Are Custom Audiences Broken?
I am trying to upload an updated customer list CSV of about 25,000 and I’ve tried it multiple times. Every time it says 0 rows uploaded (there’s definitely not 0 rows). And after it “matches” it says the list is too small and under 1000. I’ve done this many times before, I used the FB template and I’ve mapped it to 9 identifiers.
I wondered if there was something wrong with the CSV formatting so I even uploaded an older customer list from February that had matched 20,000 customers successfully. Now even that one does the same thing. Anyone else having problems with this?
The short answer is:
Your issue where a large, previously successful customer list is suddenly showing zero rows uploaded and an audience size under 1,000 is almost certainly a temporary bug on Meta’s side or a recent, hyper-sensitive change in their data validation process.
It’s highly unlikely that a working list’s formatting broke spontaneously.
While waiting for the upload process to finish – as this can take hours – sometimes resolves the “too small” message, a long-term solution to avoid this exact kind of manual list upload error and achieve superior audience matching is to use a direct, server-to-server connection.
Specifically, the Facebook Conversions API implemented via a server-side tagging solution like Google Tag Manager (GTM) with a simple hosting provider like Stape or Google Cloud Platform is an excellent, reliable, and cheap way to continuously send real-time customer data to Meta, resulting in much higher match rates for your Custom Audiences.
The long answer is:
It’s incredibly frustrating when a process that worked before suddenly fails, especially with a critical function like customer list uploads.
The fact that your old, successful list is now also failing strongly suggests an instability in Meta’s bulk upload processing, potentially an overzealous new data hash or identifier validation check that flags entire files as bad for a minor reason.
When you upload a CSV, Meta first hashes the identifiers (like email, phone, etc.) on your end to protect customer privacy, and then it attempts to match those hashes to its user base.
The “0 rows uploaded” and “too small” message happens when the matching process fails to find a minimum threshold of matched users, which is usually 1,000 for use in campaigns, or sometimes it’s just a visible error that appears before the background processing is actually complete.
Since you’re using nine identifiers, your match rate should be high, meaning you should definitely give the process a full 12 to 24 hours to update before concluding it has permanently failed.
Make sure you also haven’t missed re-accepting the updated Custom Audience Terms of Service, as this sometimes causes silent errors.
While waiting for the manual upload to finish is one short-term fix, the best long-term solution to escape the limitations and bugs of the manual CSV upload is to switch to a real-time, server-side data stream using the Facebook Conversions API (CAPI).
The Conversions API allows your server to send data directly to Meta, completely bypassing browser limitations and the clunky manual upload process.
This is where an implementation using Google Tag Manager (GTM) with a server container comes in as a surprisingly excellent and cheap alternative to expensive enterprise tools.
You would use GTM in a server-side configuration, hosted on a minimal, cost-effective platform like Stape or a basic virtual machine on Google Cloud Platform.
The key here is that instead of relying on the client-side pixel, which can be blocked by ad blockers and ITP features, the server-side setup collects events, like a Purchase or Add to Cart from your website, adds crucial customer identifying information (CI) like hashed email or phone number, and then sends it directly to Meta via the Conversions API.
This direct method results in significantly higher data quality, better deduplication between the pixel and the API, and most importantly for your issue, it enables a constant stream of high-quality customer identifiers to be fed into your Custom Audiences.
You move from manually updating a static list every few weeks to automatically updating your audience with every high-value action taken on your site, leading to more accurate targeting, better Lookalike Audiences, and an end to those frustrating “0 rows uploaded” errors.
The hosting cost for Stape, for instance, is often just a few dollars a month, making it a very cheap, scalable, and robust solution.
About The Author
