Paid media
Estimate addressable reach when scaling lookalike audiences by market.
- Category
- Paid media
- Signup
- None
- Your data
- Stays in your browser
- Also in
- Claude, via MCP
How big is a lookalike audience?
A lookalike audience is a percentage of the platform's users in the country you choose, from 1% (most similar to your source) to 10% (broadest). Pool size = market population × lookalike %. A 3% lookalike of a 4 crore market is 12 lakh people; the 1% version is 4 lakh. Source audience size changes quality, not pool size.
A 2,500 seed works but isn’t robust. Your 3% lookalike spans ~1,200,000 people, start at 1–2% and widen only if it converts.
Seed purity beats seed size, a lookalike of your best-LTV customers outperforms a bigger list of all buyers.
About this calculator
A lookalike audience is only as good as the seed it was built from, a broad reach percentage built on a thin, noisy seed list produces a large but low-signal audience. This calculator sizes your addressable lookalike pool from market population and match percentage, then separately grades your seed list against the volume a platform actually needs to build a clean model.
How to use it
- Enter your target market population, the addressable users in the country or region you're targeting.
- Enter your lookalike percentage, tighter percentages (1%) match closer to your seed, broader ones (10%) trade precision for reach.
- Enter your seed audience size, the customers or leads the model is built from.
- Read the addressable pool size and your seed quality grade, and use the tighter 1% pool as the higher-precision reference point.
Methodology
Addressable pool is market population × lookalike percentage, a direct scaling of the total population by the match tightness you selected.
Seed quality is graded against two thresholds: below 1,000 the seed is flagged as too small (the model will likely be noisy), between 1,000 and 5,000 it's usable but not robust, and 5,000+ is considered a strong, reliable seed.
The 1% pool is shown separately as a reference, since it represents the tightest, highest-similarity match available regardless of what broader percentage you're currently modeling.
This is a volume calculation, it says nothing about seed quality beyond size. A seed of 5,000 low-value trial users will still produce a weaker lookalike than a seed of 1,500 high-LTV repeat customers, seed composition matters as much as seed count.
Lookalike pool size and seed check
Addressable pool = Market population × Lookalike % 1% reference pool = Market population × 1% Seed check: under 1,000 too small, 1,000 to 4,999 usable, 5,000+ strong
- Market population
- Platform users in the target country or region, not the total population. Use the platform's audience estimate for the country.
- Lookalike %
- How broad the lookalike is: 1% is the closest match, 10% the widest.
- Seed audience
- The source list the platform learns from, such as customers, high-value buyers or qualified leads.
The seed thresholds of 1,000 and 5,000 are this tool's working guidance. Meta's technical minimum is far lower: a source of at least 100 people from one country (Meta Business Help Centre). Above the minimum, what is in the seed matters more than its size: a list of repeat buyers usually beats a larger list of everyone who ever signed up.
Worked example: a D2C brand targeting India
A brand estimates 4 crore reachable platform users in its target states. It has 2,500 repeat customers to use as a seed and is choosing between a 1% and a 3% lookalike.
- 3% pool = 4,00,00,000 × 3% = 12,00,000 people.
- 1% pool = 4,00,00,000 × 1% = 4,00,000 people.
- Seed check: 2,500 is between 1,000 and 5,000, so it is usable but not strong.
- Adding one-time buyers with an order value above the median could take the seed past 5,000 while keeping it close to the best customers.
Start with the 1% pool to test whether the lookalike finds buyers at an acceptable cost, then widen to 3% once it works and needs more room. Check how long the pool will last at your budget with the ad creative fatigue forecaster, and keep the seed fresh by refreshing it monthly from your store or CRM.
What makes a good lookalike seed?
A good seed is a list of the customers you want more of:
- Value over volume: repeat buyers or high-value customers, rather than everyone who bought once.
- Recent: customers from the last 6 to 12 months reflect today's product and price.
- One country: lookalikes are built per country, so use a seed from the market you target.
- Clean: remove employees, test orders and refunds.
If the best seed is small, value-based lookalikes or a slightly broader seed (such as all buyers with two or more orders) are better than padding it with low-value contacts.
Lookalike audiences in India vs the US
India's large user base means even a 1% lookalike is big, often larger than a small brand can reach at a useful frequency. Narrow by state, city tier or language alongside the lookalike if your product sells unevenly across the country. COD buyers who refused delivery should be removed from the seed, or the lookalike learns to find more of them; the RTO-adjusted ROAS calculator shows what those orders cost.
In the US, privacy changes have reduced the signal available for building lookalikes from website events, so seeds built from customer lists usually outperform pixel-based seeds. Many advertisers now compare lookalikes with broad targeting, letting the platform find buyers without a seed. Test both, judged on cost per acquisition. The post-iOS 14 blended attribution modeller helps read results when tracking is incomplete.
Common lookalike audience mistakes
- Using total population instead of platform users in the market.
- Building the seed from all leads or all buyers instead of the best customers.
- Never refreshing the seed, so the lookalike reflects old customers.
- Running several overlapping lookalikes in separate ad sets, which compete against each other.
- Jumping to a 10% lookalike to get scale before the 1% version has been proven.
Sources
- About lookalike audiences, Meta Business Help Centre. Accessed 2026-10-19.
FAQ
A 1% lookalike selects only the users most statistically similar to your seed, a 10% lookalike casts a much wider net that includes progressively less similar users to hit the larger reach target. Tighter percentages sacrifice reach for precision.
Widen what counts as a seed event, include high-intent actions (add-to-cart, demo request) alongside purchases, combine multiple markets if you're targeting internationally, or wait to accumulate more volume before building the lookalike, a noisy small seed often performs worse than no lookalike at all.
It's a reasonable default when seed quality is high, since it gives you the highest-similarity audience to test first. If it's too small to spend efficiently, widen gradually (2%, then 3%) only after confirming the tighter tier is converting well.
Generally yes. A lookalike built from your best-LTV customers specifically will typically outperform a larger lookalike built from every buyer regardless of value, quality of the input signal compounds through the whole modeled audience.
A lookalike audience is a group of people a platform identifies as similar to a source list you provide, such as your customers. It is used to find new prospects who resemble existing buyers.
A 1% lookalike is closer to your seed, so it often converts better at first. A wider percentage gives more room to scale. Test them against each other on cost per acquisition, not on click-through rate.
They still work, especially from customer-list seeds, but website-event seeds carry less signal than before. Compare them with broad targeting in your own account to decide.
Numbers looking off?
If the model says one thing and the pipeline says another, that gap is usually where the revenue is. A 30-minute call finds it.
It starts with a 30-minute call. Pick a time below.