Documentation checked 26 August 2026
Google quietly rewrote a sentence in its Google Ads help page for Maximize conversion value last week. The old version told you to wait four weeks or three conversion cycles before judging the strategy. The new version says one to two conversion cycles. For an ecommerce store, that edit is a real speed-up and good news for their paid search advertising. For a roofing company with a five-week gap between form fill and signed contract, the same sentence, read the way most people will read it, tells you to make a decision on data that has not finished arriving.
The sentence did not get less true. It got easier to misread.
What changed in the documentation
Here is the old guidance, as it appeared until recently and as it is still quoted in most third-party guides you will find on the subject:
wait for four weeks or three conversion cycles
And here is the sentence live on Google’s help page today, in full:
“For Search and Shopping campaigns, if you recently started reporting conversion value or changed the way conversion value is reported, we recommend you include the new values in the ‘Conversions’ column and wait for 1-2 conversion cycles for your campaign to receive conversion values at a similar rate before adopting.”
Search Engine Roundtable spotted the change on 24 August. I checked the live page on 26 August to confirm it is still there, and it is.
Two details matter more than the edit itself.
First, the four-week floor is gone. The old wording gave you a calendar backstop: whatever your conversion cycle was, you waited at least a month. The new wording removes it and leaves only cycles, which means the entire recommendation now rests on a phrase most advertisers have never actually measured for their own business.
Second, this is not a one-off. The same one-to-two-cycle language now appears on Google’s value-based bidding page for Search and Shopping, on its page about changes to target-based bid strategies, where it reads “after making budget increases, wait 1 to 2 conversion cycles to evaluate the campaign’s performance,” and on its learning period page, which says a bid strategy can take “up to 3 weeks or 1-2 conversion cycles” to calibrate to a new objective. Four pages, same phrase. This was a deliberate, systematic edit across the bidding documentation.
And to be fair to the edit: for a lot of advertisers it is genuinely better advice. If you sell direct to consumer and the conversion happens in the same session as the click, your conversion cycle is measured in hours. Three cycles was needlessly conservative. Those advertisers were sitting on their hands for a month waiting for a signal that had already arrived. For them, this change is a real speed-up.
Contractors running paid search are not those advertisers.
Why “conversion cycle” is the load-bearing word
Every argument in this article rests on one phrase, and Google never defines it in plain language on the page where it matters. Most PPC advice repeats the phrase without unpacking it either.
A conversion cycle is your lead-to-close lag, not a calendar week
A conversion cycle is the time between the click that starts a customer’s journey and the conversion you are actually optimising toward. Not the lead. The conversion whose value you are uploading.
If you are bidding to maximise conversion value and the value you upload is the value of a signed contract, then your conversion cycle is the time from click to signed contract. For a homeowner with water coming through the ceiling, that might be two days. For a full replacement with an insurance claim in the middle, it might be six weeks. For tendered commercial work, it can be a quarter or more.
Read the sentence literally, and one to two conversion cycles for a business with a five-week lag is two to ten weeks. Read it the way most people will read it, as “a couple of weeks, then,” and you start judging a Google Ads value bidding strategy on data that has not finished arriving.
Google’s own documentation agrees with the literal reading, in the fine print on a different page. The same value-based bidding guidance says to upload values for “3 weeks or 1-2 conversion cycles, whichever is longer,” recommends conversion delays “less than 7 days,” and then says this:
“If no conversion data is uploaded for 7 days post-click… the initial ramp up period for value-based bidding may take several months.”
Several months. That is Google, describing accounts like yours, on its own help pages. The headline guidance and the fine print are not in agreement, and contractors live in the fine print.
How to measure yours
You cannot use a number from an article for this, including mine. You need your own, and it takes about twenty minutes.
Export two dates from your CRM for every job won in the last twelve months: the date the lead was created, and the date it closed won. AccuLynx, JobNimbus and ServiceTitan will all give you this in a standard export. Subtract one from the other, and you have a list of lag times in days.
Now take the median of that list, not the mean. This matters more than it sounds. One tendered commercial job that took nine months to sign will drag your average into fiction, and you will end up planning around a number that describes no job you have ever actually done. The median tells you what a typical job does. That is what you want.
The median lag is your conversion cycle, and it is the number the industry usually calls conversion lag. Everything else in this article is downstream of it.
Your lag table
Fill this in from your own export. The point of the table is not the figures, it is that different job types have genuinely different cycles, and an account that mixes them is being judged on a blended number that fits nothing.
| Job type | Jobs in sample | Median lead-to-close | 1 cycle | Earliest fair evaluation date |
|---|---|---|---|---|
| Emergency repair | ||||
| Scheduled repair | ||||
| Full replacement | ||||
| Insurance or storm claim | ||||
| Tendered commercial |
Swipe sideways to see the full table.
If your replacement work runs a 38-day median, one cycle is 38 days and two cycles is roughly two and a half months. That is the honest evaluation window for that campaign, and it is nowhere near a fortnight.
Two warnings about what to do with this table.
If any median comes out above 90 days, you do not have a patience problem, you have a structural one. Google keeps the click identifier that ties a closed job back to its ad for 90 days and no longer. Its own guidance is explicit: “We only keep the GCLID for 90 days. We recommend you upload more frequently, or if your conversion happens after 90 days, upload a different offline conversion event that happens within 90 days.” Past that line, closed-won value cannot be attached to the original click at all, and the fix is not waiting longer. It is optimising toward an earlier event, which is covered below.
And measuring lag per job type does not mean splitting campaigns per job type. The conversion that best represents real value is usually the one with the least volume, and dividing a rare conversion across four campaigns leaves each one with too little for the bidding to learn anything. Measure separately, structure conservatively.
What goes wrong when you judge too early
Ask this question in any PPC or Google Ads community and you will find the same post, over and over, usually around week three: the money is going out, nothing is coming back, and as one recent thread put it, it feels like the algorithm is “just spending to spend.”
Sometimes that is exactly what is happening. Sometimes it is the opposite. In the accounts KLEXA takes over, the pattern is so consistent it is almost a script.
Weeks one and two, the campaign runs and generates leads. The values for those leads have not been uploaded yet, because the jobs have not closed yet. Week three, someone opens the account and sees spend up, conversion value flat, and a cost per conversion value that looks alarming. The strategy gets switched off, usually with a sentence like “value bidding isn’t working for us,” and the ad spend goes back to whatever came before.
Switching it off resets Smart Bidding’s learning. The account goes back to whatever it was doing before. Six weeks later the values from weeks one and two finally land, but they land in an account that is no longer using them, attributed to a strategy that is no longer running. Nobody goes back to check. The conclusion sticks, and it gets repeated to the next agency.
The cruel part is that a strategy failing and a strategy waiting look identical in week three. Same flat value column, same ugly ratio. So here is how to tell them apart, because it is not guesswork.
| Signal | The strategy is genuinely failing | The values just have not landed |
|---|---|---|
| Conversion volume | Down against the previous period | Normal or up |
| Search terms report | Drifting to junk, new irrelevant queries | Stable and on-intent |
| Lead quality per your CRM | Worse: unqualified, out of area | Fine, they are just still open |
| The value column | Flat while closed-won deals exist in the CRM | Flat because nothing has closed yet |
| Closed-won dates | Scattered, no pattern | Clustering after your evaluation date |
| Offline import log | Uploading on schedule, values genuinely low | Gaps, or uploads lagging days behind closes |
Swipe sideways to see the full table.
If the right-hand column describes your account, you do not have a bidding problem. You have a calendar problem.
The new metric that lets you audit your own values
There is a second reason this is a good week for anyone managing paid search to be looking at value bidding, and it is more useful than the wording change.
What original conversion value reports
Google added a metric called original conversion value to the Google Ads API in version 25.1 (v25.1), released on 19 August 2026. It has existed as a column in the interface since around November 2025, so depending on where you work, you may have had it for a while without noticing. Google’s definition:
“Represents the original conversion value from biddable conversions. This is the unadjusted value of conversions before any value rule adjustments, such as conversion value rules or lifecycle goals adjustments are applied.”
Put plainly: conversion value is what Google optimised on. Original conversion value is what you actually sent it. Until now, if those two numbers disagreed, you had no clean way to see it.
Why it matters if you run value rules
Conversion value rules let you adjust the value of a conversion based on conditions: audience, geo location, device, and itinerary if you are in travel. Plenty of contractors run one without thinking of it as a rule, most often a geo rule that up-weights the metro where jobs are bigger, or a device rule reflecting that mobile callers close better.
The moment you run one of those, the value in your reports is not the value you uploaded. It is your value with Google’s adjustment applied on top. Google’s API documentation says so directly: conversions value and all conversions value “reflect modifications made by conversion value rule sets.” If your value bidding performance shifts, you have had no way to tell whether your actual results changed or whether your own rule moved the number. Now you can separate them.
The diff report
Pull both metrics for the same period and put them side by side, segmented so you can see which rule dimension is responsible.
In the API, that is a GAQL query selecting metrics.original_conversion_value and metrics.conversions_value, segmented by segments.conversion_value_rule_primary_dimension, which returns values including ORIGINAL, NO_RULE_APPLIED, GEO_LOCATION, DEVICE and AUDIENCE. In the interface, add the original conversion value column alongside conversion value and segment by month.
Either way, the report you want has these columns:
| Month | Original value (what you uploaded) | Adjusted value (what Google optimised on) | Delta | Delta % | Dominant rule dimension |
|---|---|---|---|---|---|
Swipe sideways to see the full table.
Run it for the last six months. What you are looking for is a delta that grows over time, or one concentrated in a single dimension. That is your own rule quietly reshaping the picture Smart Bidding is responding to. If your value bidding looks strong in the interface and flat in the CRM, that gap is the first place I would look.
A measurement setup that survives a long sales cycle
None of the above works without plumbing that can carry a value from your CRM into Google Ads across weeks rather than minutes. Four parts, in order of how often I find them broken.
GCLID captured at form fill and stored on the CRM record. This is the single point of failure in most contractor setups. The click identifier arrives with the visitor, and if your form does not capture it into a field that survives into the CRM record, then when that job closes eight weeks later there is nothing to tie the revenue back to. No GCLID, or its wbraid and gbraid equivalents, means no offline import, no value bidding. If your uploads are already failing with “the imported Google click ID (GCLID) could not be decoded,” this is where to look first. And do not expect perfection: some proportion of clicks never match, for reasons ranging from privacy settings to people switching devices. A steady loss is normal. A sudden change in your match rate is the alarm.
Staged conversion actions, with one of them primary. Lead, qualified, quoted, won, each as its own conversion action with a value that reflects what that stage is genuinely worth to you. This is what gives Smart Bidding something to learn from in weeks one and two while the real closes are still pending, which is precisely the gap that kills accounts in week three. It is also mandatory rather than optional if your median lag runs past that 90-day GCLID window, because an earlier event is the only thing you can still attach to the click.
Two ways to get this wrong, both common. Do not set every stage as a primary conversion action. If lead, qualified and won are all primary, Google adds their values together, so a single job worth $50 gets recorded as $80 across three actions, and you have taught Smart Bidding to chase a number that does not exist. Pick the one stage you want bidding optimised toward, make that primary, and track the rest as secondary. And do not invent the values. A made-up number teaches the algorithm a made-up preference, and it will pursue it faithfully and expensively. If you do not know what a qualified lead is worth yet, work it out from your own close rates before you switch anything on.
A conversion window long enough to cover your median lag. Default windows are frequently shorter than a contractor’s sales cycle, which means conversions that arrive after the window closes are invisible to Smart Bidding no matter how diligently you upload them. Set the Google Ads conversion window against the median you measured, with headroom. And if you are considering target ROAS specifically, rather than plain Maximize conversion value, check the target ROAS entry requirements first: Google asks for at least 15 conversions in the last 30 days at the conversion tracking level, and two or more distinct values being reported. Long-cycle accounts trip that gate more often than they expect.
Offline conversion imports on a schedule, not ad hoc. Google’s guidance is that daily uploads are optimal, and it warns that Smart Bidding performance suffers from delays and from backfilling batches of old data. A weekly cron beats an enthusiastic monthly upload every time. If your imports happen when someone remembers, they are not a system.
One thing changed this summer that most guides on this subject have not caught up with. Since 15 June 2026, offline conversion imports and enhanced conversions for leads have moved to Google’s Data Manager API, and the old Google Ads API route is closed to new adopters, who get an error telling them the account is not allowlisted for the feature. Access was granted based on prior usage, so integrations that were already running keep working while they migrate. In practice this means the manual CSV workflow you will find described in older forum threads is no longer the path for someone starting today. Either your tooling already has a working integration, or your developer builds against the Data Manager API. For most contractors, that argues for using a system that already does this rather than assembling one.
Our guides to GA4 for service businesses and to CallRail cover the call side of this, since most contractor leads arrive by phone and phone leads are where GCLID capture most often breaks.
What I would tell a contractor to do this month
Three things, in this order, whether you run your own advertising or someone runs it for you.
Measure your median lag. Twenty minutes with a CRM export. Median, not mean. Do it per job type if your mix is broad.
Set your evaluation date from that number, not from the help page. One to two of your cycles, not one to two of anyone else’s. If that lands ten weeks out, it lands ten weeks out.
Put the date in the calendar before you launch. This is the part that actually saves campaigns. Write it down, share it with whoever else looks at the account, and agree in advance that nobody makes a verdict before it. The week-three panic is not a discipline problem, it is a planning problem, and a date in a calendar solves it.
Questions this raises
What is a conversion cycle in Google Ads?
It is the time between the click and the conversion you are optimising toward, measured for your own business. If you upload the value of signed contracts, your conversion cycle is click to signed contract. It is not a calendar week and it is not the same for every advertiser, which is why Google’s guidance is expressed in cycles rather than days.
How long should I wait before judging Maximize conversion value?
One to two of your own conversion cycles, per Google’s current guidance, which for a contractor with a five-week close is two to ten weeks rather than a fortnight. Google’s own value-based bidding page adds that it should be “3 weeks or 1-2 conversion cycles, whichever is longer,” and warns that accounts with no conversion data uploaded within seven days of the click may take several months to ramp.
Why did my conversion value change after I turned on value rules?
Because value rules adjust the reported value after the fact. Google’s API documentation states that conversions value and all conversions value reflect modifications made by conversion value rule sets. Compare against the original conversion value metric to see what you uploaded before any adjustment, and segment by rule dimension to see which rule is responsible.
Should contractors use value-based bidding at all?
The common advice in PPC communities is that it only beats target CPA when your sales cycle is short and your volume is decent, and that is broadly fair. The part usually left out is that a long cycle is not automatically disqualifying: it means you optimise toward an earlier event, a qualified lead or a booked estimate with a real value attached, rather than the signed contract. That earlier target is also what Google’s 90-day click-identifier limit forces on you anyway. What genuinely disqualifies you is broken plumbing. If GCLIDs are not captured and outcomes are not uploaded on a schedule, you are asking Smart Bidding to optimise for value it cannot see, and target CPA on a well-tracked lead action will serve you better until that is fixed. Whether value bidding beats target CPA in your specific account is a question only a test in your account can answer.
The bottom line
Google’s edit was not a mistake, and it was not aimed at you. It speeds up feedback for advertisers whose customers convert in a single session, and it is correct for them. But it removed the calendar backstop that was quietly protecting everyone else, and it left the entire recommendation resting on a phrase most contractors have never measured for their own business.
Measure your median lag. Set your evaluation date from it. Put the date in the calendar before you launch, so that nobody panics in week three and switches off a strategy that was still working.
If you want a second pair of eyes on it, bring your Google Ads account and a CRM export of the last twelve months to a call with KLEXA, where paid search for contractors is the whole job. We will compute your median lag, work out your honest evaluation date, and run the original-versus-adjusted diff on your value rules. If it turns out your last campaign was switched off three weeks too early, you will at least know what it cost you.
Sources, all checked 26 August 2026: Google Ads Help pages “About Maximize conversion value bidding” (answer 7684216), “About Smart Bidding using value-based bidding” (15099424), “Changes to target based bid strategies” (17061251), “Duration of the learning period” (13020501) and “Fix discrepancies and errors in offline conversion imports” (13321563); the Google Ads API release notes for v25.1; and Search Engine Roundtable’s report of 24 August 2026. Google changes these pages without notice; this article reflects them as of that date.