Most local businesses running Facebook ads use one audience: a broad demographic and interest-based targeting set they defined when they set up their first campaign and never revisited. That audience is their best performer on good days and a budget drain on bad ones. The businesses that consistently generate leads from Facebook have a different architecture: three layered custom audiences that together create a pipeline from cold awareness to warm conversion.
Why Custom Audiences Outperform Interest Targeting
The Problem With Interest-Based Targeting
Interest-based targeting — selecting demographics and categories from Facebook's menu of options — is a best-guess approximation of your customer profile. Facebook assigns interest categories based on content engagement, page likes, and behavioral signals. Someone categorized as interested in “home improvement” might be a homeowner actively seeking renovation contractors or a renter who occasionally reads HGTV articles. The signal quality is low, and the audience size is large — which means your ad budget is spread across a wide range of people with varying relevance.
Interest-based audiences also don't improve over time. The same broad category that was available when you set up your account two years ago is the same category today. There's no feedback loop, no learning from your actual customers, no compounding efficiency. The cost per lead stays roughly constant because the input stays constant.
How Custom Audiences Use Your Real Data
Custom audiences are built from your actual business data — the people who have watched your videos, visited your website, or become your customers. These are real signals of real interest, not inferred categories. A person who watched 75% of your window cleaning demonstration video and then visited your estimate page is a demonstrably high-intent prospect. Targeting that person specifically, with messaging that acknowledges their familiarity, converts at a fraction of the cost of targeting a broad interest category that might contain them — mixed with millions of irrelevant users.
Audience 1 — Video View Audiences (Awareness Layer)
Building a Video View Custom Audience
A video view custom audience captures everyone who has watched a specified percentage of a video you've run as an ad or posted organically. In Facebook Ads Manager, navigate to Audiences, create a new Custom Audience, select Video as the source, choose your video, and select a view threshold — 50%, 75%, or 95%. We recommend building separate audiences for each threshold so you can target the more engaged 75%+ viewers with different messaging than the 50%+ viewers.
The video itself can be anything that demonstrates your work: a 30 to 60 second job walkthrough, a before-and-after transformation, a customer testimonial. The production quality matters less than the authenticity. Real job footage from a smartphone consistently outperforms stock imagery or professionally produced brand videos for local service businesses. People evaluating a local contractor want to see real work. Show them real work.
Using View Depth for Qualification
View depth is a qualification signal. Someone who watches 25% of a video probably saw it autoplay in their feed and kept scrolling. Someone who watches 75% made a deliberate choice to watch something that required attention. Treat these two groups differently. The 25% viewers belong in a broad awareness retargeting bucket. The 75%+ viewers belong in a high-intent retargeting bucket with stronger conversion messaging. Segmenting by view depth typically cuts cost-per-lead in half compared to treating all video viewers as the same audience.
Audience 2 — Website Visitor Retargeting (Intent Layer)
Pixel Setup and Event Tracking
The Meta Pixel is a snippet of JavaScript placed on every page of your website. Once installed, it tracks every page visit and can be configured to track specific events — form submissions, phone number clicks, estimate page visits, thank-you page loads. If you don't have the pixel installed on your site, installing it is the single highest-priority action you can take before running any Facebook ads. A website with no pixel is generating traffic that produces zero audience data for future campaigns.
Standard event tracking should capture at minimum: PageView (all pages), ViewContent (service-specific pages), Lead (any form submission), and Contact (phone or email link clicks). These event types allow you to build segmented audiences based on what actions visitors took rather than just that they visited the site.
Segmenting Visitors by Page and Behavior
Not all website visitors are equal. Build separate custom audiences for: all website visitors (last 30 days), estimate or contact page visitors who did not submit (highest intent, did not convert), service-specific page visitors (medium intent, researching), and thank-you page visitors (past leads or customers — exclude from conversion campaigns, include in upsell or referral campaigns). Each of these segments requires different ad messaging. Serving the same “call for a free estimate” ad to someone who just submitted an estimate request as to someone who visited your homepage once is a waste of budget and a poor user experience.
Audience 3 — Customer Lookalikes (Scaling Layer)
Building the Source Audience
A lookalike audience requires a source — a custom audience that Facebook uses to identify characteristics of your best customers and find similar people. The source can be a customer email list, a list of converted leads, a website visitor audience filtered to people who submitted a form, or a video view audience of 95%+ viewers. The higher the quality and specificity of the source, the higher the quality of the resulting lookalike.
For most local service businesses, the highest-quality source is a customer email list. Upload your customer database (minimum 100 records, ideally 500+) as a Custom Audience, then create a 1% Lookalike from that audience targeting your service area. Facebook identifies the common characteristics — age, income range, behavioral signals, content preferences — of your actual customers and finds similar people who haven't encountered your brand yet.
Lookalike Percentage and Scale Trade-offs
Lookalike audiences are built in percentage increments from 1% to 10%, where 1% is the most similar to your source and 10% is the broadest match. For local service businesses, we always start at 1% — the tightest match, the smallest audience, the highest similarity to your actual customers. As you scale budget, you can expand to 2% or 3%. Going beyond 3% typically produces audiences that are too dissimilar from your customer profile to convert efficiently.
Connecting the Three Audiences Into a System
Exclusion Lists That Keep Audiences Clean
Exclusion lists prevent audience overlap and control where budget is spent. Your awareness campaigns targeting cold lookalike audiences should exclude current website visitors and video viewers (who should be in retargeting campaigns). Your retargeting campaigns should exclude past customers (who should be in a separate upsell or referral campaign). Your customer lookalike campaigns should exclude all existing customers. Without these exclusions, the same person appears in multiple campaigns simultaneously — you bid against your own campaigns for the same impression.
Refreshing and Maintaining Audience Quality
Custom audiences are not set-and-forget. Video view audiences should be set to rolling 30 to 90 day windows so they capture recent engagement, not engagement from two years ago. Website visitor audiences should be refreshed to 30-day windows unless your conversion cycle is longer. Customer lookalike source audiences should be updated whenever you add a significant number of new customers — quarterly at minimum, monthly if you have high booking volume. An audience built from stale data produces stale results. Maintain these audiences as actively as you maintain any other business system.