Customer Match now works at 100 records

Customer Match used to need 1,000 matched active users before Google would let you target the list. That single number kept most B2B and high-value service businesses out, because their entire customer file might be 400 people. The floor is now 100.

If you sell something worth thousands of dollars to a few hundred organisations, this is the most useful change in the platform in a while, and almost nobody has acted on it.

Why the old floor excluded the businesses that needed it most

The value of a customer list rises as the list gets smaller and more valuable. A retailer with 200,000 customers gets a modest lift from list-based targeting. An equipment supplier with 300 accounts, each worth six figures a year, gets an enormous one, because those 300 names are the entire market that matters.

The 1,000 floor was calibrated for the first business and locked out the second. At 100, a list of your top accounts, your lapsed customers from the last two years, or the 140 people who requested a quote and never bought becomes a usable audience.

Exclusion is usually worth more than targeting

The reflex is to build a list and target it. For most accounts the bigger win is the opposite.

Upload your existing customers and exclude them from acquisition campaigns. On a long sales cycle with automated bidding, existing customers are the easiest conversions in the auction, which means the bidder finds them, and your cost per acquisition looks excellent while your new business number flatlines. Excluding them forces the spend outward.

Same logic for anyone currently in your sales pipeline. If your team is already working the deal, you do not need to pay to reach them again, and the attribution credit muddies the read on what acquisition is doing.

Match rate is the thing to plan around

Uploading 100 records does not give you an audience of 100. Google matches your hashed data against signed-in accounts, and the matched number is what counts against the threshold.

Match rate depends almost entirely on data quality. Email alone gives you a mediocre rate. Email plus phone plus first name, last name, country and postcode gives you a much better one, because Google can match on any of several combinations. If you are running at the edge of the threshold, adding fields is the fastest fix.

Two practical points. Use personal email addresses where you have them, not just work addresses, since the signed-in Google account is often personal. And clean the file before upload, because trailing spaces, capitalisation inconsistencies and formatted phone numbers all cost you matches.

Consent, plainly

You need permission to use customer data this way, and that obligation sits with you, not Google. Under Australian privacy law your collection notice and privacy policy need to cover the use, and the practical test is whether the customer would be surprised. Someone who bought from you and agreed to marketing is fine. A list scraped from an industry directory is not.

Google's own policy requires that the data was collected directly from the customer with appropriate notice. Do not upload purchased lists. It is against the terms and it produces terrible match rates anyway.

Getting it live

1. 1.Export the segment from your CRM with as many identity fields as you hold. Do not send anything you would not want in a spreadsheet.
2. 2.Hash locally with SHA-256 after normalising, or let Google Ads Data Manager handle the hashing on upload. Data Manager is now the single ingestion path for customer data, so build the connection there rather than through legacy uploads.
3. 3.Set a refresh. A static list decays. Weekly or fortnightly sync from the CRM is the difference between an audience and a snapshot.
4. 4.Start with exclusions on acquisition campaigns and observation-only on everything else. Read the data for a fortnight before you act on it.
5. 5.Then build the targeted use cases: win-back for lapsed accounts, upsell for single-product customers, and high-value seeding.

The limits worth naming

Small audiences produce thin reporting. At 100 to 300 matched users you will not get statistically meaningful performance data at the audience level, and you should not pretend otherwise. Treat these lists as steering inputs and exclusions, not as reportable line items.

Reach will also be small by definition. This is not a volume play. It is a precision play, and the value shows up in who you stopped paying to reach as much as who you started reaching.

Pull your customer file this week and count it. If it is over 100, you already own an audience you have never used.