports

A test shop with no real customers that runs standard analytics and ad tags (Google Analytics 4, the Meta Pixel in its test mode, and the open-source Plausible) next to a server log of every page view and order. AI browsing agents and AI browsers are given shopping tasks, and the test compares what each tag recorded (visits, source, browser, purchases) with the server's record.

Built with what exists

Browser Use driving Chrome (see the Note "Deleting 80% to start over"); consumer AI browsers with agent modes; a WooCommerce shop or Shopify development store with payments in test mode; Google Analytics 4, Meta's Test Events tool and Plausible; Python for the comparison. No real people and no real payments are involved.

Credit · unclaimed

Open · Privacy

Check declared cookie lifetimes

A crawler that loads 1,000 European websites using the TCF in a fresh automated browser, accepts all purposes, records every cookie each listed ad vendor sets and when it expires, and compares that with the maximum lifetime the vendor declares in IAB Europe's Global Vendor List (GVL) and in its own published storage disclosure file.

Built with what exists

the public GVL file and vendors' disclosure files; OpenWPM or Playwright for automated browsing from an EU network address; Python. No real user and no logged-in account are involved.

Credit · unclaimed

Where it sits