How to build a Facebook custom audience from your FE CRM

Every FE agent who has run Facebook ads for more than six months has a list sitting in their CRM that they're not using: their issued policyholders. These are the people who saw the ad, submitted a lead, answered the phone, and bought. Meta's algorithm doesn't know who they are. You do.

Uploading that list to Meta as a custom audience lets you do two things. First, build a lookalike audience modeled on real converters instead of Meta's generic guess at what a final expense buyer looks like. Second, exclude your existing customers from prospecting so you're not paying to advertise to people who already bought. This post walks through the full process: which contacts to export, how to format them, what to do inside Ads Manager, and what to expect in the first 30 days.

What a Meta custom audience is (and why it beats cold targeting)

A Meta custom audience built from a customer list is a segment Meta creates by matching the contact data you upload (emails, phone numbers, names) against its own user database. Once matched, you can target those people directly, exclude them from other campaigns, or use them as the seed for a lookalike audience that finds people who look like them.

Cold interest targeting on Meta is a blunt tool. Broad age targeting (50 to 75), income signals, and interest clusters like funeral planning or AARP memberships find people who fit the FE demographic in the abstract. A custom audience built from your actual policyholders tells Meta exactly who completed the full funnel for you specifically. A 1 percent lookalike from that seed will usually run a lower CPL than pure interest targeting, because the model is trained on real buyers rather than Meta's approximation of potential buyers.

The three lists worth building from your FE CRM

Not all CRM contacts are equally useful as audience seeds. Before exporting anything, decide what signal you want Meta to find more of.

Issued policyholders

This is the most valuable list you have. These are people who completed the entire funnel: saw an ad, submitted a lead, talked to you, and bought. Export every contact tagged as issued or active in your CRM. Include anyone who issued with you in the last two to three years. The bigger this list, the better the lookalike.

High-intent leads who didn't buy

People who answered the phone, ran through a presentation, and said no or not yet cleared the intent bar even if they didn't buy. As a standalone retargeting audience, this list is useful. Mixed into a lookalike seed with your policyholders, they dilute the signal. Keep them in a separate audience and use them separately.

Lapsed or cancelled policies

Upload cancelled policyholders as an exclusion audience on your prospecting campaigns if nothing else. You don't want to pay to re-prospect someone who already went through your funnel and churned. Some agents also run a dedicated win-back campaign to this list with a re-engagement angle.

How to export from the most common FE CRMs

The goal is a CSV with email address or phone number (or both) for each contact. The exact steps vary by platform.

GoHighLevel

In GHL, go to Contacts and apply a filter for your issued-policyholder tag. Click Export in the top right and choose CSV. The export includes whatever fields your contacts have populated. You want at minimum: email, phone, first name, last name. If you capture zip code in GHL, include that column too. Meta uses zip as an additional matching signal.

AgencyZoom

Navigate to the Clients section in AgencyZoom and filter by policy status (Active or Issued). Use the export function to download to CSV. If the default export doesn't include email, check the column settings before exporting. Phone and email are both useful and you can include both in the same file.

Radius Bob

Radius Bob has a Leads export under Reports. Filter by status (Sold or Issued) and your preferred date range, then export to CSV. Phone numbers in Radius Bob may be formatted with dashes or parentheses. Leave them as-is. Meta strips formatting during matching and doesn't need clean numbers.

Any other CRM or spreadsheet

If your CRM doesn't have a clean export option, copy the relevant contacts into a Google Sheet, add column headers (email, phone, first_name, last_name, zip), and download as CSV. Meta doesn't care about column order as long as you map them correctly during upload. One row per contact, no duplicate header rows.

Uploading your list to Meta step by step

The upload happens in Audiences inside Meta Ads Manager. The process takes about ten minutes. Meta then takes 24 to 48 hours to match the list and populate the audience.

1. Open Meta Ads Manager, click the grid icon in the top left, and select Audiences.
2. Click Create audience and choose Custom audience.
3. On the source screen, select Customer list.
4. Download Meta's CSV template if this is your first time, or click Upload file and select your exported CSV directly.
5. On the column-mapping screen, match each column in your file to the corresponding Meta identifier. Map email to Email, phone to Phone, first name to First Name, and so on. If your CSV has extra columns like a lead ID or policy number, map them to “Do not upload.”
6. Review the preview, then click Upload & create. Meta processes the file and returns a match rate once it's done.
7. Name the audience clearly: something like “FE Issued Policyholders – Oct 2026” so you can find it when building the lookalike and know which version is current.

A match rate of 40 to 65 percent for email is normal. Phone numbers often land higher, in the 50 to 75 percent range for the FE demographic, because people in this age bracket are more likely to have their mobile number tied to their Facebook account than a formal email address. If your match rate is under 30 percent, check that phone numbers are formatted with country code (+1) or that the email addresses are the same ones contacts use for Facebook.

Building the lookalike audience

Once your customer list has at least 100 matched users, Meta will let you build a lookalike. With 500 to 1,000 matched users the lookalike is meaningfully stronger. With exactly 100, it will work but produces a looser match.

1. In Audiences, find your uploaded customer list, click the three-dot menu, and select Create lookalike.
2. Set the source to your uploaded policyholder audience.
3. Choose the location: United States, or your specific state if you only work one market.
4. Set the audience size. A 1 percent lookalike is the tightest match, closest to your actual customers. A 2 to 3 percent lookalike is broader, produces more volume, and usually runs at higher CPL. Start with 1 percent.
5. Click Create audience. Meta takes one to three days to populate the lookalike.

You can create multiple sizes at once. Testing 1%, 1-2%, and 2-3% in separate ad sets (equal budgets, ABO) shows which size delivers the best CPL for your specific book. If you haven't read the CBO vs ABO guide for FE, that's the right structure for this kind of head-to-head test.

Where to use your custom audiences in campaigns

A CRM-based audience has two distinct uses in your campaign structure: as a targeting seed for prospecting and as an exclusion to clean up your cold campaigns. Set up both.

Lookalike for prospecting

Add your 1 percent lookalike as the audience on a new prospecting ad set. Layer age (50 to 75) to keep it inside the FE bracket. Do not stack interests on top of a lookalike. Meta already knows who to find based on the seed. Adding interests narrows the pool without improving relevance and limits reach unnecessarily. Geographic exclusions (specific states or counties you don't work) are the only useful overlay. See the FE audience targeting guide for how a lookalike fits into a full campaign structure alongside interest and broad audiences.

Exclusion for existing customers

Add your issued-policyholder list as an exclusion audience on every cold prospecting ad set. There is no reason to pay to advertise to someone who already bought from you. This also gives Meta's algorithm a cleaner signal about what kind of person you don't want to target, which can improve lead quality on the prospecting side.

Retargeting your high-intent leads

Upload your “answered the phone but didn't buy” list as a separate custom audience and run a retargeting ad set to it with a distinct budget. The creative for this audience can be more direct than a cold prospect ad. These people already know what final expense insurance is. A simple re-engagement hook works better than re-running the full education angle at them.

Keeping your audiences fresh

Custom audiences built from customer lists do not auto-update when you add contacts to your CRM. You have to re-upload manually. For most FE agents, once a month is the right cadence. Pick a date (the first of the month works), export the updated list, and upload a new version. Name each version with the date.

Meta does not automatically replace the old audience when you upload a new version. Archive or delete the old one each time, or you will end up with a dozen audiences named “FE Issued Policyholders” with no clear indication which is current. Clean up old versions as part of the monthly refresh.

An audience not refreshed in 90 or more days will still function technically, but the lookalike built from it drifts from your current customer profile. If you're actively issuing policies, your most recent policyholders are the most predictive signal. Keep the seed current.

What to expect in the first 30 days

A lookalike from a real policyholder list is not an instant fix. It takes 7 to 14 days of spending at sufficient budget before Meta has enough conversion data to optimize properly. The first several days are often noisy. Don't pause a lookalike ad set in the first week because the CPL looks high.

Meta targets 50 conversion events per ad set per week to exit the learning phase. At a $35 CPL, that's roughly $250 a day to clear learning within a week. At $100 a day it may take two to three weeks. Check the budget guide for FE agents if you're working out what spend level runs a proper test.

After 30 days with consistent spend, compare the lookalike ad set's CPL against your broad age-targeted ad set running the same creative. If the lookalike is within $5 to $10 of broad targeting on CPL, it's working as expected. If it's running $15 or more higher with no improvement in lead quality, the seed list may be too small or too mixed. Check the match rate on your upload and consider adding more contacts to the seed before rebuilding the lookalike.

Common questions

Does Meta share my customer data with anyone? Meta uses uploaded contact data solely for matching. According to Meta's custom audiences terms, the data is hashed before matching and the hashed version is discarded after the audience is built. Meta does not sell or share customer list data. Review the terms inside your Business Manager under Audiences for the full language.

Can I use a CRM audience for a campaign if my account is in the learning phase? Yes. You can add a lookalike audience to an ad set regardless of where the account is in its learning history. The lookalike itself takes one to three days to populate after you create it, but that is separate from the campaign learning phase.

What if my CRM has duplicate contacts? Meta de-duplicates during the upload process. A contact appearing twice in your CSV with the same email address counts as one matched user. You don't need to clean duplicates before uploading, though a cleaner file uploads faster and produces a cleaner match report.

If you want us to set this up for you

Apply on the FexAds homepage. Part of our setup process is configuring your audiences correctly before the first dollar is spent, including any CRM lists you want uploaded as custom audiences or exclusions. We build this into the campaign structure on day one.