Buy a one-off list when you need a single campaign and accept decay. Licence a database when you run continuous programmes and need refresh. Build in-house when your ICP is under a few hundred companies, where manual depth genuinely beats index breadth.
The three options compared
| Dimension | Buy a one-off list | Licence a database | Build in-house |
|---|---|---|---|
| Freshness | Decays from day one | Refreshed on cadence | As fresh as your last sprint |
| Filter granularity | Whatever the seller offers | Title, function, geo, industry, tech stack | Whatever you can research |
| Verification | Rarely disclosed | Continuous, method disclosed | Manual, inconsistent |
| Cost profile | Low upfront, high waste | Scoped to volume and filters | High labour cost, hidden |
| Time to usable data | Days | Days | Weeks to months |
| Compliance basis | Often undocumented | Documented per geography | Yours to establish |
| Best for | One campaign, one test | Continuous programmes | Very narrow ICPs |
When a one-off list is the right answer
Buying a list makes sense in a genuinely narrow case: you want to test whether a segment responds at all, the test will run within a few weeks, and you accept that the file is disposable. As a cheap experiment it is defensible.
It stops being defensible the moment it becomes the ongoing data source. Teams buy a list, run it for a quarter, watch response rates fall, and conclude the channel does not work — when what actually happened is that the file decayed and nobody replaced it.
When licensing is the right answer
Licensing earns its cost when you are running continuous programmes: outbound sequences that need fresh records monthly, ABM lists that need committee coverage maintained, paid campaigns that need matched audiences refreshed, or syndication filtered to a live account list.
The decisive factor is refresh rather than volume. If your working list needs to be current in six months’ time, licensing with scheduled re-extracts costs less than repeatedly buying files and cleaning them.
When building in-house genuinely wins
There is a real case for manual research and it is worth stating honestly, because most data vendors will not. If your ICP is under a few hundred companies, manual depth beats index breadth every time.
At that scale a researcher can map the full buying committee, find the trigger events, read the annual report, note the recent hires and understand the organisational structure. No index provides that. The economics work because the account count is small and the deal values are usually large.
- ICP under roughly 300 companies → build in-house, or licence the base layer and research on top
- ICP in the thousands with continuous programmes → licence
- One test campaign, disposable file → buy a list
- Anything requiring committee depth on named accounts → manual research, whatever the source
The hybrid most teams end up with
In practice the effective setup for mid-market B2B is usually licensing for breadth plus manual research for depth. The database supplies the addressable universe, firmographic filters and the bulk of contact records. Manual work adds committee mapping and trigger research on the Tier 1 accounts that justify it.
That split maps neatly onto how ABM tiering works: Tier 3 runs entirely on licensed data, Tier 2 gets partial enrichment, Tier 1 gets human research. Spending manual research time uniformly across a large list is the most common way to waste an SDR’s month.
The question that settles it
Before comparing vendors, count your addressable market. Not the TAM slide — the actual number of companies that match your ICP, could buy this year, and are reachable through channels you can afford.
If that number is under a few hundred, no database purchase will help as much as a researcher and a spreadsheet. If it is in the tens of thousands, manual research is a rounding error against the opportunity and licensing is obviously correct. Most of the difficult decisions sit between those poles, and they are usually resolved by the hybrid.