Ad platforms stopped being auction houses a while ago. They are prediction machines. You give them examples of what a customer looks like, they go find more. The quality of your examples is now the biggest lever you control, bigger than creative testing cadence, bigger than bid strategy choice.
That claim used to be arguable. Google Marketing Live 2026 settled it.
Qualified Future Conversions, announced there, has Gemini estimate the future value of today’s interactions: the repeat visit, the brand search three weeks later, the second purchase. Smart Bidding then optimizes toward predicted long-term revenue rather than the last click. Read that again from the model’s point of view. It learns from the conversion and value data you feed it. Feed it complete, well-matched, consented signal and it predicts well for you. Feed it a pixel losing a third of its events and it predicts well for someone else.
Meta’s version of the same story is Event Match Quality: a visible score on how well your events match to people, directly correlated with cost per result. Snap and TikTok publish equivalent guidance. Google’s framework name for the discipline is Data Strength, four steps from connecting data sources to proving ROI, and it now shapes how Google’s own teams engage advertisers.
The uncomfortable part: signal quality compounds. A competitor with better match rates gets better optimization this week, which produces more conversions to learn from next week. It is the quiet reason two advertisers with identical budgets and comparable creative see diverging returns.
The encouraging part: the fix list is short and known. Server-side delivery so events stop leaking. Hashed identifiers on every event. Consent Mode implemented properly so Google can actually use what you send. Offline and CRM outcomes flowing back, because your best conversions probably do not happen in a browser. None of this is exotic. It is plumbing, and plumbing is checkable.
We run that check as a working session. Baseline your signal, rank the gaps by revenue impact, fix the top one first. Data Strength assessment is the Google-flavored version; the same discipline pays on every platform you buy.