The Algorithmic Hunt: Mastering Meta and Google

Transcript

When someone is eligible to see an ad on Meta or Google, an auction has likely already happened. It lasted milliseconds. It happens in the background. But it determined whether your ad appeared at all, what it cost, and who saw it. Both Meta and Google rely on real-time, auction-based delivery for performance ads. Your bid enters. So does your estimated action rate — the platform's prediction of whether that specific person will actually convert. Ad quality joins the calculation too. The result is your ad rank. Win the rank, win the placement. Now, the crucial insight: a higher bid does not automatically win. Algorithmic optimization and conversion data can lower your costs even against a competitor spending more. Last time, Hugh, we established that the clearest, fastest ad wins — not the most polished one. That principle connects directly to media buying. Algorithmic optimization and conversion data are key drivers of auction performance on both Meta and Google. The platforms say so explicitly. Broad targeting and conversion data can improve your auction position and reduce costs for the same bid. So the ad you build and the budget you set are not separate decisions. They are the same decision. Here is where most founders make a costly mistake. The instinct is to rely on manual targeting — narrow interests, demographics, and locations layered together. It feels precise. It is actually limiting. Meta reports that broader audiences can improve conversion volume and lower costs. Algorithmic optimization explores larger pools and finds patterns humans cannot manually construct. Meta explicitly recommends fewer, larger audiences and simplified account structures. Let the algorithm optimize delivery based on conversion data rather than your assumptions about who the customer is. Think of it like casting a wide net versus hand-picking fish one by one. The net wins at scale. Now, the algorithm needs fuel to learn. That fuel is conversion data. And this is where tracking becomes non-negotiable, Hugh. Both Meta and Google recommend implementing pixels or tags alongside server-side signal options — Conversions API for Meta, and enhanced conversions or server-side tagging for Google. Why? Privacy changes, including Apple's App Tracking Transparency, have reduced observable user-level data. Relying on browser-based signals alone is less dependable than combining them with server-side signals. Brands using both pixel events and the Conversions API together typically see more stable performance than those relying on browser tracking alone. For a subscription brand, the deeper the event you optimize for — a subscription start, a first recurring charge — the higher the long-term value of the customers the algorithm finds. Cheap traffic is easy to buy. Subscribers are what you need. Meta’s Advantage+ shopping campaigns and Google’s Performance Max campaigns automate placements and audiences based on conversion goals. Performance Max runs across Search, Display, YouTube, Discover, Gmail, and Maps simultaneously — all driven by your conversion goals and audience signals. You give up granular manual control. In return, you get scale the algorithm can find that you cannot. Google notes that advertisers using automated bidding like Target ROAS often see improved performance versus manual bidding once enough conversion data exists. The system reacts in real time to signals humans simply cannot process. For Grub Club, that means feeding these campaigns your best creative assets and your clearest conversion goal — a subscription start — then letting the machine hunt. Broad targeting handles discovery. Retargeting closes the loop. Site visitors and cart abandoners remain among the highest-converting audiences on both platforms. But privacy rules require consent and limit frequency. That makes your consented first-party lists — email subscribers, past customers — critical remarketing assets. For a subscription brand, the campaign goal cannot stop at initial acquisition cost. You need to integrate lifetime value back into your bidding targets. Set your ROAS targets based on what a subscriber is worth over twelve months, not just their first order. That single adjustment changes which customers the algorithm prioritises finding for you. The key idea is this. Effective media buying for Grub Club is not about picking the right interest checkbox. It is about giving two powerful algorithms the right objective, the right signal, and the right creative — then trusting them to find your subscribers. Broad targeting for discovery. Deeper-funnel events for optimization. First-party data for retargeting. And remember: platforms can over-attribute conversions. Running holdout tests or geo-tests helps confirm whether your spend is driving net new subscribers or just claiming credit for people who would have converted anyway. That means your job, Hugh, is not to outsmart the algorithm. It is to feed it well and measure it honestly.