Intent segments beat demographics, and it is not close
Knowing someone is a 34-year-old woman in Bogotá tells you very little. Knowing she has compared three refrigerators this week tells you almost everything.
Demographic targeting survives in media planning for two reasons: it is easy to buy, and it is easy to explain to someone who is not in the room. Neither reason has anything to do with whether it works.
For commerce advertising specifically, the gap between demographic and behavioural targeting is not marginal. In the tests we run, it is routinely the difference between a campaign that clears its target and one that does not.
Why demographics underperform for commerce
A demographic segment describes who someone is. A purchase decision is about what someone is doing this week.
Those two things overlap loosely at best. Household appliance purchases are made by people of every age and income band, clustered around events — moving house, a breakdown, a renovation, a promotion — that demographic data cannot see. Targeting the demographic reaches the ninety-eight percent of that group who are not currently in the market, at full price.
The result is a campaign that looks well-targeted in the plan and performs like broad reach in the report.
What an intent segment actually contains
The useful signals in commerce are behavioural and recent:
Category consideration. Repeated engagement with a product category over a short window. Someone who has visited comparison pages for the same category four times in ten days is in a different state from someone who did it once.
Comparison depth. How many products, how many brands, how many price points. Depth correlates with proximity to purchase far better than volume of pageviews does.
Price sensitivity. Whether the behaviour clusters around discount content, filters by price, or engages with the premium end of the range. This changes the creative, not just the bid.
Store proximity. For categories where purchase is physical, distance to a relevant store is a strong modifier — and one that is invisible to any demographic model.
Stage. Early research behaves differently from late-stage comparison. Both are worth reaching; they are not worth reaching with the same message.
The perishability problem
The critical property of intent data is that it decays quickly. A consideration signal from six weeks ago is close to worthless — the person has either bought or moved on.
This is where a great many intent-targeting products quietly fail. Segments are built on ninety-day windows because that produces satisfying audience sizes in the planning tool. A ninety-day commerce intent segment is mostly people who are no longer shopping, sold at a premium on the strength of the word “intent”.
Short windows produce smaller, more expensive, dramatically more effective audiences. Buyers who have run the comparison properly stop objecting to the size.
What this asks of the seller
Selling intent honestly requires being specific about things the industry usually leaves vague: the observation window, the recency distribution, the signal source, and the refresh cadence.
A buyer who knows a segment is built from seven-day comparison behaviour, refreshed daily, can price it correctly and plan against it. A buyer handed a segment called “Appliance Intenders” with no window disclosed is being asked to take it on faith — and after enough disappointing campaigns, they stop.
The advantage of doing this properly is not really the performance. It is that the second campaign gets funded.