Blog / The leads were too cheap: what we found in a client’s CRM, row by row

Your team doesn’t care about pipeline if you measure them on leads.

I’ve worked with so many clients over the years where the target is lead volume, and so often when we look under the hood at what the marketing is producing, there’s so much crap going through that it was never going to convert into anything that really affects the business positively.

But no one on the team had spotted it because they were hitting the targets they were set and nobody was measuring past the form fill. Once you measure deeper, into MQLs or sales qualified leads or, even better, pipeline, and most of those leads stop existing.

And that’s a good thing.

On a call last year we told one of those clients that their lead volume was about to fall off a cliff, and we were going to be the ones pushing it. We’d traced what their campaigns were producing and found leads that were never going to buy anything, from placements no human being was really looking at.

The plan was to switch off the sources feeding them.

A few months later their MQL qualification rate had gone from 10% to 70%. Sales stopped complaining about chasing dead leads  and the account was producing more real pipeline on the same budget than it ever managed in the volume era, with a fraction of the leads.

Auditing CRM Data: What Was Actually Under the Hood

The client shared their CRM screen on a call and we went through the lead rows one by one.

I’d recommend every marketing leader do that at least once a quarter to keep your finger on the pulse of what quality looks like to the individuals and teams inside your company.

There were personal Gmail addresses attached to companies that don’t exist, students and another market delivering truck drivers to an enterprise software company. So much fluff at the eyeball test.

It gets missed because none of that was visible in the ad platform, where the same campaign looked like the best thing the account had ever done. The platform’s job ends the moment the form is submitted, but everything wrong with the lead shows up further down the line, in systems the platform never sees (unless you have set it up to send that data back, which most have not).

The tell had been sitting there for months, and it’s the one I’d teach any marketing leader to spot first. The leads were simply too cheap.

I guess I’m such a jaded agency owner after all these years that when I see a key metric flying in the right direction, I first think it’s an issue rather than real, so I don’t celebrate any success too early. If it looks too good to be true, sorry, but it probably is.

This account’s history said a decent lead cost around $300, and one campaign type was suddenly producing them at $15. Nobody asked any hard questions about the quality when it was making the reports look good, so it ran and ran. When we finally pulled the placements apart, the overwhelming majority of those impressions were being served on low-quality website inventory, the kind of sites and apps that exist to be accidentally tapped. Leave Google unchecked for too long and it can drift like this, so it’s critical to stay vigilant.

We’ve made this a standing rule across accounts since then. Any conversion spike gets checked for quality BEFORE anyone celebrates it. I’ve watched campaigns spend thousands producing leads that never touched the pipeline because no one was checking or, let’s face it, being measured on anything deeper than the form fill.

And the people we’re dealing with know this. It’s been the message hammered out on LinkedIn for years! Leads mean nothing. So I guarantee that in nearly every one of these accounts the marketing team could have told me which sources were rubbish, because they sit closer to the numbers than anybody else in the company. But they also know exactly what they’re going to be judged on when the quarter closes, and if we walk in and cut half the budget and half the leads go with it, they’re the ones who get asked a lot of questions.

So when we propose the cut, they start to panic, because they know leadership is going to start panicking too. I have a lot of sympathy for that. The marketing lead sitting in front of me did not design what gets counted in their company, and most of these companies count department metrics. It’s a problem they inherited, and they have to survive inside it for as long as it lasts.

The ladder underneath all of this, form fill to MQL to SQL to pipeline, is worth getting straight before you touch any of the media. I’ve written that part up separately, because it’s really a board conversation rather than a campaign one.

How Cutting Low-Quality Leads Improved MQL Rates

The first of the two examples here is the client this article opened with. We took their demand generation off the placement networks producing junk and added exclusion lists a thousand terms deep. Reported lead volume collapsed by around 90%, and the MQL rate went from 10% to 70%. What sales received stopped being a lottery.

The moment that stayed with me came later, when the client was presenting the quarter internally and the smaller number was the one they were proud of, from the delivery team right through to the executive team.

The second client sold cloud infrastructure. There we cut reported signups hard across a year, on purpose, and the rate at which those signups turned into paying customers went up by more than the volume we lost. That was the best the account had ever recorded. I’ve written that one up in more detail elsewhere, including the figures and what their marketing lead took into his board meeting.

Part of it was killing two competitor campaigns responsible for a quarter of all signup volume. They had produced zero customers in months, with a quarter of the volume, but none of the business. On the old dashboard those campaigns looked essential, and every dollar they absorbed was taken from campaigns that were actually producing customers.

The thing that made both of these survivable was the least technical thing we did. Before we touched anything, we told the client roughly how far the reported numbers would fall, why, and which number to watch instead. So when conversions dropped 30% the following month, leadership didn’t raise a red flag. Their own marketing team had predicted it to within a few points. I think that’s most of the job, honestly. The number was going to fall either way, and all we did was say so first.

And I want to be honest about the trade-off. Cutting volume looks worse before it looks better, and the gap between the two can be a full quarter. If the people above you haven’t been prepared for the shape of that curve, all of this measurement discipline looks like performance decline.

How to Audit Your Lead Quality Without Breaking the Account

The mechanics matter, because using a chainsaw rather than a scalpel cuts good spend along with the bad. Judge every source by the deepest signal you can reach. Platform conversions are the shallowest end, signups are better, activation is better still, and a paying customer is gold.

At the cloud client we graded every keyword by whether its signups ever deployed anything. You don’t need perfect attribution for that. A monthly export from the CRM, joined by hand against campaigns, changed every spending decision we made for a year.

Cut the sources and leave the measurement signals alone. The instinct once you distrust a number is to rip it out of the account everywhere, and that breaks the algorithm’s learning along with the junk. Change one campaign at a time, and keep anything that produced customers recently. Run the old and the new setups alongside each other long enough to compare them honestly.

And respect small numbers. When a campaign produces seven customers a month, one more customer halves its cost figure. Realize that a single bad fortnight proves nothing, and a single good one proves even less. We hold kill decisions until we see the pattern has survived a few reads. I’ve seen that patience save campaigns a snap decision would have removed for the crime of a slow week.

The questions to ask your team on Monday

You can run most of this diagnosis in one meeting, without opening an ad platform. If you’re a marketing leader who wants to know whether any of this is in your account, these are the four questions I’d ask.

1. “Take our biggest lead source and tell me how many of last quarter’s leads became customers.” If nobody can answer at all, that silence is the finding. It’s usually fixable within a month.
2. “What does our cheapest lead cost, and why is it that cheap?” Cheap always has a reason. Make somebody go and find it.
3. “Which of our reported conversions would sales actually thank us for?” Ask a salesperson, in the room, while everybody is listening.
4. “If we cut every source that has never produced a customer, what happens to this quarter’s report?” If the honest answer is that the chart would look terrible, you’ve found where the incentive problem lives.

Ask those four in one meeting and you’ll know inside an hour whether any of this is sitting in your account. It’s a cheaper diagnosis than anything I could sell you.