What Is Customer Match? How It Works on Google and Meta
Published:
Last Updated:
Category: Ad Operations
Authors: Shusaku Yosa
Published:
Last Updated:
Category: Ad Operations
Authors: Shusaku Yosa
In digital advertising, targeting built on your existing customer data has a large bearing on return on spend. The mechanism at the centre of this is Customer Match: you upload a list of customers you already hold and serve ads directly to those users. Both Google Ads and Meta Ads implement it.
This article covers how Customer Match works, how the implementations differ between Google and Meta, where it applies in practice, and what to watch for on the privacy side when operating it.
Customer Match is a targeting feature in which an advertiser uploads customer data they hold, such as email addresses, phone numbers, or postal addresses, to an advertising platform. The platform matches that data against its own user accounts, allowing ads to be served to a specific set of customers.
Conventional targeting relies on attributes the platform infers, such as interests or browsing behaviour. Customer Match instead starts from first-party data the advertiser holds directly. This allows a far more precise approach to people you already have a relationship with.
The same mechanism goes by different names depending on the platform.
This article uses Customer Match as the general term, and refers to each platform by its own naming when discussing platform-specific behaviour.
The process breaks down into four steps.
Hashing is fundamental to how Customer Match works. It converts the original data into a fixed-length string according to a set rule, and the original data cannot be recovered from the resulting value.
This allows the advertiser and the platform to determine whether two records refer to the same person without either side exchanging raw email addresses. It is the technique that makes privacy protection and matching accuracy compatible.
Normalisation must happen before hashing. Unless you strip leading and trailing whitespace and convert everything to lowercase first, the same address will produce different hash values and fail to match.
The proportion of an uploaded customer list that successfully matches accounts on the platform is known as the match rate. It depends on how current the list is and how accurately the data is formatted, and commonly falls somewhere between 50 and 80 percent.
If the match rate is low, check the following causes.
Customer Match in Google Ads can be used across a wide range of placements, including Google Search, Shopping, YouTube, Gmail, and the Display Network.
In addition to email addresses, Google accepts phone numbers, combinations of name and address, and mobile device IDs. Uploading multiple identifiers at once improves the match rate, since a match on any one of them is sufficient.
Using Customer Match is subject to account requirements, which may include a record of policy compliance and a minimum level of ad spend. It is also restricted for sensitive categories such as housing, employment, and credit-related advertising, as well as adult and alcohol-related content. These requirements change, so check the current official help documentation before implementing.
Delivery only begins once the matched audience reaches a certain size. The threshold varies by placement, and if the audience is too small, the list can be created but will not serve. When designing lists, estimate in advance whether you can reach the volume needed to meet the minimum.
Meta Ads provides customer list uploads as one type of Custom Audience. Delivery covers Facebook, Instagram, Messenger, and Audience Network.
Meta is designed to improve matching accuracy by combining multiple identifiers. Alongside email addresses and phone numbers, attributes such as name, date of birth, gender, and location can serve as supplementary matching keys. Including more columns tends to raise the match rate.
Meta also offers Lookalike Audiences, which take an uploaded customer list as a starting point and extend delivery to new users with similar characteristics. Google Ads has an equivalent in similar segments, but the two differ in how they are generated and in their availability, so they should be tested separately rather than treated as interchangeable.
Meta allows a customer value column to be added to the list. Passing purchase amounts or lifetime value as a number lets you generate lookalike audiences weighted toward high-value customers, which can produce higher-quality delivery than expansion based purely on headcount.
Exclusion is where the effect is most reliable. Specifying a list of customers who have already purchased or subscribed as an exclusion prevents new-acquisition budget from being spent on people you already have. Where remarketing runs alongside acquisition campaigns, this substantially reduces waste from overlapping delivery.
Build a list of purchasers of a specific product and serve ads for related products or higher-tier plans. You might promote lenses and accessories to people who bought a camera body, or a paid plan to users on a free tier. Because delivery goes to an audience that already trusts you, conversion rates tend to be higher.
Segment customers whose last purchase falls beyond a set period and serve ads encouraging a return visit. The key difference from CRM campaigns is that you can reach people who never open your email newsletters.
For subscription models, build lists of users whose usage has declined or whose renewal date is approaching, and serve content on how to get more from the product or where to find support, with the aim of improving retention.
Existing customer lists can be used for bid adjustment rather than exclusion. Bidding more aggressively for high-LTV segments and pulling back on lower-value ones improves return within the same budget.
Importing closed-deal data from physical stores or phone sales as a customer list allows store visits and sales driven by online advertising to be reflected in measurement. For business models that do not complete online, this is valuable for understanding what advertising actually delivers.
Using customer data for ad delivery presupposes compliance with the relevant regulations in each jurisdiction, starting with data protection law. In Japan, this includes obtaining consent where personal data is provided to a third party, and specifying and publishing the purpose of use. State the use of data for advertising purposes in your privacy policy, and upload only data for which the necessary consent has been obtained.
If EU users are included, GDPR requirements come into scope; for California residents, CCPA and CPRA. Work with your legal team or an external specialist to establish operating rules appropriate to your situation.
A customer list is not something you upload once and forget. Design an operating process that reliably removes people who have lapsed or opted out, and refreshes the list on a regular schedule. Most platforms support automated updates via API, which is better than manual operation on both accuracy and workload.
When a user asks to stop receiving advertising, you need a process that removes them promptly. Beyond the platform-side setting, manage a flag in your own customer database so that the exclusion carries through to subsequent uploads.
If the matched audience falls below the threshold, it will not serve. Segmenting too finely leaves individual lists below the minimum and therefore non-functional. Starting at a coarser granularity and subdividing once data accumulates is the more workable approach.
If you are implementing Customer Match for the first time, this sequence keeps the risk down.
Rather than aiming for a complex segmentation design from the outset, starting with exclusion, where the effect is most reliable, and expanding in stages tends to produce better results.
Customer Match lets you apply customer data you already hold directly to ad delivery. Hashing protects privacy while allowing both a precise approach to existing customers and more efficient new-customer acquisition.
Google and Meta differ in the data fields they accept and the expansion features they offer, so it is important to design around the specifics of each. When adopting it, settling the operational side first, meaning regulatory compliance and a process for keeping lists current, is the precondition for sustained results.
A clear explanation of callouts (callout assets, formerly ad extensions), how they differ from sitelinks, the keys to wr...
A carousel is a mechanism that displays multiple images or videos by switching them within one frame. Learn the types fo...
An MCC account is a parent account that centrally manages multiple Google Ads accounts. This article explains what an MC...
