What we were looking at
We had a list of 1,151 companies. We had pulled them out of the CRM because they matched the criteria we were targeting, and 195 of them had already been given to SDRs to work on.
Then the usual question came down from above. Are these companies any good, and what has happened on them since we handed them over.
The first check, and why it was not enough
The second half of that question looked simple, because the CRM has the answer. There were 507 contacts saved on those companies, and 115 of them had ever been contacted by anyone.
The easy reading is that the team did not work the list. I could not draw that conclusion, because the CRM only knows about people who are already saved in it. If most of those 507 contacts were the wrong people, then a rep who skipped them did the right thing, and the low number says something about our data rather than about the team.
So before I said anything about how the list had been worked, I needed to know something the CRM could not tell me. Can we even find the right people at these companies?
The test
I took a sample out of the same list of 1,151 companies. Two companies first, to check that my setup did what I thought it did, then 10 properly.
We use Clay for this. It searches several contact databases at once and gives back the people it can find at a company. I only let it do the first step, which is finding out who works there. The second step, looking up their email address and phone number, is the part you pay for per person, and I did not want to pay for that until I knew what kind of people were coming back.
What came back gets smaller at every step:
| Step | Count | What this step removes |
|---|---|---|
| Companies in the sample | 10 | |
| Companies where the tool found at least one person | 9 | One company had nobody findable |
| People the tool returned | about 92 | Each one had a LinkedIn profile to check |
| Of those, working at the company itself | 80 | about 12 who work somewhere else |
| Of those, in a role we sell to | about 15 | the rest are not decision makers |
Before the explanations, one word about the 90 percent, because "match rate" gets used for two different things and they are easy to mix up.
| What gets counted | The question it answers | Who usually quotes it |
|---|---|---|
| Companies where at least one person was found | Can this tool reach the companies on my list | What I measured here, 9/10 |
| Submitted contacts that came back with data | How complete is their database on people I already know | What a data provider usually means |
Both get called the match rate. Mine is the company-level one, and it is the one that matters when you are starting from a list of target companies rather than from a list of names.
Two of the steps in the table need explaining, because neither is obvious.
The LinkedIn profile is the check. A profile is how you confirm that the person is real and still has that job. A name and a job title with nothing behind them can be years old, and once it sits in your CRM it looks just as reliable as a verified one.
"Working at the company itself" is not a given. The dozen or so who did not worked at a subsidiary, at an agency working for the company, or at a company they left years ago that still shows on their profile. In a spreadsheet you cannot tell them apart from the rest.
The last step is the one no tool does for you. I went through the 80 people by hand and checked which of them held one of the job roles we sell to.
The number I would plan with
18 percent. Of everything the tool gave back, that was the share we could actually put in front of a rep.
About 15 of the 80 people who really worked at those companies held a role we sell to.
It is also a good result, and I want to say that clearly. 10 companies gave us 15 named decision makers a rep could write to that week. There is nothing wrong with that.
The problem was that nobody was measuring it. Our reporting showed the match rate, the 90 percent, which tells you whether the tool can find people at these companies. What nobody measured was whether it finds our people, and that is the number that decides whether a list of 1,151 companies is worth working at all.
From six months inside a B2B GTM organization with 36,000 accounts in its CRM.
Why the two numbers are so far apart
They answer different questions, and only one of them is anybody else's job to produce:
| Match rate | The number you plan with | |
|---|---|---|
| What it measures | Can the provider find anyone at these companies | Can it find the people we sell to |
| In this run | 90 percent | 18 percent |
| Who can report it | The provider. It is their database | Only you. Only you know who you sell to |
| What it decides | Whether a provider is worth using at all | Whether this list is worth a rep's time |
The match rate belongs to the provider and describes how good their database is. The second number comes out of the list of job titles you told the tool to look for, which is why no provider can work it out for you.
That list of job titles decides everything, and I got it wrong twice in one afternoon:
| The list I used | What came back |
|---|---|
| Narrow and exact titles | Nothing at all. Not a small number, zero |
| Wide titles | Everyone the tool could find, whatever their role |
Neither run was usable. The first was too strict to return a single person, the second stopped telling roles apart and gave back volume instead of a buying committee.
The setting that works is a list of job titles somebody keeps working on. In most companies nobody owns that list, it sits inside a tool where nobody looks at it, and it gets changed by whoever last ran into a problem with it.
What to do instead
- Test before you buy volume. Two companies, then 10, and only the first step where the tool finds out who works there. It costs almost nothing and it changes what you buy.
- Count the right people, not the matches. Take what came back, remove everyone who works somewhere else, and count how many of the rest your reps would really write to. That percentage is what you plan with.
- Put a name against the job title list. It decides the second number, and in most companies nobody owns it.
- Run the test again whenever that list changes. The number moves with it, and nothing in the tool will tell you that it moved.
What the match rate is still good for
It tells you whether a provider can find anyone at all at the kind of company you care about. One that cannot is out before you go any further, and that is worth knowing.
What it does not tell you is whether the list you just bought is worth working on. For that you need the second number, and you have to work it out yourself.