Changing Google Ads Bid Strategies: When to Switch and How to Avoid Crashing Your Campaign

The bidding mistake we see most often in eCommerce accounts is a strategy change made too soon, too aggressively, or without knowing what sets off the learning period. A campaign that was doing fine goes quiet for a fortnight. ROAS falls through the floor and spend nearly stops. Someone then makes another change to fix it, and the whole cycle starts again.

This guide explains how Smart Bidding actually works, when an account has enough data to move between strategies, how to make the move without destabilising the campaign, and how to recognise the specific ways bidding goes wrong.

How Smart Bidding Works

The Smart Bidding strategies (Maximise Conversions, Maximise Conversion Value, Target CPA and Target ROAS) use Google’s machine learning to set a bid at the moment of each auction. Rather than you choosing a fixed bid, Google estimates how likely that particular auction is to end in a conversion and bids to match.

That estimate draws on signals from the auction itself: device, location, time of day, the search query, browsing history and hundreds of other factors. The more conversion data a campaign has, the better Google can judge conversion probability and the more accurate the bids become.

This is why conversion volume matters so much. With only a handful of conversions, the estimates are shaky and bids jump around. As conversions build up, the model sharpens and bids settle down.

The Four Strategies and When to Use Each

Maximise Conversions: Google bids to win as many conversions as it can within your daily budget, with no ROAS or CPA limit. This is the right choice while you are still gathering data, because a campaign needs conversion history before an efficiency target means anything.

Use it for: new campaigns, account restructures, campaigns with fewer than 30 conversions in the last 30 days, or situations where you have a budget to spend and volume comes first.

Maximise Conversion Value: Google aims for the highest total conversion value (revenue) instead of the highest number of conversions. It suits accounts where conversions are worth different amounts (products at different price points) and you want Google to lean towards bigger orders rather than simply more orders.

Use it when: purchase revenue is tracked correctly, your catalogue has real price variation, and you want Google to favour higher value sales.

Target CPA (tCPA): Google bids to hit the cost per acquisition you set. It fits lead generation, or any case where a steady CPA ceiling matters more than volume.

For eCommerce it is usually a weaker fit than tROAS because order values vary. tCPA treats a £15 sale and a £300 sale as the same conversion.

Target ROAS (tROAS): Google bids to hit the return on ad spend you set. You enter a ratio (for example 400%, meaning £4 of revenue for every £1 spent) and Google works towards it. For most mature eCommerce campaigns this is the end goal, because it ties efficiency to the value of each sale.

Use it when: conversion volume is steady, conversion values are accurate, and profitability matters more than raw volume.

How Many Conversions Do You Need?

“How many conversions before I switch to tROAS?” has a real answer, but it is not one fixed number.

Google’s official guidance is 50 conversions in 30 days before you add a ROAS target, and that is a sensible floor. Below 50, the model cannot estimate conversion probability well, and putting a ROAS target on top of an uncertain model tends to go one of two ways: the campaign cuts spend hard because it cannot tell which auctions will hit the target, or it overshoots in ways that are difficult to predict.

In practice, accounts with 15 to 30 conversions a month can sometimes run tROAS successfully when the conversion data is very consistent and the target is cautious. The risk of instability below 50 is much higher, though.

Consistency matters more than the raw count. Fifty conversions spread evenly over 30 days give the model far more to work with than fifty clustered around a single sale.

If you are under the threshold, run Maximise Conversions without a CPA limit. The job at this stage is building conversion history, not squeezing efficiency. Putting constraints on a campaign that lacks data works against you.

When to Move From Maximise Conversions to Target ROAS

Make the switch when:

• The specific campaign you are changing has 50 or more conversions in the last 30 days
• It has run for at least two weeks without major structural changes
• Conversion tracking has been checked (the right events, no double firing, purchase revenue recorded correctly)
• You have a realistic ROAS target based on what the campaign has actually achieved

That last point is where most transitions go wrong. People set the ROAS they would like rather than the ROAS the campaign delivers. If a campaign has been running at 350% on Maximise Conversions and you set a 600% target on day one, it will cut spend sharply. It will only enter auctions where it expects 600%, those are rare, and volume collapses.

Setting the first target: take the campaign’s average ROAS over the last 30 days on its current strategy and set tROAS at about 85 to 90% of that figure. This leaves the campaign room to work while nudging efficiency up. If it holds volume and hits target for two weeks, raise the target by 10 to 15% and watch for another two weeks.

Tightening step by step is far less disruptive than jumping straight to an aggressive target.

The Learning Period and Why It Matters

Whenever you make a significant change to a Smart Bidding campaign, Google marks it as learning. While it is learning:

• Bids are less accurate than usual
• Performance normally sits below the campaign’s baseline
• Google’s documentation says this lasts one to two weeks, but campaigns with lower conversion volume can take three to four

Changes that start a learning period:

• Switching bid strategy
• Moving a tROAS or tCPA target by more than 15 to 20%
• Changing the budget significantly
• Adding or removing conversion actions
• Major structural changes (new ad groups, large keyword additions)

The takeaway: never stack significant changes on top of each other. Each one restarts learning, and a campaign that is constantly being changed never leaves the learning phase or reaches its potential.

If several changes are needed, space them at least two weeks apart. Make the most important one first, let the campaign settle, then make the next.

Why tROAS Is Cutting Your Spend Too Hard

If spend dropped sharply after moving to tROAS, the usual causes are:

The target is higher than the campaign can realistically achieve. It enters fewer auctions because few are predicted to reach your number. Fix: bring the target closer to the historical average and let volume rebuild before tightening again.

Conversion volume fell around the same time. A strategy change sometimes lines up with a natural dip in demand (seasonality, competitor activity) or a tracking problem. Separate the effect of the bid strategy from outside factors before blaming the switch.

The campaign is still learning. Unstable spend in the first one to two weeks after a change is normal. If spend is still suppressed after three weeks, dig deeper.

The budget is too small for the target. With a £50 daily budget and an 800% tROAS, the campaign may find very few auctions it is confident about. A budget that heavily limits available auctions can make tROAS underdeliver. Raise the budget or lower the target.

Rescuing a Campaign That Crashed After a Strategy Change

When a strategy change has knocked a campaign over, the order in which you recover matters.

Do not jump straight back to Maximise Conversions. Switching again restarts learning for a third time and makes the hole deeper.

Instead:

• If you moved to tROAS with an aggressive target, lower it to match historical ROAS and give it two weeks before judging.
• If the campaign is spending but ROAS is weak, confirm nothing changed in conversion tracking. A tracking issue that appeared alongside the strategy change is a common source of confusion.
• If the campaign has almost stopped spending, check whether the budget is now the bottleneck. A tROAS target on a small budget is often the cause.
• If it is still clearly underperforming after three to four weeks, going back to Maximise Conversions to rebuild data is a valid option. Just accept that it costs another two weeks of learning.

The Bid Strategy Lifecycle for a New eCommerce Campaign

For a campaign starting from zero:

Months 1 to 2: Maximise Conversions with no CPA or ROAS target. Set a budget that allows at least two to three conversions a day. The aim is conversion history, so do not judge efficiency yet.

Months 2 to 3 (once you have 50+ conversions): Add tROAS at a cautious level, 80 to 90% of the ROAS you have actually achieved. Watch for two weeks. If volume holds and the target is met, raise it by 10%.

After that: keep tightening in 10 to 15% steps, with two weeks of observation between each. The ceiling is the point where the campaign still delivers acceptable volume. If volume collapses at a certain target, you have found the efficiency limit at your current scale. Growing past it means accepting a lower target or improving conversion rate and feed quality so the campaign can compete in more auctions.

The same lifecycle applies to Shopping, PMax and Search. Thresholds and timings are similar across campaign types, although PMax usually needs more conversion data before a tight target makes sense.

The One Rule That Prevents Most Bidding Problems

Change one significant thing at a time, and let the learning period finish before you judge the result.

Most bidding crises follow the same pattern: a change is made, it does not work straight away, a second change is made to fix it, then a third, and six weeks later the campaign has been learning the whole time and its performance no longer reflects any clear strategy.

Change something. Wait two weeks. Review. Then make the next change. It is slower, but it is the only way to know what is actually working.