Skip to main content
MethodMarketingPaid advertisingAdvanced

Use customer lists, custom audiences and lookalikes

A plan for using your own customer data in Google Customer Match and Meta custom and lookalike audiences: eligibility, consent, per-platform file preparation, list sizes, opt-out removal, and what the platforms treat as a suggestion versus a hard limit.

Use this task 5 quick questions, then open it in ChatGPT, Claude or Gemini.

This is a practical, copy-paste process for you have a customer or lead list and want to target, exclude or find similar people. Use it when you want a clear structure, then add your own details.

Verified on 2026-10-2212 sources cited

Quick start

1. Only use customer data you collected yourself.
2. For people in Europe, the UK or Switzerland, upload only those who agreed to advertising use. A newsletter sign-up is not enough.
3. Give each list one job: keep customers out, win them back, or find similar people.
4. Make a separate clean file for each platform: Google wants phone numbers with a +, Meta wants digits only.
5. On Meta, customer lists and lookalikes are hints unless you choose "Further limit the reach of your ads". Exclusions are real limits, but only for the people Meta can match.
6. Small Google accounts can use lists to exclude people, but not to target them.
7. Every month, add new customers and remove anyone who unsubscribed.

Before you start

Full method

How to use this method

  1. 1. Copy or downloadCopy the prompt below, or download the method file and attach it to your chat.
  2. 2. Paste into any AIChatGPT, Claude, Gemini or Copilot. Replace each [PLACEHOLDER] with your details, or let the AI ask you.
  3. 3. Add your materialAttach your own files, data or brief so the AI works from your facts, not guesses.

The prompt

Paste it as-is. The AI will ask for anything still in [BRACKETS].

Download .md
# Use customer lists, custom audiences and lookalikes
Method by Deepstory Research: deepstoryresearch.com/method/custom-audiences-and-customer-match

You are a paid-media data lead who sets up first-party audiences under GDPR and platform policies. Follow the method below step by step. Goal: A plan for using your own customer data in Google Customer Match and Meta custom and lookalike audiences: eligibility, consent, per-platform file preparation, list sizes, opt-out removal, and what the platforms treat as a suggestion versus a hard limit.

## My inputs
- Customer data you have: [CUSTOMER_DATA_YOU_HAVE]  (example: 4,200 emails and phones from our CRM, across DE, FR, NL)
- How it was collected and consent: [HOW_IT_WAS_COLLECTED_AND_CONSENT]  (example: checkout option for personalised advertising, privacy notice names Google and Meta)
- Goal: [GOAL]  (example: exclude buyers, find lookalikes)
- Platforms: [PLATFORMS]  (example: Google Ads and Meta)
- Where customers live: [WHERE_CUSTOMERS_LIVE]  (example: Germany, France, Netherlands)

Before you start: if any input above is still in [BRACKETS] or missing, ask me for it in one short numbered list and wait. Do not guess. If I attached files (data, briefs, Deepstory evidence files), treat them as the primary source.

## Steps to follow
1. Confirm the data is first-party and that recorded consent and the privacy notice allow advertising use, especially for EEA, UK and Swiss people.
2. Give each list one job: exclude, re-engage, seed lookalikes, or observe and bid.
3. Check platform eligibility, regulated categories and list-size needs for each use.
4. Prepare a separate file per platform: normalise the way each platform requires, then hash.
5. Choose the upload route on each platform and set consent fields for EEA records.
6. Explain which settings are suggestions and which are hard limits under Advantage+ and Performance Max.
7. Set a refresh schedule that also removes opt-outs and deletion requests.
8. List the checks for processing, match rate and how far exclusions really reach.

## Deliver the output in this structure
- Data and consent check
- List purpose table
- Eligibility and regulated-category check
- File preparation spec per platform
- Upload route and consent fields
- Suggestion vs control table
- Refresh and opt-out schedule
- Post-upload checks

## Quality rules
- Never upload data the user did not collect directly, or data on under-13s.
- A newsletter opt-in alone is not consent for advertising use; say so.
- Mark audience-size targets that are not platform minimums as rules of thumb.
- Label every number as FACT (with source and date), ESTIMATE (show the calculation) or ASSUMPTION.
- If a Deepstory evidence file is attached, use its figures as the reference values and cite its source references; tag claims that are not from it as [UNVERIFIED].
- End with: the 3 assumptions that matter most, what would change the answer, and the single best next step.

What you need to give the AI

InputPlaceholderExample
Customer data you have[CUSTOMER_DATA_YOU_HAVE]4,200 emails and phones from our CRM, across DE, FR, NL
How it was collected and consent[HOW_IT_WAS_COLLECTED_AND_CONSENT]checkout option for personalised advertising, privacy notice names Google and Meta
Goal[GOAL]exclude buyers, find lookalikes
Platforms[PLATFORMS]Google Ads and Meta
Where customers live[WHERE_CUSTOMERS_LIVE]Germany, France, Netherlands

Use it when

  • You have a customer or lead list and want to target, exclude or find similar people.
  • You need to exclude existing customers from acquisition ads.
  • Your audience uploads are failing or matching poorly.

Best with: A chat AI for the plan, using a small fake sample of the file. Do the upload yourself, or through your CRM connector after approving it.

What you get

  • Data and consent check
  • List purpose table
  • Eligibility and regulated-category check
  • File preparation spec per platform
  • Upload route and consent fields
  • Suggestion vs control table
  • Refresh and opt-out schedule
  • Post-upload checks

The AI will ask you first

Answer these before the AI advises you; each one changes what it recommends.

  1. 1

    Which privacy laws apply where your customers live (for example GDPR in Europe, the UAE PDPL, India's DPDP Act), and is any of the data health-related?

    Consent rules outside Europe still apply, and health data is usually treated as sensitive. This task is not legal advice.

    If unknown: Ask the user to confirm with their privacy adviser before any upload; never upload patient or health data.

  2. 2

    Did you collect this data directly, and does your privacy notice say you share it with advertising partners?

    Google Customer Match only allows first-party data with a privacy policy that discloses sharing with service providers.

    If unknown: Do not upload until the user confirms. Draft the privacy-notice wording for their adviser to check.

  3. 3

    For people in the EEA, UK or Switzerland, is consent for advertising use recorded per person (not just a newsletter opt-in)?

    Google does not process EEA records without consent, and EU consent rules apply on both platforms.

    If unknown: Treat the list as containing EEA people; upload only records with recorded advertising consent, and confirm with the privacy lead.

  4. 4

    Which identifiers do you have (email, phone, name and postcode, app ID), and is it B2B or B2C?

    More identifiers raise the match rate; work emails match poorly, so phone numbers matter more for B2B.

    If unknown: Plan for email and phone.

  5. 5

    What is each list for: exclusion, re-engagement, lookalike seed, or observation and bidding?

    Each use has different eligibility rules and size needs.

    If unknown: Start with excluding existing customers.

  6. 6

    How old is the Google Ads account, its lifetime spend, and its policy and payment history?

    Google needs a good compliance and payment history for any use, and 90 days plus more than USD 50,000 lifetime spend for targeting and bid adjustments.

    If unknown: Assume observation and exclusion only.

  7. 7

    Do the ads touch health, finance, housing, jobs or other sensitive topics?

    Google sensitive categories cannot use advertiser-curated audiences, Meta Special Ad Categories remove lookalikes, and Meta blocks audiences suggesting health conditions or financial status.

    If unknown: Ask before building any audience.

  8. 8

    Who in the business can accept Meta's Custom Audience Terms, and which campaign types are live (Performance Max, Demand Gen, Advantage+ sales)?

    Uploads fail until the terms are accepted by an authorised person, and each campaign type uses lists differently.

    If unknown: Ask the business owner to accept the terms themselves; plan for Search and Advantage+ sales.

  9. 9

    Which CRM or shop platform holds the data, and is there an unsubscribe or deletion list?

    Connectors make syncing easier, and opt-outs must be removed from audiences on every refresh.

    If unknown: Plan a manual monthly export with opt-outs removed.

The procedure, step by step

Read this before you run it, so you can check the AI actually followed the method.

  1. 1

    Check data and consent

    First-party only. Privacy notice names ad partners. Recorded advertising consent for EEA, UK and Swiss records; a newsletter opt-in alone is not enough. Confirm with the privacy lead.

  2. 2

    Give each list one job

    Exclude buyers, re-engage lapsed buyers, seed lookalikes, or observe for bidding.

  3. 3

    Check eligibility

    Google: good history for any use; 90 days and USD 50,000 lifetime spend for targeting. Meta: lookalike seeds of 100 or more matched people. Regulated categories: no lookalikes on Meta Special Ad Categories; no advertiser-curated audiences in Google sensitive categories.

  4. 4

    Prepare files

    Remove duplicates, test records and opt-outs. Normalise per platform (phones with + for Google, digits only for Meta). Hash with SHA-256, or upload unhashed through the interface.

  5. 5

    Upload

    Google: Audience manager > Customer list, or Data Manager, with consent fields for EEA records. Meta: Audiences > Create audience > Custom audience > Customer list.

  6. 6

    Wait for processing

    Do not launch campaigns that depend on an exclusion until the list shows as ready with a size.

  7. 7

    Attach to campaigns

    Exclusions as controls. Inclusions as suggestions under Advantage+ or as Performance Max audience signals, after confirming the account may use the list that way.

  8. 8

    Refresh and remove

    At least monthly (rule of thumb): add new consented customers and remove anyone who unsubscribed, withdrew consent or asked to be deleted. Never only append.

Platform settings(show)
PlatformPathSettingNote
Google AdsTools > Shared library > Audience manager > Segments > Customer listCustomer Match list upload or Data Manager connectionGoogle recommends Data Manager for Customer Match workflows.
Google Data Manager APIAudience member ingestion with consent fieldsCustomer Match uploads with ad_user_data and ad_personalization consentReplaces Google Ads API uploads since 1 April 2026 for developers without recent activity.
Meta Ads ManagerAudiences > Create audience > Custom audience > Customer listCustomer list custom audience, or exclusion-only customer list
Meta Ads ManagerAd set > Audience > Controls > Show more controlsCustom audience exclusion (always a control)
Meta Ads ManagerAd set > Audience > Add suggestionsCustom audience inclusion as a suggestion; Advantage+ lookalike

Worked example

The figures below are illustrative. Use your own numbers and facts.

Scenario: An EEA D2C brand selling in Germany, France and the Netherlands with 4,200 customer emails and phones.

Input: Checkout has a separate, unticked-by-default option to use the customer's details for personalised advertising on Google and Meta; 2,900 customers ticked it. The privacy notice names both. Goals: exclude buyers from prospecting, and seed a lookalike. Google account: 14 months old, EUR 30,000 lifetime spend, clean history.

Output excerpt: Upload only the 2,900 records with recorded advertising consent, with ad_user_data and ad_personalization granted on the Google upload. Google: lifetime spend is under USD 50,000, so use the list for exclusion and observation only. Meta: a separate file with phones as digits only; use it as an exclusion in prospecting ad sets, knowing unmatched customers will still see ads; build one 1% lookalike covering DE, FR and NL. Expect Advantage+ lookalike on by default. Monthly refresh from the CRM, removing anyone who unsubscribed or withdrew consent.

Check your result

Before you trust the AI's output, confirm each of these.

  • ☐ Every uploaded record is first-party and has recorded advertising consent where required.
  • ☐ Each list has one stated job.
  • ☐ Google eligibility (exclusion only vs targeting) and regulated-category limits are stated.
  • ☐ Files were normalised per platform before hashing.
  • ☐ EEA uploads carry consent fields.
  • ☐ A refresh schedule exists that removes opt-outs and fits the 540-day Google window.
What you'll get(show template)
## First-party audience plan: [BUSINESS]

### Data and consent
Source: [how collected] | Privacy notice names ad partners: [yes/no] | Advertising consent recorded: [n of total] | Confirmed by: [privacy lead]

### Lists
List | Job | Size | Platforms | Exclusion or suggestion

### Eligibility
Google: [exclusion and observation only / targeting allowed], because [history, spend]
Regulated categories: [none / ...]
Meta: lookalike seed size: [n matched], countries: [..]

### File preparation
Google: emails [rules], phones E.164 with + | Meta: emails [rules], phones digits only | Hash: SHA-256

### Upload route
Google: [Data Manager / Audience manager], consent fields: [set] | Meta: [Customer list custom audience]

### Suggestions vs controls
Setting | Google | Meta

### Refresh
[Cadence], opt-outs removed: [yes], next date: [date]

Quality checks built into the prompt

  • • Never upload data the user did not collect directly, or data on under-13s.
  • • A newsletter opt-in alone is not consent for advertising use; say so.
  • • Mark audience-size targets that are not platform minimums as rules of thumb.
  • • Every number labelled FACT, ESTIMATE or ASSUMPTION; claims outside attached evidence tagged [UNVERIFIED].

Decision rules

If/then rules the AI should apply. Heuristic rules are a practitioner's rule of thumb, not a platform rule.

IfThenType
The data was not collected directly by the business, or includes people under 13.Do not upload it to Google Customer Match.Platform rule
You use Google Customer Match at all.The account (or its manager account) needs a good policy-compliance and payment history. Targeting and bid adjustments also need 90 days of history and more than USD 50,000 lifetime spend; otherwise use observation and exclusion only.Platform rule
A Google upload includes EEA, UK or Swiss people.Send consent for ad_user_data and ad_personalization with the upload, and leave out records without consent. Google does not process EEA data without consent for Customer Match.Platform rule
You build a new automated Customer Match upload.Use Google Data Manager or the Data Manager API. Since 1 April 2026 Google Ads API Customer Match uploads fail unless there was Customer Match activity between 1 October 2025 and 31 March 2026.Platform rule
A Customer Match list has not been refreshed for a long time.Refresh it. Membership lasts at most 540 days, and at least 100 members must be added or updated in that window to keep the list eligible.Platform rule
You prepare a hashed file for Google.Trim and lowercase emails; for gmail.com and googlemail.com remove dots and plus-suffixes; phones in E.164 with the plus sign (+4915112345678). Hash with SHA-256.Platform rule
You prepare a hashed file for Meta.Trim and lowercase emails (no gmail dot removal); phones as digits only with the country code, no plus sign, symbols or leading zero (4915112345678). Hash with SHA-256. Keep separate files per platform, or upload unhashed through each interface and let it hash.Platform rule
You add a custom audience or lookalike to a Meta Advantage+ audience.Treat it as a suggestion; Meta can deliver beyond it. Only locations, minimum age, languages and custom audience exclusions are hard limits.Platform rule
You use a lookalike in a new Meta ad set.Expect Advantage+ lookalike to be on by default, which lets delivery go beyond the chosen percentage.Platform rule
You build a Meta lookalike.Use a seed of at least 100 matched people from at least one country. One lookalike can cover several countries. The API size range is 1% to 20%; in Ads Manager use the sizes offered.Platform rule
The ads are for housing, employment or financial products on Meta (US-based advertiser, or reaching the US, Canada or Europe).Declare the Special Ad Category: lookalikes are not available, but custom audience inclusions and exclusions still are, within the category's audience limits.Platform rule
The ads fall in a Google sensitive interest category.Do not use Customer Match or other advertiser-curated audiences. For US and Canada housing, employment and consumer finance ads, Customer Match is allowed, but age, gender, parental status, marital status and ZIP targeting are not.Platform rule
A Meta win-back campaign must reach only people on the list.Use "Further limit the reach of your ads" to make the custom audience inclusion a control; this may turn Advantage+ audience off, which is expected.Platform rule
You need existing customers kept out of acquisition ads on Meta.Use the customer list as a custom audience exclusion; exclusions are always controls. They only remove people Meta could match, so some customers will still see the ads.Platform rule
A Meta list is under about 1,000 matched people.Use it for exclusion or as a lookalike seed, not as the only targeting. On Google, check the list size per network in Audience manager.Threshold: about 1,000 matched people (rule of thumb)Heuristic

Failure modes

Common mistakes with this method, how to spot them, and how to fix them.

MistakeSymptomFix
Uploading a purchased or event-partner list.Policy breach risk and possible loss of Customer Match access.Upload only data you collected directly, with a privacy notice that covers ad partners.
Treating a newsletter opt-in as consent for advertising.GDPR risk; Google drops EEA records without consent.Collect and record advertising consent separately.
One hashed file for both platforms.Phone matches lost on one platform.Normalise and hash per platform.
Expecting a Meta lookalike or custom audience to limit who sees the ad.Delivery to people outside the list.Under Advantage+ these are suggestions; use exclusions for hard limits.
Expecting an exclusion to remove every customer.Existing customers still appear in "new customer" results.Compare matched size with list size and add more identifiers.
Only ever appending to lists.People who unsubscribed keep being targeted.Remove opt-outs and deletion requests on every refresh.
Building a new Google Ads API Customer Match job.Requests fail with CUSTOMER_NOT_ALLOWLISTED_FOR_THIS_FEATURE.Use the Data Manager API.
Naming or building audiences around a health condition or income.Meta flags the audience and blocks it.Build audiences around behaviour with the business, such as purchase recency.

Agent mode: if an AI agent runs this unattended

May do without asking

  • • Write the file preparation rules and check a small fake sample for formatting problems.
  • • Draft privacy-notice wording for the user to check with their adviser.
  • • Explain which audience settings are suggestions and which are controls.

Must stop before (the user decides and does these)

  • • Uploading any customer list or personal data to Google or Meta.
  • • Setting up or approving a CRM connector or automatic sync, which uploads personal data on an ongoing basis.
  • • Downloading or exporting customer lists from ad platforms or the CRM into chat.
  • • Attaching an audience to, or removing an exclusion from, a live campaign in a way that changes who you spend money on.
  • • Publishing or enabling any campaign or ad set.
  • • Accepting Custom Audience or Customer Match terms, or changing billing.

Evidence for this method

This method runs on your own material: your files, brief or code. If your task needs market or economic numbers, attach a sourced context file from the evidence library alongside the prompt.

Browse evidence

Run next

Methods that chain well after this one.

Deepstory methods are process guidance for AI tools. Review AI output before relying on it; legal, tax, medical and investment decisions need a qualified professional.