Target ROAS for subscriptions: bid on the first payment, or on predicted LTV?
The value you send decides what Smart Bidding chases. A practical way to send predicted value without lying to the algorithm, and when to switch.
Notes
Working notes on measurement and paid media.
Tracking that breaks, bidding that misreads, and the Gulf-market details that do not appear in platform documentation. Written by the team, from client accounts, roughly once a month.
Enhanced conversions through server-side GTM: what actually changes in your bidding
Everyone turns it on; few check whether it is doing anything. How to read the match-rate diagnostics, why normalization before hashing decides the result, and what a healthy lift looks like.
Read the noteImporting from Google is where Microsoft Ads starts, not where it ends
The five things we change after every import: match types, device bids, audience signals, UET goals, and the budget split by country.
Consent mode v2 in practice: how much of your EU data is modeled, and does it matter?
Observed versus modeled conversions by country, what the ratio does to target CPA, and the consent-banner choices that move it.
Performance Max next to Search: keeping the comparison honest
Brand exclusions, channel reporting, and the experiment design that tells you whether PMax is adding conversions or taking them from campaigns you already had.
Arabic search campaigns: the structure mistakes that waste half the budget
Mixed-language ad groups, translated instead of written ads, and landing pages that switch language mid-funnel. What a clean EN/AR account looks like.
Offline conversion import for lead gen: from form fill to won deal in the bidding
Capturing the click ID, mapping CRM stages to conversion actions, and the schedule and values that let Smart Bidding learn from deals instead of forms.
Meta and Google will both claim the same sale: a budget rule that survives it
Why attribution reports cannot settle cross-channel budgets, and the geo-holdout and baseline methods we use instead when budgets are too small for a formal lift study.