There are accounts still on manual bidding in 2026, almost always because of a bad experience: "we tried Smart Bidding and the CPA went through the roof". In nearly all those cases the transition was made in one go, without preparing the measurement and with a made-up target. This lesson is the plan for doing it properly: four phases, no sharp drop and clear criteria for going back.
Phase 0 · Is the account ready?
Before touching the strategy, three checks:
- Honest measurement: one or two primary actions that mean money, values assigned, no duplicates, advanced Consent Mode. If this fails, the transition will fail: Smart Bidding will learn from bad data (the Measurement module).
- Volume: at least 30 conversions a month in the campaign (or in the portfolio you build). With fewer, use a portfolio or "Maximize conversions" with no target as a bridge.
- Stability: no recent structural changes (new campaigns, a website redesign, a change of offer). Pick a quiet month.
Phase 1 · The bridge strategy
Do not jump from manual to target CPA. The gentle route:
| From | Intermediate step | Destination |
|---|---|---|
| Manual CPC | Enhanced CPC (eCPC) for 2-4 weeks: Google adjusts your bids up and down, you keep the base | Maximize conversions → target CPA |
| Enhanced CPC | Maximize conversions with no target, on the current budget, for 2-4 weeks | Target CPA |
| Maximize clicks | Maximize conversions with no target | Target CPA / ROAS |
The bridge has two jobs: letting the algorithm build up learning from your conversions, and letting you see what the real CPA is with automated bidding before you set a target.
Phase 2 · The first target
After the bridge, look at the real CPA (or ROAS) of the last 3-4 weeks and set the target at that figure or 10% worse (higher on CPA, lower on ROAS). Yes, worse: the first week with a target is a calibration week, and a demanding target from day one strangles the volume. Then you tighten it in steps (Module 5 of the Basic level).
Keep the budget where it is: if you raise it at the same time, you will not know what caused what.
Phase 3 · The four weeks of watching
| Week | What is normal | What is alarming |
|---|---|---|
| 1 | "Learning" status; the average CPC changes (usually up); slightly irregular spend | Spend halves overnight (an impossible target) |
| 2 | Conversions similar to the previous period; CPA ±20% | Double the CPA with no sign of improvement; conversions at zero (measurement?) |
| 3 | It settles; the strategy report shows top signals | Still learning (have you touched something?) |
| 4 | Decision time: compare the 4 weeks with the previous 4 (conversions, CPA, IS) | — |
Absolute rule: do not touch anything during those four weeks unless there is a confirmed alarm. Every change resets the clock.
The symptoms that look frightening and are not a problem
- The average CPC goes up: Smart Bidding pays more for the clicks that convert and almost nothing for the ones that do not. The average CPC can rise while the CPA falls. Look at the CPA.
- The average position drops on some keywords: the algorithm has decided they do not deserve the top spot. Look at the CPA.
- Daily spend swings about: normal with an average daily budget; judge by the week.
- "Star" keywords with less traffic: they may have been converting worse than you thought. The search terms report will tell you.
When to go back (and how)
Go back to the previous strategy if, after four weeks of touching nothing and with the measurement verified, conversions have fallen by more than 30% and the CPA has not improved. Before going back: check that the target was not impossible and that there were no external changes. And if you do go back, use a bridge for that too (Enhanced CPC), not pure manual.
The transition in large accounts
Campaign by campaign, not the whole account at once: start with the one with the most conversions (it learns fastest and the risk is diluted), carry on with similar ones grouped into portfolios, and leave brand until last (or on impression share).
💡 Ninja trick: the most delicate moment of the transition is phase 3: the temptation to meddle. The Suite's Smart Bidding (SBNS) has a TEST mode for exactly this — every night it analyses and logs what it would do (move a target, a budget) without applying it — and a judge that compares performance against the business target over adequate windows. Going through the transition with the script in TEST is having a second pair of eyes that never panics.
What you should remember
- Phase 0: honest measurement, volume, a quiet month.
- Phase 1: a bridge (eCPC or Maximize conversions with no target) for 2-4 weeks.
- Phase 2: the first target = the real figure or 10% worse; budget left alone.
- Phase 3: four weeks without touching anything; a rising CPC is not an alarm, a sustained double CPA is.
- Only go back with conversions down 30% and no CPA improvement — and with a bridge there too.