Lookalike Audience Size vs. Cost Per Qualified Lead: Where Broad Lookalikes Start Wasting Budget for Local Service Meta Campaigns

The Problem with Defaulting to 1%

When you build a lookalike audience in Meta, the platform lets you choose a similarity range from 1% to 10% of a target country's population. The conventional wisdom is: tighter = better. A 1% lookalike is the most statistically similar to your seed audience (your customer list, pixel purchasers, or highest-value leads).

That logic isn't wrong — but it's incomplete. For local service businesses operating in a defined geography, defaulting to 1% without ever stress-testing 3%, 5%, or 7% means you're leaving potential reach on the table and making a false assumption: that all the uplift from going narrower survives contact with your actual close rate.

The real metric that matters isn't nominal cost-per-lead (CPL). It's *cost per qualified lead* — or better yet, cost per acquired customer (CAC). A $20 lead is not cheaper than a $35 lead if the $20 lead closes at half the rate.

How Meta's Reach-vs-Specificity Curve Actually Works

Meta's own documentation describes lookalike audiences as a spectrum: at 1%, the platform finds users whose behaviors, demographics, and interest signals most closely mirror your seed. As you expand toward 10%, the algorithm casts a wider statistical net — similarity loosens, but addressable reach grows.

For a national advertiser, 1% of the US adult population is roughly 2–3 million people — plenty of volume. For a local HVAC company targeting a single metro, that same 1% lookalike might resolve to a few thousand in-geography users, causing audience fatigue within weeks and frequency-driven CPL spikes.

This is where local businesses face a structural tension that national brands don't:

• Too narrow: Audience exhausts quickly, frequency climbs, CPL rises from fatigue — not from poor targeting.
• Too broad: Similarity degrades, lead-to-appointment rate drops, and nominal CPL may look better while real CAC quietly inflates.

The goal is finding your local inflection point — the tier where incremental reach still delivers qualified leads before dilution overwhelms the efficiency gain.

A Framework for Modeling the Quality-Cost Curve

Because Meta doesn't publish tier-by-tier CPL benchmarks for local verticals, we use a labeled illustrative model to reason through the curve. Plug in your own numbers to validate.

Illustrative model — local home-services campaign, $3,000/month spend:

| Audience Tier | Est. In-Geo Reach | Nominal CPL (illustrative) | Est. Lead-to-Appt Rate | Cost per Booked Appt | |---|---|---|---|---| | 1% LAL | Low (fatigue risk) | $38 | 55% | $69 | | 3% LAL | Moderate | $32 | 50% | $64 | | 5% LAL | Higher | $27 | 38% | $71 | | 7% LAL | High | $22 | 26% | $85 | | 10% LAL | Very High | $19 | 18% | $106 |

All figures are illustrative estimates built for demonstration. Your vertical, seed quality, and geography will shift every number.

The key insight from this model: nominal CPL falls continuously as you go broader — but cost per booked appointment bottoms out around the 3–5% tier, then climbs steeply. The 10% audience looks cheapest on a dashboard and is actually the most expensive campaign when you account for downstream qualification.

This is the same principle we explore in Time-to-Lead vs. Cost-per-Lead: Which Predicts ROI? — the metric your dashboard shows you is rarely the metric that predicts revenue.

What Degrades Lead Quality as Lookalikes Expand?

Understanding why quality drops prevents you from misreading the data when you test this yourself.

Signal dilution: Your seed audience is the ground truth. At 1–3%, Meta is matching on strong, overlapping signals. By 7–10%, the algorithm is filling seats with users who share a subset of signals — maybe demographics, but not intent behaviors.

Geographic dilution: If your seed is 400 past customers in one city, a 5%+ lookalike may pull in users from adjacent metros or suburbs outside your service area. They may convert on the ad but can't book — or they book and cancel when they realize you don't serve their zip code.

Offer-fit mismatch: Broader audiences include users who are loosely similar but in a different life stage or housing situation. A roofing company's 1% lookalike skews toward homeowners who recently searched storm damage. A 10% lookalike starts including renters who clicked one home-improvement article.

A strong seed list is the single biggest lever you have. If your seed is thin (under 300–500 matched users), even your 1% lookalike is working from a noisy foundation — a point worth revisiting alongside GBP Call vs. Website Click: Which Predicts Lower CAC?, since GBP callers often make higher-quality seed events than generic site visitors.

The Testing Protocol: How to Find Your Inflection Point

Don't guess your optimal tier. Run a structured test with clear success criteria before committing budget.

Step 1 — Define your qualification signal. Decide in advance what counts as a qualified lead: booked appointment, answered call over 90 seconds, form fill with a valid service address. CPL without this definition is just noise.

Step 2 — Split test at minimum 3 tiers. Run 1%, 3%, and 5% as separate ad sets within the same campaign, same creative, same budget allocation. Don't test more tiers than your monthly spend can fund to statistical confidence — a rough rule of thumb is at least 30–50 lead events per ad set before drawing conclusions.

Step 3 — Track downstream, not just dashboard. Tag leads by audience tier in your CRM or intake sheet. Measure lead-to-appointment rate and lead-to-closed rate at the 30-day mark. If you're running Smart Bidding on Google in parallel, the same principle applies — data volume drives decision quality, as outlined in Google Smart Bidding: How Much Data Local Ads Need.

Step 4 — Calculate cost per booked appointment, not CPL. Multiply nominal CPL by your tier-specific lead-to-appointment rate. The tier with the lowest result is your current optimum. Reassess every 60–90 days as your seed audience grows and audience composition shifts.

When Broad Lookalikes Are the Right Call

Broader tiers aren't always wrong. There are specific conditions where expanding past 3% makes strategic sense:

• Aggressive growth phase: If you're scaling spend significantly month-over-month, a 1–3% tier will exhaust before your budget does. Expanding reach is necessary — just pair it with tighter creative targeting or lead form qualification questions.
• Thin local population: In smaller metros, a 1% lookalike may simply not contain enough users to deliver. Expand until you have a viable audience size (Meta recommends seed audiences of 1,000–50,000 users for best results — that's a published platform guideline, not an estimate).
• Retargeting as a backstop: Broader lookalikes become more efficient when paired with a strong retargeting layer that recaptures the higher-intent users who engaged but didn't convert. Broad prospecting fills the top; retargeting does the qualifying work.

The Bottom Line

Most local advertisers are optimizing the wrong number. A falling CPL from a broader lookalike audience isn't a win — it's a warning sign unless you've confirmed that downstream qualification held.

The framework is simple: 1. Build a high-quality seed audience (past customers, GBP callers, closed deals). 2. Test 3 tiers minimum with a pre-defined qualification signal. 3. Calculate cost per booked appointment, not cost per lead. 4. Find the tier where that number bottoms — that's your baseline. 5. Reassess when spend, creative, or seed audience changes materially.

If you want a second set of eyes on your current lookalike setup — and a clear read on where your real CAC is hiding — [book a free strategy call with Nika Spark](/contact). We'll model your specific numbers, not a generic benchmark.