Why Target Households Instead of Lookalikes?

The lookalike paradox

Lookalikes are a confident answer to a quiet question: who else looks like my best customers? The trouble is that "looks like" is a behavioural fingerprint on a platform. Not a description of a person, a household or a buying context.

Scale a guess and you get more guess. The output is wider, not sharper.

How lookalikes actually work

You hand Meta (or TikTok, or Google) a seed list. The platform matches it to logged-in identities, then expands outward to users with similar on-platform behaviour. Pages liked, sessions, dwell, click patterns. You pick a percentile band (1%, 5%, 10%) and that becomes your prospecting pool.

None of those signals describe the household behind the screen. They describe a behavioural shadow.

What household-level actually means

A household audience starts from the opposite end: 30M+ UK households described by named attributes. Property type, tenure, family stage, purchasing power, move stage. You write a brief in plain English and the matching households become a Custom Audience.

Same Meta surface. Completely different seed quality.

Three things households tell you that lookalikes can't

• Tenure and move stage. Owner vs renter, recently moved vs settled. Decisive for furniture, kitchens, mortgages, energy.
• Life stage. Family composition, age band, children present. Decisive for everything from sleep to insurance.
• Neighbours. Affluence bands, area type, density. Decisive for AOV, channel mix and creative tone.

Side-by-side: a real test pattern

Run the same creative and offer through two audiences for one full purchase cycle:

1. Audience A: a 1% lookalike of your top-LTV customer file.
2. Audience B: a household brief. For example, "Owner-occupied, 3+ bedrooms, High Purchasing Power, Recently Moved".

In considered, home-anchored categories the household audience consistently shows lower CPA and higher AOV, because the seed isn't probabilistic. It's the brief you would have written if Meta had let you.

When lookalikes still make sense

• Pure top-of-funnel reach where audience precision genuinely doesn't matter.
• Impulse, low-AOV categories with weak household signal.
• Markets without a high-coverage household graph.

Outside those, the lookalike is the easy answer, not the right one.

How to start

1. Pick one campaign that currently leans on lookalikes.
2. Write a household brief that describes your best customers in plain English.
3. Run them side by side, same creative, for one full purchase cycle.
4. Read CPA, AOV and incrementality, not just CTR.

See the related head-to-head in household audiences vs lookalike audiences and the broader case in beyond lookalike audiences.

The bottom line

Lookalikes scale similarity. Household targeting scales fit. In any category where context drives the purchase, fit wins.