Your best lead source might also be your worst. Here's how to track lead source performance by fit, so you can see which channels bring the customers you actually win.
Picture the Monday pipeline meeting. Marketing puts up the source report. Trade shows brought in 140 leads last quarter, webinars 60, referrals 12. Everyone nods. Trade shows get next year's budget.
Nobody asks the question that matters: which of those leads looked like the customers we actually win?
I've worked plenty of trade show stands as a sales rep. Scanning badges, swapping business cards, then getting back to the office and thinking: that might have been a complete waste of time, and I had no way of knowing. So I followed up with every contact as if they were a good fit to buy from us.
Volume is easy to count, so it's what gets reported. But a source that sends you 140 leads that look nothing like your customers isn't your best source. It's your busiest. Those are different things, and the gap between them is where sales time and marketing budget leak away.
Why volume and conversion rate both mislead
Volume rewards noise. A channel that sends lots of leads looks great on a bar chart. It says nothing about whether those leads are the right companies, at the right size, with the right buyer.
Conversion rate is better, but slow and fragile. Win rate by source only tells you something once deals have closed, which can take months. On small numbers it swings wildly: two wins from five referrals reads as 40%, and one more loss drops it to 33%. And it can't tell you anything about the deals still open, which is where you can still act.
What you actually want to know is fit. For each source, how much do its leads look like the customers you already win? That's a question you can answer today, about your open deals, without waiting for anything to close. (If you're not sure what good fit looks like for you, here's how to tell if a company is a good fit.)
How to measure lead source quality by fit
You can do this in a spreadsheet in an afternoon. You need a HubSpot export (or any CRM) of your deals with their source, plus a few facts about each company.
- Export your deals. Include closed-won deals and open deals, with the deal's original source, the company's industry, employee count and country, and the deal amount.
- Describe your wins. Look only at closed-won deals. Which industries, size bands and countries come up again and again? That's your real customer profile: not the one on the slide, the one in the data. (We wrote a whole guide to analysing your closed-won deals.)
- Score each open deal for resemblance. Give each open deal one point for each fact it shares with your typical win: same industry, same size band, same country, deal amount in your usual range. Four points means it looks just like your wins. Zero means it looks like nothing you've ever closed.
- Group by source. For each lead source, take the median score of its open deals. Use the median, not the average, so one outlier can't drag a whole channel up or down.
- Put the two numbers side by side. For each source: how many leads it sent, and how much those leads look like your wins. Don't combine them into one figure. The point is to see where they disagree.
- Add one more column: share of wins. Of all your closed-won deals, how many came from each source? It's a count, not a rate, so it can't be distorted by a source that sends lots of junk.
A worked example
Here's what it looks like for an illustrative B2B software company. The numbers are made up, but the pattern is one you'll recognise.
| Source | Leads last quarter | Median fit of open deals (out of 4) | Share of all wins |
|---|---|---|---|
| Trade shows | 140 | 1 | 9% |
| Paid search | 85 | 2 | 22% |
| Webinars | 60 | 3 | 31% |
| Referrals | 12 | 4 | 38% |
On volume, trade shows win easily. On fit, they come last: most of those leads share only one fact with the companies this business actually wins. Referrals are the smallest channel, but their leads look the most like existing customers, and they account for more of the wins than any other source.
None of this says trade shows are bad. It says the trade show leads in this pipeline don't look like this company's customers. That's a question about which shows, which stands and which conversations, not a reason to shut the channel down.
Three things that will trip you up
Small numbers. A source with three open deals can show any median it likes. If a source has fewer than about five deals behind a figure, leave the figure blank and show the count instead. A blank is more honest than a confident-looking number built on almost nothing.
Blank source fields. In most CRMs a big chunk of deals have no source recorded. Don't quietly drop them. Give them their own "Not recorded" row. If it's your biggest row, that's the first thing to fix, and it's a finding in its own right.
Messy labels. "LinkedIn", "linkedin", "LinkedIn Ads" and "LI" are probably the same channel. Tidy them into one before you group, or each will look too small to read.
What to do with the answer
The useful moments are where volume and fit disagree.
- High volume, low fit: the channel is working hard and bringing the wrong companies. Look at the targeting before you look at the budget.
- Low volume, high fit: this is the one to understand. What is it about these leads? Can you find more of them on purpose?
- A big "Not recorded" row: you can't manage what you don't record. Make source a required field and revisit in a quarter.
Then run it again next quarter. The aim isn't one perfect report. It's to see whether the changes marketing and sales make are bringing in more leads that look like your best customers.
Doing this without the spreadsheet
Telepath does the fit part inside HubSpot. It learns what your won deals have in common and scores your open deals with a T-Score: how much each one looks like your wins, from 0 to 100. Any figure resting on too few deals is left blank rather than guessed.
If you'd like to see who you actually win with first, the free ICP report reads an export of your closed-won deals and shows you the pattern in a couple of minutes. Related reading: what pipeline intelligence is and why sales ignores marketing leads.
