Definition
When an advertiser uploads a list of people or organisations, the platform does not take the file at its word. It looks for each record among its own signed-in users, or, for an account list, among its company pages, and adds the ones it finds to an audience. Match rate is the proportion of the upload that made it across.
Matched records ÷ Records uploaded × 100(%)
- Records uploaded is the number of rows in the file after duplicates and blank rows are removed. A row is one person or one company, however many identifiers it carries; counting identifiers instead of rows makes a list look bigger than it is.
- Matched records is the number of those rows the platform tied to a user or company page. Where the platform shows only an estimated audience size, that estimate stands in for the count, with the caveat that it is rounded and may be capped at a floor label for small lists.
Meta documents that a customer list is hashed before it leaves the advertiser's environment and compared with Meta's own hashed identifiers, with matches added to the Custom AudienceAn ad audience built from the advertiser's own sources: customer lists, site visitors or engagers.. The custom audiences guide notes that the result is shown as an estimated size rather than a precise match figure. LinkedIn matches contact lists to members and company lists to company pages, and Google matches hashed customer data to signed-in accounts without publishing match rates at all.
Person lists and company lists
A person list asks whether an email address or phone number belongs to someone the platform knows; a company list asks whether a row identifies a company page. The second is the basis of ABMA B2B approach that targets a chosen list of companies with coordinated marketing and sales. on LinkedIn, and it fails in its own way: subsidiaries, trading names and common words get confused when a row carries only a company name.
How to interpret it
Higher is better, because every unmatched row is a customer who cannot be excluded, a prospect who cannot be reached, or a conversion that cannot be credited. But the number needs context before it means anything:
- By identifier. A list of work email addresses usually matches poorly on consumer and professional platforms alike, because people register with personal addresses. The LinkedIn targeting and lead gen forms guide gives exactly that reason for contact lists matching well below the number of rows uploaded.
- By list source. Buyers who checked out recently with a personal email and phone number match far better than a trade-show list from years ago. Compare uploads of the same kind of list, not a customer file with a prospect file.
- By platform. Each platform holds different people and accepts different identifiers, so a gap between platforms describes their user bases, not a fault in the upload.
- By use. A suppression or exclusion list that matches poorly leaves existing customers seeing acquisition ads; a seed list that matches poorly gives a lookalike model too little to learn from. The same rate can be fine for one job and too low for another.
When an account list matches too few companies to run, the diagnosis page for a low Matched Audience match rate separates faults in the file from the narrowing that role filters add afterwards.
An accounting software firm rebuilds its exclusion list
Tallowbeck Accounting Software wanted to stop showing trial offers to its own customers. It exported 8,000 trial users and customers from its CRM, carrying only the work email address each had signed up with, and the platform's estimated audience came back at about 1,600, a match rate near 20%. Most addresses were company domains the people had never used to register on the platform. The team added phone numbers for the 5,200 records that had one, first and last names, and country, then trimmed and lower-cased every email before hashing. The re-upload of the same 8,000 rows matched about 3,900, close to 49%. That was enough for the exclusion audience to remove most existing customers from the trial campaign, and the team now refreshes the file monthly from the same export.
What moves it
- The identifiers supplied. Each extra identifier gives the platform another way to find the person: email, phone number, name and postcode or country together match far more rows than email alone. For company lists, the website domain, the company page URL and the headquarters country do the same work as a phone number does for a person.
- Formatting before hashing. Hashing is deterministic, so an email with a trailing space or a capital letter produces a different hash from the platform's copy and never matches. The enhanced conversions guide sets out the normalisation Google documents and warns that a wrong format silently produces zero matches.
- The age of the list. People change jobs, email addresses and phone numbers. An old list decays even if nothing in the file changes.
- Where the data came from. Data the business collected directly from its own customers, its First-Party DataData a business collects directly from its own customers, visitors and users., carries the identifiers people actually use; bought or scraped lists rarely do, and platform terms ask the advertiser to declare the data's origin in any case.
- Processing time. A freshly uploaded list starts unmatched and fills over hours. The LinkedIn Ads launch QA checklist makes "matched lists have finished matching" a gate before launch, because an exclusion that looks applied can be removing almost nobody.
What good looks like
There is no reliable benchmark for match rate, and the site does not cite one. Platforms do not publish comparable figures, several show only estimates, and the rate depends so heavily on the identifiers and the audience that a number from another business says almost nothing about yours. The useful comparisons are internal: this upload against the previous upload of the same list, the same list with and without additional identifiers, and the matched size against the size the job needs. An exclusion list should match enough to remove most of the people it is meant to remove; a seed list should clear the platform's documented minimum with room to spare. A fall between uploads of the same export means the export or the formatting changed.
Related metrics
The matched size is the starting size of any Custom AudienceAn ad audience built from the advertiser's own sources: customer lists, site visitors or engagers. built from a list, before role or location filters narrow it further. Enhanced ConversionsA Google Ads feature that sends hashed first-party customer data to improve conversion matching. rely on the same hashing and matching to tie conversions back to ad interactions, so the formatting rules that raise a list's match rate also raise the share of conversions the feature can recover. The quality of a business's First-Party DataData a business collects directly from its own customers, visitors and users. sets the ceiling on both. In ABMA B2B approach that targets a chosen list of companies with coordinated marketing and sales., the rate of the company list decides how many target accounts can be reached at all; the guide to account-based marketing on LinkedIn covers how to build and tier the lists that go into it. Where match rates for bought third-party segments are the question, the comparison of CDPs and DMPs explains why renewing a data platform without testing them is a common waste.
Common mistakes
- Reading rows uploaded as audience size, and planning reach or exclusions around people who were never matched.
- Uploading work emails alone to a platform where people register with personal addresses.
- Hashing unnormalised data, so spaces and capital letters turn real customers into non-matches.
- Launching before matching finishes, so an exclusion or seed list does almost nothing in the first days.
- Comparing rates across platforms or list types as if they measured the same thing.
- Uploading once and forgetting, so the list ages, the match rate drifts down and nobody notices.
