Here's the trap. A vendor demos their database by searching for companies in your industry, the results look excellent, and you sign a year. The search worked because they chose it. Your own list is a different list, and you find that out in month two.
You can test properly for about $52. Two hundred records from your real target segment, run through two vendors, verified through a third, and actually sent to. That is the only sequence that produces a number rather than a claim.
The awkward part is that no free tier in the data category is large enough to run it. The biggest recurring free allowance is 100 credits, which is half a test.
So this page gives the method, the exact budget, and the three numbers you have to count separately because the vendors report them as one.
Why 200 records?
It is large enough to tell a 60 percent hit rate from a 90 percent one clearly enough to act on, and small enough to run on a single $15 plan. We are not going to quote you a confidence interval, because we have not done that statistics and inventing one would be exactly the kind of unearned precision this whole subject already has too much of.
What matters more than the number is where the records come from. Build the sample from the awkward part of your ICP. Small companies, non-US domains, non-technical job titles, the segments you already suspect are thin. A vendor's demo list is selected to succeed, and a sample of your best-known accounts will flatter every tool equally.
Can you run it on free tiers?
Not in one go. The allowances are all smaller than the test.
| Tool | Free credits | How much of a 200-record test that covers |
|---|---|---|
| Prospeo | 100 a month | Half |
| Clearout | 100 at signup | Half |
| Hunter | 50 a month | A quarter |
| Persana | 50 a month | A quarter |
| FullEnrich | 50 once | A quarter |
| BetterContact | 50 once | A quarter |
Two months of Prospeo's free tier gets you to 200, which is a real option if you are not in a hurry. Otherwise you pay, and paying is cheap. We set out which free tiers do real work separately, including the three that will not let you export at all.
What does the test cost?
The cheapest route is exactly $15, and it is exact in a way that looks almost deliberate.
| Tool | Entry price | Credits | Covers 200 records? |
|---|---|---|---|
| BetterContact Starter | $15 | 200 | Yes, exactly |
| Clearout Starter | $23 | 3,000 | Yes |
| FullEnrich Pro | $29 | 500 | Yes |
| Snov Starter | $39 | 1,000 | Yes |
| Prospeo or Hunter | $49 | 2,000 | Yes |
| LeadMagic Basic | $49.99 | 2,000 | Yes |
| UpLead Essentials | $99 | 170 | No |
The most expensive entry tier in the group is the one that cannot complete the test. UpLead's $99 plan carries 170 credits, thirty short. In fairness its credit buys a contact with the email and the mobile together, which is a different product, but if you want to run a standard 200-record evaluation you cannot do it on their entry plan.
Why does testing phone numbers cost three times as much?
Because a mobile number costs ten credits at most vendors and an email costs one. Two hundred credits buys 200 emails or 20 phone numbers, and twenty results is not a sample.
To test phone coverage across the same 200 records you need around 2,000 credits, which moves you from the $15 tier to the $49 one. At LeadMagic's five-times multiplier it is 1,200 credits, which is why that tool comes out cheapest per complete contact in the cost comparison.
What is the actual method?
- 1. Build the sample. 200 records from the hard part of your ICP. Keep the company name, the person's name and the domain, and nothing else, so you are testing the vendor rather than your own existing data.
- 2. Run it through two vendors, not one. A single result tells you nothing about whether 62 percent is good. One single-source tool and one waterfall is the most informative pairing, and the gap between them is the thing you are actually buying.
- 3. Count three things separately. How many came back with anything. How many came back with a mobile. How many actually reached a human. The category reports all three as one number and they are not one number.
- 4. Verify through a third party. Not the tool that supplied the data, for obvious reasons. Clearout's 100 free credits cover half the sample and NeverBounce at $8 a thousand covers the rest.
- 5. Send to it and count hard bounces. This is the only step that produces accuracy rather than a claim about accuracy, and it is the step everyone skips.
What does a good result look like?
Judge it against decay rather than against the vendor's claim. A list sold as 95 percent accurate is around 88 percent at three months on the standard decay range, so a fresh sample bouncing at five or six percent is behaving normally rather than badly.
On coverage, the useful comparison is between your two vendors rather than against a published figure. If the waterfall finds thirty points more than the single source, that gap is worth about eleven cents a contact and is a straightforward buy. If it finds five points more, you are paying a premium to fill in the last few rows.
And if both come back low, the problem may be your ICP definition rather than the vendors. Two independent databases missing the same segment usually means the segment is genuinely under-covered, which is itself worth knowing before you build a quarter's pipeline plan on it.
The whole exercise costs about $52 and half a day. Against an annual data contract, and against the vendor accuracy claims that nobody independent has ever tested, it is the cheapest thing you will do all quarter.