Open this doc once a month. Pick the variant matching your campaign’s current bidding state. Do the routine. Close it. Total time: 60 to 90 minutes per campaign.
Which variant to use
| Campaign is on. | Use variant |
|---|---|
| Maximize Clicks (or Manual CPC) | Variant A below: manual bid adjustments work |
| Maximize Conversions / Target CPA | Variant B below: manual bid adjustments are ignored; exclude instead |
Not sure? Check Settings → Bidding. The strategy name is at the top.
The two variants exist because Smart Bidding ignores manual bid adjustments. Trying to optimize a Smart Bidding campaign with the Variant A playbook wastes time and changes nothing on the platform. The mindset and the levers are fundamentally different.
Variant A: Monthly on Maximize Clicks
What this delivers: A deeper dive than the weekly rhythm, run once a month while still on Maximize Clicks. Covers the four bid-adjustment levers Google still respects: location, ad schedule, device, and demographic.
Use when: the campaign is on Maximize Clicks (Phase 1 bidding). After The Switch, use Variant B instead.
Before you start
- Campaign has at least 2 weeks of data
- Campaign is still on Maximize Clicks
- Block 60 to 90 minutes on the calendar
Steps
-
Review ad-copy performance. After 30+ days, your 3 ads per ad group have collected real data. Identify which Position 2 angle won (speed / risk / social proof / location / free quote). Document the winner. Plan to refresh next month: keep the winning angle, write 2 new ads with the winner plus one new test angle.
-
Run location optimization. Strong-CPA cities get +10% to +30% bid; weak-CPA cities get -10% to -30%; out-of-service-area cities get added as Excluded Locations.,.
-
Use Matched Locations for sub-city granularity. Sculpt bid adjustments down to ZIP codes when one ZIP within a city converts at 3x another.,.
-
Run ad schedule optimization. Hour-by-hour bid adjustments (you set 1-hour blocks during onboarding, so the data is granular). Bid up 10% to 30% on top 3 converting hours; bid down on dead hours.
-
Run device optimization. Bid up on the converting devices, bid down on the underperforming ones.
-
Run demographic optimization. Audiences → Demographics. Adjust bids by Age, Gender, Household Income tier per the data.
-
Review audience segments. Audience segments were added in Observation mode at launch. After 30 days, bid up on audiences that over-index on conversions.
-
Apply advanced call-extension bid adjustments. If 60%+ of conversions come from clicks-to-call, +10% to 20% on the call extension.
-
Send the monthly client report.
[Inferred client-comms cadence]One-page summary: spend vs budget, leads / conversions, CPL, what changed, what’s coming next month.
Watch out for (Variant A)
- Manual bid adjustments are about reallocation, not creation. You’re bidding more aggressively where conversions already happen. You’re not creating leads out of thin air.
- Don’t bid up something that’s “almost converting.” If the data is thin, bid up makes the data thinner. Wait for stable signal.
- Don’t move all 4 levers at once. If everything changes, attribution is impossible. One or two levers per monthly cycle.
Done when (Variant A, each month)
- ✅ Ad-copy performance reviewed; winning Position 2 angle documented
- ✅ Locations bid-adjusted; out-of-service cities excluded
- ✅ Matched locations checked for ZIP-level sculpting
- ✅ Ad schedule bid-adjusted by hour
- ✅ Device + demographic + audience adjustments applied
- ✅ Call extension adjustments applied if call-heavy
- ✅ Monthly client report sent
Then check Gate 2
After each Variant A run, check whether Gate 2: Transition Readiness is green. The thresholds: 15-30+ conversions accumulated, weekly CPA within ±25% of monthly average, search terms mature, daily conversion volume consistent. When Gate 2 is green, open optimization-the-switch.md and execute The Switch. From the next monthly cycle forward, you use Variant B.
Variant B: Monthly on Smart Bidding
What this delivers: The post-Switch monthly rhythm. The mindset is fundamentally different from Variant A. Under Smart Bidding, manual bid adjustments are ignored by Google’s algorithm. The model shifts from “steer with bid adjustments” to “give the algorithm clean data and structural control.”
Mindset, in one line: exclude, don’t adjust.
Use when: the campaign is on Maximize Conversions or Target CPA (Phase 2 bidding). Before The Switch, use Variant A.
What still works under Smart Bidding
| Lever | Works under Smart Bidding? |
|---|---|
| Location bid adjustments (%) | ❌ Ignored |
| Device bid adjustments (%) | ⚠️ Only -100% (full blackout) is respected |
| Ad schedule % adjustments | ❌ Ignored: only hour-removal is respected |
| Demographic bid adjustments (%) | ❌ Ignored |
| Audience bid adjustments (%) | ❌ Ignored |
| Excluding locations entirely | ✅ Works |
| Removing hours from the schedule | ✅ Works |
| Splitting campaigns by performance tier | ✅ Works |
| Tuning the Target CPA | ✅ Works (your primary lever) |
Before you start
- The Switch (optimization-the-switch.md) is complete and Gate 3 was passed
- Campaign has been on Smart Bidding for at least 30 days; performance has stabilized post-learning
- Block 60 to 90 minutes on the calendar
Steps
-
Run the Hour & Day report. Use the Report Editor for hour-by-day granularity. Exclude bottom-performing hours entirely; don’t bid-adjust them down (ignored).,.
-
Run the locations report. For underperforming locations, exclude entirely or split into a separate campaign with a different Target CPA. Don’t bid-adjust them down.
-
Audit audience segments. Bid modifiers ignored. Either let observation-mode segments run for data, or restrict the campaign to high-converting audiences.
-
Audit demographic performance. Same rule: bid modifiers ignored. To act on insights, exclude underperforming demographics entirely.
-
Audit device performance. Only -100% (full blackout) is respected; any other percentage is ignored. Use -100% to disable a device entirely if data supports it.
-
Tune the Target CPA itself. This is your primary lever now. CPA running 20%+ below target → lower Target CPA by 10%. CPA 20%+ above target → raise Target CPA by 10% or accept lower volume.
-
Consider campaign splits for performance tiers. If one geo or service segment converts at 3x the rate of another, split into separate campaigns with independent Target CPAs.
-
(Optional) Consider offline conversion uploads. Pointer only: see M11-Offline-L02 (Conversion Depth Method), M11-Offline-L05 (booked appointments), M11-Offline-L06 (revenue uploads). Advanced layer for stable Smart Bidding campaigns; covered in a future advanced reporting playbook.
-
Send the monthly client report.
[Inferred client-comms cadence]Same shape as Variant A’s report. Highlight what changed under Smart Bidding (Target CPA, exclusions) and what didn’t (no bid adjustments).
Watch out for (Variant B)
- Don’t add bid modifiers expecting them to work. They don’t. If you find yourself reaching for a +20% city bid, either exclude that city or split it into its own campaign.
- Don’t tune Target CPA more than once per month. Smart Bidding needs time to settle after each change. Constant tweaking resets the learning.
- Don’t exclude segments based on one bad week. Wait for at least 30 days of consistent underperformance before excluding.
Done when (Variant B, each month)
- ✅ Hour & Day report run; bottom hours excluded
- ✅ Location performance reviewed; underperformers excluded or split
- ✅ Audience + demographic + device exclusions applied where data supports
- ✅ Target CPA tuned ±10% based on 30-day post-Switch performance
- ✅ Campaign splits considered if performance tiers diverged
- ✅ Offline reporting (M11) considered as future advanced layer
- ✅ Monthly client report sent
Where to go from here
- Tomorrow: optimization-daily.md
- End of week: optimization-weekly.md
- End of next month: back here, run Variant B again
Variant B repeats every 30 days indefinitely. There is no “graduation” from this rhythm.