Google Ads Value-Based Bidding Migration Guide: What to Expect When You Switch

This is a Google Ads guide. The mechanics below are Google's: bid strategies, conversion goals, the learning period. Meta and LinkedIn behave differently and are not covered here.

Switching a live campaign to value-based bidding is the one change in Google Ads that lets the same budget buy better customers instead of more form fills. It works, and it works for most lead-gen accounts that clear the volume floor. What stops it working is rarely the values. It is the transition: a campaign that is re-learning looks worse before it looks better, and somebody switches it back.

The good news is that Google publishes the order to do it in, and the order is designed to keep the campaign stable the whole way through. Done in that order, the switch is a migration with a few quiet weeks in it, not a leap.

This is what to expect when you switch a live Google Ads campaign, how long each stage takes, how to tell normal adjustment from something actually going wrong, and how to get to the other side with the improvement intact.

Two changes, made one at a time

Moving to value-based bidding is two separate changes:

1. Changing the conversion signal. The campaign starts receiving a meaningful value on each lead, through a new conversion action or values added to the existing one.
2. Changing the bid strategy. The campaign moves from bidding on how many leads it gets to bidding on what they are worth.

Google's guidance is to make them one at a time, in that order. Set up the values first and keep the bid strategy you already have while Smart Bidding observes them. Google's guide to changing both puts that observation period at about three conversion cycles or four weeks, and its value-based bidding guide says four weeks or one to two conversion cycles, whichever is longer. Only then switch the strategy.

The reason is simple. Smart Bidding adjusts to every change in what it is optimising toward, and two unfamiliar inputs on the same day are harder to adapt to than one. Staging also means a wrong value scale, a wrong currency or a doubled conversion gets caught while it is still harmless, because nothing is bidding on it yet.

What triggers an adjustment period

Smart Bidding recalibrates when the definition of success changes. Google names these:

• Changing the bid strategy. Moving from Maximize conversions or Target CPA to Maximize conversion value or Target ROAS is not a tuning adjustment. The old strategy was bidding on how many leads arrived; the new one bids on what they are worth.
• Changing the conversion goal or its composition. Adding a value-carrying action to the campaign's goal, removing the old form-fill action, or changing which actions are primary.
• Changing the scale of the values. Moving from revenue to profit, or from one currency to another, is a change Smart Bidding has to re-fit to.
• Large budget or target changes. A big move in daily budget or in the ROAS target is a change of scale.

Google Ads shows the strategy status as Learning after a new strategy, a setting change or a conversion composition change. The campaign keeps serving during it. Learning means the system is adapting, not that bidding has stopped, and it is continuous: a change causes adjustment, not a reset to zero.

Keep unrelated changes out of the window. New landing pages, new ads, new targeting and a big budget change all make it impossible to tell what the switch did.

How long the learning period lasts: conversions, not days

This is the part that matters most for lead generation, and the part almost every article gets wrong.

Google gives two figures for two different things. After a change to the conversion goal a campaign bids on, it says to expect one to two conversion cycles of re-learning. For the learning period in general it says it can take up to around 50 conversion events or 3 conversion cycles for a strategy to calibrate, and it names what decides the length: how many conversions your campaigns get, how long your conversion cycles are, and which bid strategy you chose. A conversion cycle is the typical time from ad click to the conversion being recorded, and for outcomes imported from a CRM it includes the delay before they reach Google Ads.

None of that is a number of days. Work the conversion half of it on lead-gen volume:

The table is arithmetic, not a promise: campaign history, conversion delay and the strategy also affect it. Thirty conversions a month is the floor we hold to before publishing a value feed at all, so seven weeks is the realistic worst case for an account that should be attempting this. It is our benchmark, not a Google eligibility rule; Google publishes different requirements for different campaign types. Under it, learning is not your problem yet; volume is.

The conversion cycle half is independent and often binds harder. If it takes 21 days from click to the conversion you are counting, three conversion cycles is nine weeks whatever your volume does. An advertiser sending a qualified-stage conversion rather than a form fill has deliberately lengthened their own conversion cycle, and should expect to wait longer for the same reason.

How to migrate, in Google's order

1. Stabilise the budget. Set the campaign's budget to what you are prepared to spend before touching the conversion setup. A budget change on top of a goal change is two adjustment periods stacked, and Maximize conversion value without a target will try to spend the whole daily budget, so the number has to be a real limit rather than a ceiling you assume it will not reach.
2. Create the value-carrying conversion action, and leave it Secondary. Secondary actions are recorded under "All conversions" and not bid on. Confirm conversions are landing with the right times, currency, values, click IDs or enhanced-conversion data, and order IDs where you use them. One exception to check: a secondary action inside a custom goal is still used for bidding, so look at the campaign's custom goals before assuming an action is observation-only.
3. Make it Primary and move the old form-fill action to Secondary, on the same day. A lead counted twice, once as a form fill and once with a value, is a doubled conversion the strategy has to unlearn. Keep your current bid strategy. Google says to expect one to two conversion cycles of adjustment on the goal change alone.
4. Keep the count-based strategy while the values accumulate. Stay on Maximize conversions or Target CPA for at least four weeks, and longer if your conversion cycle is long. Smart Bidding is receiving the new value distribution without being asked to optimise toward it yet, and you are watching the values for anything wrong. If you are adding values to an existing primary action rather than creating a new one, the same applies: report them consistently for that period before changing the strategy.
5. Switch to Maximize conversion value. Our recommendation, below, is to start with no target. This is a second adjustment period: up to around 50 conversion events or three conversion cycles.
6. Leave it alone. Monitor delivery and the values arriving, and do not react to ordinary daily movement.

Account-level where you can, one switch across every campaign that shares the goal. Campaign by campaign only if you need the gradual version, with the experiment trade-off from the section below in mind.

Two naming notes. "Include in Conversions", the switch older guides tell you to set, was removed in 2022; Primary and Secondary are the same choice under its current name. And since June 2026 Google labels the target strategies as "Target CPA" and "Target ROAS" again, where for a few years they read "Maximize conversions with a target CPA" and "Maximize conversion value with a target ROAS". Older screenshots show the long names. The bidding is the same.

Start on Maximize conversion value, not Target ROAS

Our recommendation is to always start on Maximize conversion value with no target, and to move to Target ROAS only once the switch is mature and validated: Google out of learning, the sales cycle run, and the values confirmed against real closed deals.

The reason is what a target does during learning. Target ROAS is a promise about a ratio, and the campaign has little history on the new values to keep it with. Set the target to the return you want rather than the return you currently get, which is what almost everyone does, and the strategy responds to an unreachable target by not bidding. Spend falls to a fraction of the budget, volume disappears, and the conclusion in the room is that the values did not work. The values were never tested.

Maximize conversion value has no such failure mode. It spends the budget toward the highest total value it can find and asks you to decide nothing you do not yet know. The one thing it does ask is that the budget is the real limit, which is why the budget is step one. It is also the only setting under which the learning period measures the values themselves rather than the values plus a guess about the target.

There is a case for a target from day one, and it is worth stating fairly: after the observation period you have several weeks of observed return on the new values, which is a defensible starting point. If efficiency is a hard business constraint and you must open with a target, set it the way Google says to: from the return the account actually achieved over a recent window, such as the last 28 days, excluding the most recent days where conversions are still arriving. Never from the return you hope for. Then wait one to two conversion cycles before comparing the target with actual return in the bid strategy report, and move it in steps. Gradual is not caution here, it is the only way the strategy keeps a stable base to move from.

What normal adjustment looks like

The campaign keeps running. Learning is not a pause, and your ads do not stop serving.

What you will see is movement: daily spend that does not sit where it used to, conversion volume that shifts, cost per lead moving around, and for a while a gap between any target and observed return. The status reads Learning, and hovering over it names the reason.

Cost per lead is no longer the measure of success, because the strategy is no longer optimising for it. It may rise, and on a working switch it often does, because the campaign is buying fewer, better leads on purpose. But a higher cost per lead on its own is not evidence that the values are working either. What you are looking for is in the next section.

Separate adjustment from a problem. Wider swings around roughly the same average is learning. A sharp and sustained collapse in impressions and spend from the first day is a target set too early or too high. Missing values, duplicated conversions, a wrong currency or a sudden change in the value scale are implementation faults, and they should be fixed, not waited out.

All campaigns at once, or a few at a time?

The cleanest test is a Google Ads campaign experiment. It splits traffic and budget between the original campaign and a copy on the value-based strategy, so both halves see the same season, the same competitors and the same landing pages, and the difference between them is closer to what the switch actually did. Google supports these specifically for value-based bidding. Keep the experiment to the one change: same goals, same creative, same targeting.

If an experiment is not practical, stage the rollout, starting with the campaigns that have meaningful spend, stable tracking, enough conversion history, reliable values and a manageable conversion delay. Be honest about what that gives you: an untouched campaign is not automatically a control, because it may differ in audience, intent, geography or offer.

And each bid strategy learns on its own conversions, unless campaigns share a portfolio strategy, in which case they pool. Splitting one campaign with 60 conversions a month into two with 30 each does not give you a staged rollout; it gives you two strategies that both learn slowly. If the campaigns you would hold back clear the volume floor on their own, stage it. If they do not, switch everything and compare against your own history.

New campaigns learn differently

A brand new campaign has no history of its own, although Google says conversion data elsewhere in the account helps its initial calibration, which is why a new campaign inside an established account settles faster than the same campaign in a new one.

It is still a poor place to evaluate value-based bidding. If the campaign, the audience, the creative, the conversion action and the strategy are all new, no result tells you which one moved. Test the migration on an established campaign with known performance.

The second clock: your sales cycle

Google exiting the learning period does not mean you can read the result. Two clocks are running, and the second is much slower.

The first is Google's: 50 conversion events or three conversion cycles, as above. When it finishes, the bidding is stable.

The second is your sales cycle. The leads that arrived after the switch have to actually close before you know whether they were better leads, and a lead that arrived the week you switched is still open the week Google finishes learning. Judging a 60-day sales cycle after 30 days compares mature pre-switch leads with immature post-switch ones. Compare cohorts at the same age, and count the leads still open.

As a working rule, wait until roughly one and a half sales cycles have passed since the switch and you have at least 25 resolved deals on each side of it. That is a screening threshold, not a proof: how many deals it takes to show a real improvement depends on your baseline, its natural variation and the size of the effect, so report the uncertainty and keep an observed change apart from a proven cause.

That is a long time, and it is worth saying plainly rather than discovering it in month two. It is also why the interesting question during the wait is not "did revenue go up" but "did the mix of leads change". The composition of what arrives moves within weeks, long before anything closes: the predicted values, the company sizes and industries, the qualification rate, early pipeline progression, the disqualification reasons. Those are leading indicators, not proof of revenue, and they are the earliest honest signal that the bidding is doing something.

What not to do

• Do not change the conversion signal and the bid strategy on the same day. Let Smart Bidding observe the values under your current strategy first.
• Do not open with an aspirational target. If you must open with one, base it on recent observed return.
• Do not judge on cost per lead. Measure lead value, lead mix and mature outcomes, with cost per lead as a supporting figure.
• Do not combine the switch with unrelated changes. They make the result unreadable.
• Do not treat every fluctuation as a full reset. Smart Bidding learns continuously. Avoid unnecessary edits while you evaluate, but a change is an adjustment, not a stopwatch back to zero.
• Do not wait through broken tracking. Missing values, duplicate conversions, a wrong currency and unmatchable leads need fixing, not patience.

Before you switch at all

The bid strategy cannot rescue a weak signal. Confirm, before anything is sent:

• Google can match enough leads, through a click ID or enhanced conversions for leads.
• Values use only what is known when the lead arrives.
• Values come from resolved outcomes, not intuition.
• Outliers are capped, so one unusually large deal cannot dominate the signal.
• The values vary enough to give Smart Bidding something to prefer.
• The same lead is not counted through more than one primary action.

Five checks on the CRM records themselves covers the first two in detail. The complete guide to value-based bidding for lead generation covers the rest, including close rate times average deal size, the cap and what to do with groups too small to price.

Common questions

How long does Google Ads take to learn after switching to value-based bidding? Google's yardstick is up to around 50 conversion events or three conversion cycles, and one to two conversion cycles after a change to the conversion goal specifically. On an account getting 50 conversions a month, 50 events is about 30 days. Days are the output, not the input.

Should I change the conversion values and the bid strategy at the same time? No. Set up the values first, keep your current strategy while Smart Bidding observes them for at least four weeks, then switch. That is Google's order, and it means a wrong value gets caught before anything bids on it.

Does changing the bid strategy reset the learning period? It triggers an adjustment and a Learning status, as do conversion goal changes and setting changes. Avoid unnecessary changes while you evaluate, but Smart Bidding learns continuously and a change is not a full reset from zero.

Will my cost per lead get worse? It often rises, and on a working switch that is by design: fewer, better leads. But it does not have to, and a higher cost per lead on its own is not evidence of anything. Judge on lead value and mature outcomes.

Should I switch every campaign at once? Use a campaign experiment where you can. Otherwise stage the rollout through stable campaigns with enough conversion data, and remember that an untouched campaign is not automatically a valid control.

Should I start on Target ROAS or Maximize conversion value? Maximize conversion value, with no target. Move to Target ROAS once the switch is mature and validated. If efficiency is a hard constraint and you must open with a target, set it from the return the account actually achieved over a recent window, excluding the days still waiting on conversions.

When can I tell whether it worked? When both clocks have run: Google out of learning, and the post-switch leads as mature as the pre-switch ones you are comparing them with, which for most lead-gen accounts means one and a half sales cycles and at least 25 resolved deals on each side. Lead mix shows earlier. Revenue follows the sales cycle.

The honest summary

On the other side of the switch is a campaign that spends the same budget on the leads that actually become customers, and keeps getting better at it as every closed deal feeds the next refit. Getting there is a migration, not a switch: validate the values, let your current strategy observe them for a month, move to Maximize conversion value, allow one more adjustment period, then wait for the new leads to mature before you judge. On lead-gen volume that is two to three months end to end. The cost is real, it is temporary, and it is the price of the thing working afterwards.

What makes the wait easy is knowing which clock you are on, and not confusing adjustment with failure. Spend collapsing from day one is a target set too early or too high. Nothing changing at all after both clocks have run is a value spread problem, where the numbers you sent were too similar to rank anything. Broken tracking is broken tracking. None of those is fixed by waiting. The learning period is.

Hold the line through it and the improvement is yours to keep.

Where ValueBasedBidding.com fits

It does the part of this that is yours to do. It reads your CRM export or your HubSpot, checks every condition in the list above before anything is sent, prices each new lead from your own closed deals, and sends that value to Google Ads the moment the lead is created. The connect step walks you through the sequence above in Google's order, with the value action, the primary and secondary settings and the strategy switch each as a step. And once you have switched, it holds the verdict until the second clock has run, one and a half sales cycles and 25 resolved deals on each side, and shows you whether the mix of leads has moved while you wait. Start with the sample data, then your own export.

Sources: Google Ads Help: Duration of the learning period for campaigns and what affects it, Changing conversion goals and actions used for Smart Bidding, Change Smart Bidding strategies and change conversion goal, About Smart Bidding using value-based bidding for Search and Shopping, About value-based bidding using campaign experiments, About primary and secondary conversion actions, About Target ROAS bidding, About Maximize conversion value bidding, About bid strategy statuses and Changes to target based bid strategies. The per-volume arithmetic in the table is ours, worked from Google's stated 50 conversion events.