The Initial Results Are In for Google's Fledge Tests

The first results from Google's solution to prevent third-party browser tracking have arrived. Currently, three companies on the buy side—Google, Criteo and RTB House—are testing Fledge, an acronym for "first locally executed decision over groups experiment." RTB House, in collaboration with 17 publishers including Café Media, ran Fledge testing from early April to Sept. 15. They tested basic functions, including adding and removing users from interest groups, while running ad auctions, and serving users with ad impression while registering clicks and conversions. One surprising takeaway was the lack of support from ad servers and SSPs that did not allow testing of a complete programmatic integration, said Lukasz Wlodarczyk, vp of programmatic ecosystem growth and innovation. "According to one of the interviews with our partners, this is caused by a lack of clarity on how the integrations between SSPs and publishers’ ad servers will work in the Fledge environment," he said. This is part of Google's attempt to shepherd the industry toward a cookieless solution for ad delivery. Fledge aims to serve behaviorally targeted ads via actions taking place on a device instead of an external server, which is the current industry standard. Advertisers were allowed to use their choice of server while seeing an increase in impression numbers toward the end. While Google has delayed its plan to axe third-party cookies to 2024, the tech giant needs to incentivize market participants to test Fledge, according to Wlodarczyk, otherwise it risks all Privacy Sandbox initiatives being left behind. Despite the SSP workarounds, publishers like Café Media found the Fledge proposal to work "pretty easily," according to Paul Bannister, chief strategy officer at Café Media. Through this trial, RTB House served close to 7 million ads across 2,700 advertisers in 50 countries. Chrome changes the scale Previously, Google's Privacy Sandbox testing was made available only to a small percentage of Chrome users, limiting marketers to basic-level testing, according to Bannister. However, Fledge was widely implemented in the main version of Chrome. "That scale differential made a very big difference," he said. RTB House added and removed users from interest groups while running ad auctions. A total of 1.2 million users were included in interest groups across 225 countries. The Fledge trial allowed participants to use the log-level reporting, where RTB House was able to use the current mechanics of measuring impressions and clicks. Although most integrations with publishers were completed between June and July, it wasn’t until August that the number of Fledge impressions rapidly increased, with more than 6.5 million ad impressions served in August compared to almost 213,000 in June. More than 7 million impressions were served throughout the trial, meeting the expectations of RTB House and all publishers. "The fact it [reached that number] is positive," said Bannister, “because prior to Chrome changing the scale of the testing, I don't think it would have gotten [there] in a fairly short period." RTB House mostly experimented with ad placements bought directly from publisher partners. The experiment simulated a single-level Fledge auction with a single buyer, and ran in pre-bought placements restricted to trial-enabled Chrome users. More scope on behavioral targeting RTB House tested ads composed of multiple pieces, including its PLTD, which stands for product-level Turtledove (Two Uncorrelated Requests, Then Locally Executed Decision On Victory). This engine built recommendations that suggested products previously not seen by people but could be liked by them, according to Wlodarczyk. This significantly increased the scope of tailored ads and capabilities for creating interest groups. "This not only helps create interest groups based on user behavior, but also based on particular items the user has seen on the internet," he said. Ideally, Fledge enables decision-making to take place in a single browser, but for the initial trial, Google allowed participants to "bring your own server." This enabled RTB House to use its server to fetch additional signals that are usable in real-time bidding. "This allowed us to put some of the computation on the server instead of putting it 100% on the device," Wlodarczyk added. No Cookies? You Don’t Need Them When You Have Motivation AI

Ad Tech Industry News

The Initial Results Are In for Google's Fledge Tests

Baking a post-cookie privacy solution

Trishla Ostwal

About

Subscriptions

Events

Publications

WORK SMARTER - LEARN, GROW AND BE INSPIRED.

Subscribe today!

Trishla Ostwal