Every data conversation starts in the same place. How many records, and how much each? It is an easy number to compare across suppliers, which is exactly why it has become the number everyone negotiates on.
It is also close to meaningless on its own.
The arithmetic nobody does
Say you buy 20,000 records at a low rate because the price is good. The selection is broad — a handful of postcodes, an age range, a gender split. You send a campaign. A small fraction of those people were ever plausible buyers, and your response reflects that.
Now say you spend the same money on 5,000 records, selected far more tightly. Four times the cost per record. If the response rate more than quadruples, you are ahead — and when a selection goes from “lives in this general area” to “lives in a pocket where this behaviour genuinely concentrates”, it often does.
The number that matters is not cost per record. It is cost per reply, or cost per sale. Those are the only figures that connect to the business.
Why the market sells volume anyway
Because volume is easy to compare and easy to deliver. Any supplier can produce a large file quickly. Producing a small, well-reasoned one takes modelling, behavioural data and an argument about why these records and not those.
It is also easier to sell. “500,000 records” sounds like more than “12,000 records”, and a buyer being measured on reach will pick the first every time.
The problem surfaces later, when someone asks what the campaign actually produced.
A worked example
These figures are illustrative, but the shape is typical.
A client with a narrow consumer product has been buying broad lists and getting results they describe as acceptable. Not bad. Not worth scaling.
We rebuild the selection. Instead of postcodes, we work at SA1 level — areas of 200 to 800 people — and weight toward the ones where the relevant purchasing behaviour actually concentrates. The list comes back a fraction of the size.
The client spends roughly the same money. They contact far fewer people. More of those people are interested, because more of them were plausible buyers before the first message went out.
The uncomfortable part, if you are the supplier, is that the invoice is similar while the volume is much lower. Which is fine, as long as everyone is being measured on the right thing.
What makes a small list work
Geography fine enough to mean something. A postcode can span dozens of suburbs. An SA1 holds a few hundred people. Only one of those tells you where buyers concentrate.
Behaviour, not just demographics. High income with kids is not a buying signal. What people actually spend money on is.
Channels you can actually use. A tightly selected list is wasted if you can only reach it one way. Consented data means email, SMS and phone are all open, so one audience supports a sequence instead of a single shot.
Someone to send it. A good list that sits in a spreadsheet for three weeks while you arrange suppliers is not a good list yet.
What to ask next time
When a supplier quotes you a rate per record, ask what proportion of the file they would expect to be genuinely in market, and how they know. The answer tells you whether you are buying a selection or an extract.
Then ask what it would cost to do the same campaign with a quarter of the records and four times the thought. That conversation is usually more interesting.
Curious what a tighter selection looks like for your category? We will show you the counts before you commit to anything. Get in touch.
