The 8-Week Bidding Decision Log
Objective: Walk an 8-week Demand Gen campaign through its learning phase, the Maximize Conversions to tCPA switch, and the view-through conversion signal, making the same timing decisions a live account manager would face.
You manage paid media for RXBAR, the protein bar brand, and just launched a Demand Gen campaign to warm shoppers who've been engaging with fitness and meal-prep content but haven't searched for a protein bar yet.
Decide when to leave Maximize Conversions alone, when to switch to tCPA and at what target, and how to read view-through conversions once tCPA is running.
Before you start
What you'll need
Free path (everything below is enough to finish)
Native to the campaign, no export needed for weekly checks
Free once linked to the Google Ads account, no separate license
The process
Simulation
Week 1, first dashboard check
Day 5 of 56You launched the campaign on Maximize Conversions 5 days ago and are opening the dashboard for the first time.
Demand Gen · RXBAR Fitness Content Prospecting · Day 5 of 56 Impressions 142,000 CTR 0.9% Conversions 6 Cost per conv $31.40
Five days in, CPA looks high and volume looks thin. What do you do?
Final deliverable
A week-by-week bidding decision log across the 8-week campaign, documenting when to hold, when to switch to tCPA and at what target, and how the view-through signal was interpreted.
See a reference example
Instacart, Demand Gen bidding log (excerpt) Week 1: Maximize Conversions, CPA $34.80 on 4 conversions. Held, sample too small. Week 4: 61 trailing-30-day conversions, observed CPA $24.60. Set tCPA at $31 (26% above observed). Week 8: Direct conversions flat at ~50/week, view-through up 72%, assisted conversions up 64%. Read as mid-funnel success, kept running as a Search feeder.
Success criteria
You're done when you can:
- Does not switch bidding strategy before the learning phase completes
- Sets tCPA within the 20-30% above observed CPA band
- Correctly interprets rising view-through/assisted conversions as a positive mid-funnel signal, not a failure