Customer type labeling and lifecycle goals: Google now classifies your lists
On 17 June 2026 Google started activating automatic customer type labeling on eligible Google Ads accounts, and from 18 August the system began processing data and assigning labels. The subject is conversion-based lists — the audiences Google builds itself from on-site actions. If you never classified them by hand, the platform does it for you, sorting users into a nine-category taxonomy that includes Purchasers, Cart Abandoners, Qualified Leads and Disengaged Customers. Those labels then start feeding Smart Bidding.
The labeling itself is reasonable. The problem is that if your lists were built on different logic than Google’s taxonomy assumes, the labels land wrong and bid strategies quietly optimise toward the wrong people. This article covers how customer types and lifecycle goals actually work, exactly where the logic breaks, how to audit your audiences in an hour, and what to do if the labels have already been applied and performance moved.
What customer types are and why Google wants them
A customer type is a label on an audience answering one question: where in the relationship with your business do these people sit. The top-level split is three groups:
- Existing customers — they have bought;
- New customers — they have not bought yet;
- Other segments — everything else: subscribers, cart abandoners, dormant users.
Underneath that sits a finer nine-category taxonomy describing behaviour more precisely: purchasers, repeat purchasers, cart abandoners, qualified leads, subscribers, registered non-buyers, inactive, disengaged, all visitors.
The labels are not for reporting. They are the foundation of customer lifecycle goals — modes that make the algorithm treat new and existing buyers differently:
| Mode | What it does | When it fits |
|---|---|---|
| New Customer Value Mode | Adds extra value to a conversion from a new customer | The default choice: grow the base without abandoning repeat sales |
| High Value New Customer Mode | Same principle with a more aggressive premium | When base growth outweighs short-term ROAS |
| New Customers Only Mode | Serves primarily to new customers | Product launches, new market entry; otherwise it cuts volume hard |
The logic is straightforward. If the algorithm cannot tell a first-time buyer from someone who would have returned through remarketing anyway, it collects the cheap conversions among existing customers. The report shows a great ROAS while the customer base stands still. The mechanics of the new-customer premium are covered in detail in our article on conversion value and value rules, including the formulas behind the multipliers.
What changed in 2026: customer type labeling became automatic
The two dates that matter: 17 June 2026 — rollout begins on eligible accounts; 18 August 2026 — data processing and auto-labeling start. Before that, classification was optional: an unlabelled list simply did not participate in lifecycle goals. Now unlabelled conversion-based lists receive a label automatically.
One important boundary: this affects auto-generated conversion-based lists only. Files uploaded manually through Customer Match are not auto-labelled — you set the customer type yourself at upload. That is not a reason to relax, because in most accounts remarketing runs on the automatic lists.
The cheapest fix is the one you make before the labels spread across campaigns. Afterwards you are repairing the lists and waiting for strategies to relearn at the same time.
Where the logic breaks
Customer type labeling reads your conversion events. If the events are configured differently from what Google assumes, the labels will be wrong. Four common scenarios:
- One list for everyone. An audience of “anyone who submitted a form” contains both first-time and repeat buyers. Google must assign it to one category, and either choice is half wrong.
- Purchase is not tracked as a purchase. In B2B, the tracked conversion is an enquiry while the deal closes in the CRM a month later. To Google, everyone is a qualified lead and actual customers are invisible.
- Subscription treated as purchase. If a newsletter signup is flagged as a purchase event, the entire subscriber base is classified as existing customers and excluded from acquisition.
- Multi-brand or multi-market accounts. A customer of one brand counts as existing across every campaign in the account, even though they are new to the second brand.
The common denominator is conversion event quality. Clean events, properly split by type, produce correct labels. How to get events in order is covered in our breakdown of micro-conversions and funnel structure.
A one-hour audience audit
Step 1. Inventory the lists
Open Tools → Shared Library → Audience Manager → Segments. Export the list with columns for name, type (conversion-based / tag-based / Customer Match), size for Search and Display, membership duration, and where it is used. The usual discovery is 15–30 segments of which four or five are actually applied anywhere.
Step 2. Mark what each list really is
For every conversion-based list, answer one question: are all the people in this list at the same stage of the relationship? If the answer is no, the list needs splitting. The easiest split is by event: one audience for purchase, one for enquiry, one for cart abandonment.
Step 3. Set customer types manually
For every list used in campaigns, set the customer type explicitly. Manual labels take priority over automatic ones: once you have set it, the system does not overwrite. That is the fastest way to remove the risk entirely.
Step 4. Decide on the ambiguous lists
There is a third path besides “split” and “label”: opting out of conversion-based lists in account settings. That is reasonable if you run entirely on uploaded Customer Match data. But remember it also removes what those lists were giving you — above all the ability to exclude existing customers from acquisition campaigns.
Step 5. Review campaigns using lifecycle goals
For each campaign with a new customer acquisition mode enabled, check three things: which mode is set, which lists are designated as existing customers, and whether an additional new-customer value is defined. A frequent mistake is having the mode on with no existing-customer list specified at all — Google then treats everybody as new and the premium becomes meaningless.
Step 6. Capture the “before” state
Before changing anything, export 28 days of conversions, conversion value, new customer share, CPA and ROAS per campaign. Without that baseline you cannot demonstrate to yourself or a client that the fix worked. Measure the effect afterwards through Google Ads experiments or geo experiments for incrementality.
The nine categories: what each one actually means
The granular taxonomy is not decoration — which category a list lands in determines whether it counts as new, existing or other. Category by category:
| Category | Who lands there | Top level |
|---|---|---|
| Purchasers | Made at least one purchase | Existing |
| Repeat purchasers | Two or more purchases | Existing |
| Cart abandoners | Added to cart, never checked out | Other |
| Qualified leads | Submitted an enquiry, no purchase recorded | Other (often wrong in B2B) |
| Subscribers | Signed up, never bought | Other |
| Registered users | Created an account, never bought | Other |
| Inactive | No recent activity | Other |
| Disengaged customers | Used to buy, stopped | Existing |
| All visitors | Any session | Other |
Pay attention to the “qualified leads” row. It is the single biggest source of trouble in B2B and services: people who have in fact already bought — because the deal closed over the phone — remain leads as far as Google is concerned. Acquisition campaigns keep paying a new-customer premium for people who have been in the database for months. The only cure is sending deal status back into Google, covered below.
The second subtle row is “disengaged customers”. Formally they are existing, so by default they receive no new-customer premium. Economically, though, winning back a lapsed customer often costs more than retaining an active one and sits close to acquisition in value. If reactivation is a distinct objective for you, keep that list separate and manage it through value rules rather than lifecycle modes.
Feeding the system proper customer data
Auto-labeling is a fallback for when you never told Google who is who. Telling it directly is far more reliable. Three sources, in descending order of signal quality:
| Source | Signal quality | What it requires |
|---|---|---|
| Customer Match from CRM with explicit type | High | Hashed emails/phones, a regular export, valid user consent |
| Offline conversion import with deal status | High | Stored click identifier, CRM ↔ Google Ads connection |
| Automatic conversion-based lists | Medium | Correctly configured conversion events |
The practical minimum: once a week, export two lists from the CRM — “purchased in the last 24 months” and “purchased more than once” — and upload them to Customer Match with the customer type set explicitly. Auto-labeling then stops mattering, because manual labels win. Upload mechanics and data requirements are covered in our article on Customer Match and first-party data.
The second layer of reliability is offline conversions. When the deal closes away from the website, only a CRM import tells Google who genuinely became a customer. Without it, auto-labeling is condemned to treat anyone who left a phone number as a buyer. The mechanics are in our guide to offline conversion import, and the question of preserving the click identifier in the piece on GCLID, GBRAID and tracking templates.
A note on consent
Working with customer lists rests on a legal basis: only contacts you are entitled to use may be uploaded to an ad platform. In European traffic it is also a technical matter — consent signals have to reach Google or part of the data is simply rejected. That is covered in our breakdown of Consent Mode v2 and advertising privacy.
Calculating the additional value of a new customer
Lifecycle modes ask you to define how much more a new customer is worth. Guessing is pointless; there is a defensible method.
Premium = (new customer LTV − value of the single purchase), where LTV is measured over a horizon you can actually plan against — usually 12 or 24 months.
Example: an average first order of $90, and a cohort that produces $240 over 24 months. A new customer is therefore worth $150 more than the moment shows. If your conversion passes $90, the new-customer premium is that $150. In practice be conservative and take 60–70% of the calculation to absorb cohort forecasting error.
Three checks without which the number is fiction:
- The cohort must be large enough — at least 200–300 customers, or the average swings wildly.
- Calculate on margin, not revenue, otherwise loss-making acquisition looks profitable on paper.
- Strip one-off spikes: a single enterprise order distorts the entire cohort average.
The cohort LTV methodology is set out in detail in unit economics and LTV in media buying.
Confirming the mode works rather than merely being switched on
A toggle proves nothing. Three checks that give a factual answer:
- Segment the campaign report by customer type. A healthy picture shows both categories present with a new-customer share that is stable week over week. If 100% of conversions fall into one category, the signal is not arriving.
- Check the existing-customer list size. In Audience Manager the list must show a non-empty size for Search. A dash or a “list too small” note means Google physically cannot use it and the mode runs empty.
- Check when the list was last refreshed. A Customer Match file that has not been updated in three months is stale: some of those customers have bought again, others have churned. Aim for weekly refreshes; monthly is the minimum acceptable cadence.
Also read the new-customer share as a trend over 8–12 weeks rather than a snapshot. Swings of 5–7 percentage points are ordinary noise from seasonality and demand mix. A sustained 15-point shift with no campaign changes is almost always a consequence of list labeling moving, not the market.
If the labels already applied and things moved
Symptoms: the new-customer share in reporting shifted without you touching anything; CPA rose in acquisition campaigns; volume dropped in campaigns set to New Customers Only. The repair order:
- Do not switch everything off at once. First establish what changed — segment the report by customer type to see the conversion distribution.
- Apply manual labels to every list used in campaigns. That overrides automatic classification.
- Upload a fresh Customer Match file with an explicit type — the strongest available signal.
- Temporarily reduce mode aggressiveness. If New Customers Only was set, switch to New Customer Value Mode until the data settles.
- Allow 10–14 days. Strategies do not relearn instantly, and daily edits only extend the unstable period.
If behaviour still looks strange after that, stop looking at audiences and check the foundation: conversion completeness, duplicate tags, attribution windows. The order of checks is in the account audit checklist.
A routine so it does not happen again
| Frequency | Action |
|---|---|
| Weekly | Export “purchasers” and “repeat purchasers” from the CRM into Customer Match |
| Monthly | Check customer types on all active segments, plus list sizes and membership durations |
| Quarterly | Recalculate the new-customer premium against fresh cohorts |
| Every launch | Explicitly name the existing-customer list in the campaign’s lifecycle goal settings |
For accounts with a large number of campaigns, checking customer types and lifecycle settings is far quicker through a bulk export than through the web interface — approaches are described in our article on bulk operations in Google Ads Editor.
And a sober framing: lifecycle goals are a refinement on top of an account that already works. They will not fix a campaign with access, limit or structural problems. If that is where you are stuck, sort the foundation first — see how Google Ads agency accounts are structured and what a dedicated PPC team handles — and come back to customer segmentation once the base is stable.
FAQ
Does customer type labeling apply to every account?
Rollout targets eligible accounts, primarily those with conversion-based lists and enough data volume. You can check status in account settings and in Audience Manager, where a customer type field appears on the lists.
Can I opt out of conversion-based lists entirely?
Yes, account settings include an opt-out. But the benefits leave with the risks: existing-customer exclusions and ready-made remarketing segments. Opting out only makes sense if you run purely on Customer Match.
Will the system overwrite a label I set manually?
No. Manual classification takes priority — auto-labeling only applies to lists with no type set. Setting types by hand is therefore the most reliable protection.
Does this affect manually uploaded Customer Match files?
No, auto-labeling leaves them alone. You set the customer type at upload and it stays as you set it.
New Customer Value Mode or New Customers Only?
In the vast majority of cases, the first. New Customers Only cuts volume substantially and is justified at product launch, on entering a new market, or when repeat sales are handled by another channel.
How much data do lifecycle modes need?
A practical benchmark is an existing-customer list of at least a few thousand records plus steady campaign conversion volume. At small scale Google cannot reliably assign a user to a segment and the premium rarely applies.
How do I confirm the mode is actually working?
Segment the report by customer type: you should see conversions split between new and existing. An empty segment, or everything falling into one category, means either the signal is not arriving or the lists are mislabelled.
Is defining the additional new-customer value mandatory?
For New Customer Value Mode, yes — otherwise the algorithm has nothing to differentiate with. Derive it from the gap between cohort LTV and the single purchase, and take 60–70% of the result as a conservative estimate.
What should a multi-brand account do when a customer of one brand is new to another?
Either separate the brands into different accounts, or maintain separate customer lists and name the correct one in each campaign’s goal settings. A single account-wide list will always misclassify in that configuration.
How fast will strategies adjust after labels are fixed?
Plan for 10–14 days. During that window leave budgets, bid targets and audience composition alone — every fresh change restarts the adaptation.
Should existing customers be excluded from acquisition campaigns?
Not always. A hard exclusion saves budget but forfeits repeat sales that would have come through the same search queries. It is usually more effective to keep them in and separate them by value through value rules.
Does any of this affect Performance Max?
Yes — lifecycle goals are one of the few genuine control levers inside PMax. Correctly labelled existing-customer lists matter especially there, since without them the campaign happily harvests conversions from people who would have returned anyway.
