Quick Answer: Reverting to Max Conversions after a failed tCPA test doesn't restore your campaign's old performance. Every bid strategy change resets the learning phase. If your account is still underperforming months after switching back, the fix isn't patience -- it's a fresh campaign with a single, consistent bid strategy from the start.
The Setup Everyone Recognizes
You had a Max Conversions campaign running well. Consistent leads, stable CPA, no major complaints. You decided to introduce a tCPA target to push for more efficiency. For the first week or two, things seemed fine. Then conversion volume dropped. CPA climbed. You gave it a few more weeks hoping the algorithm would sort itself out.
It didn't. So you switched back to Max Conversions.
That was months ago. Performance is still off -- not catastrophically, but noticeably worse than before you touched anything. You're wondering whether you've permanently damaged the campaign or whether you just need to wait longer.
Neither framing is quite right.
What Actually Happens When You Switch Bid Strategies
Every time you change your bid strategy in Google Ads, the campaign re-enters the learning phase. This is documented Google behavior, not speculation. The algorithm was building a predictive model around one optimization goal -- finding users likely to convert at whatever CPA it was targeting. When you change that goal, it has to start building a new model.
Your historical conversion data doesn't disappear. It's still in the account. But the behavioral model the algorithm built from that data -- the one that was driving your results -- doesn't carry over. There's no restore point. Switching back to Max Conversions doesn't re-activate the old model. It starts a new learning cycle with Max Conversions as the objective.
If you switched from Max Conversions to tCPA and then back to Max Conversions within a short window, you reset the learning phase twice. Each reset consumes budget and time. Each one requires a fresh data accumulation period before Smart Bidding can make confident decisions. Do that in rapid succession and you've spent a significant chunk of your conversion volume just on reorientation -- with nothing stable to show for it.
Why the Campaign Feels Stuck
The algorithm isn't broken. It's not holding a grudge. What's happening is more mundane: the campaign's recent conversion history is fragmented across multiple bid strategy modes, and the current model is working from an inconsistent signal.
Smart Bidding works best when conversion data is dense, consistent, and all pointing at the same goal. When the account history shows gaps, shifts, and reversals in optimization objective, the algorithm treats the recent data as the most relevant signal. If recent data is thin or noisy from the strategy changes, performance reflects that.
Waiting for the campaign to "settle" can work -- but only if you've stabilized the bid strategy and conversion volume is high enough for the algorithm to rebuild its model at a meaningful pace. If you're generating fewer than 30 conversions in 30 days, recovery on a campaign that's been through multiple bid strategy changes can take a very long time, if it happens at all.
If you want Signal Rich Structures to walk you through how campaign architecture affects the quality of signal you're feeding the algorithm -- from bid strategy to conversion setup to budget thresholds -- it's a free course at freak.marketing/signal-rich-structures.
How to Actually Fix It
Step 1: Audit the current campaign's recent conversion volume
Look at your campaign's conversion data for the last 30 and 60 days. Not account-level -- campaign-level. You're checking whether conversion volume is below the threshold where Smart Bidding can make reliable decisions (roughly 30 conversions in 30 days for tCPA; Max Conversions has a lower bar but still needs consistent volume).
If volume is thin and irregular, the campaign is working with degraded signal. Waiting longer on this campaign won't substantially change that.
Step 2: Pick one bid strategy before you do anything else
The most important decision here is not which bid strategy is "best." It's which one you're willing to commit to without touching for at least 30 to 45 days. Consistency beats optimization in this situation.
If your conversion volume is below 30 conversions in 30 days, Max Conversions is the right choice. Do not add a tCPA constraint. The constraint limits the algorithm's ability to find conversions during the learning phase, which is the opposite of what you need right now.
If your conversion volume is consistently above 30 conversions in 30 days, you can consider tCPA -- but only with a target set at or above your recent average CPA. Setting a tCPA significantly below your current CPA is one of the most common ways to tank a campaign immediately after switching.
Step 3: Duplicate the campaign
Don't try to nurse the existing campaign back to health. Duplicate it. Copy the structure -- ad groups, keywords, ads, settings -- into a new campaign. Launch the new campaign with the bid strategy you chose in Step 2.
The duplicate starts fresh. No fragmented learning history. The algorithm gets a clean run from the start with one consistent optimization goal.
Step 4: Let conversion data accumulate before touching anything
The most damaging thing you can do at this point is intervene early because performance looks inconsistent in the first two weeks. It will look inconsistent. That's the learning phase. Set a calendar reminder for 30 days and leave the campaign alone unless something is genuinely alarming (like CPA at 5x your target with zero signs of improvement).
Once the campaign has 30 to 50 conversions under the new bid strategy, you'll have a much cleaner read on whether it's working.
Step 5: Pause the old campaign once the new one stabilizes
Don't run both simultaneously for extended periods. If both campaigns are targeting the same keywords and audiences, you're splitting your conversion data between two campaigns and slowing the learning on both. Once the new campaign has stabilized and is delivering comparable or better results, pause the original.
FAQ
Can I just wait for the original campaign to recover on its own?
Sometimes, yes -- but only if you've stopped changing the bid strategy and conversion volume is sufficient for the algorithm to rebuild its model. If it's been more than 60 days with no meaningful improvement, the campaign is likely settled into a degraded equilibrium. A fresh duplicate is faster and more reliable than waiting.
Will duplicating lose my quality score and ad history?
Quality Score is keyword and ad-level data that lives at the account level, not inside a single campaign. A duplicate campaign targeting the same keywords will inherit the quality signals attached to those keywords. You don't lose that by starting a new campaign.
How do I avoid this next time I want to test tCPA?
Use campaign experiments. You can run a tCPA test as a split against your existing Max Conversions campaign. If the test underperforms, you end it and the base campaign never left its original bid strategy. If it wins, you roll it out. No resets, no recovery periods.
What's the right tCPA target when I'm ready to try it?
Start at or slightly above your recent average CPA, not below it. If your average CPA is $80, a tCPA of $60 puts the algorithm in a position where it has to be overly selective to hit the target -- which usually means sharply reduced volume. Give it room to find conversions first, then tighten the target incrementally once volume is stable.