How to review Google Ads bid strategies with simulations
What each bid strategy is doing, what Google's own simulations say a change would do, and where targets fight the budget.
IntentJudge bid strategies on the platform's own counterfactuals instead of folklore, and find the strategies working against their budgets.
- 4
- Steps
- 5
- Tools
- 3
- Rules
- 3
- Failure modes
When to use it
Use it when
- Quarterly, or when CPA/ROAS targets moved.
- Campaigns limited by budget or by bid strategy flags.
- Before switching bidding modes on meaningful spend.
Do not use it when
- Fresh strategies still learning: simulations there extrapolate noise.
- Tiny campaigns: simulations need volume to exist at all.
What it needs first
Required
Preconditions
- Campaigns with automated bidding and enough history for simulations to be generated.
Inputs
scope- Campaigns to review.Default: all automated-bidding campaigns above median spend.
Procedure
In order. Every tool named is one the gateway ships, and links to its reference.
Map strategies: type, target, attached campaigns.
Read performance against each strategy's own target.
Pull the simulations: what target or budget changes would trade.
Cross with budgets to find strategies throttled by money rather than targets.
Decision rules
The observable condition, what it lets you conclude, and what takes the conclusion back.
When
A campaign hits its target CPA but is limited by budget, and the simulation shows volume at flat CPA
Conclude
The constraint is the budget, not the strategy: quantify the forgone volume from the simulation.
When
The simulation shows steeply rising CPA for more volume
Conclude
The strategy is at the efficient frontier; more volume is a price decision, present the curve.
When
Target CPA sits far below what the campaign has ever achieved
Conclude
The target throttles delivery: the strategy obeys a number nobody meant literally; propose the simulated realistic target.
Evidence every conclusion must carry
- Every recommendation quotes the simulation's own points, not extrapolation beyond them.
- Strategy performance is judged against its own target, stated.
Where the agent stops
- Target changes are strategy decisions; budget changes route through budget-reallocation and Safe Writes.
- Mode switches (manual to auto, auto to manual) are proposed with their learning-phase cost stated, never applied.
How it goes wrong quietly
The cases where the analysis is wrong and still looks right. Read them before trusting a number.
- Simulations reflect the recent auction; they expire fast and say nothing about next month.
- Portfolio strategies aggregate campaigns; judging one member campaign alone misreads the portfolio's optimisation.
- Seasonal simulations (generated during a peak) overstate what a normal week can buy.
What the answer contains
- 01Strategy map
- Type, target, campaigns, performance vs target.
- 02Simulation findings
- The trade-off curves that matter, read out.
- 03Proposals
- Target and budget changes, each with its simulated effect.