295 people ticked a box on a registration form saying they were in the market for HR software. About eighty days later a person on our team reached the same people and asked the same question out loud.
32 were still shopping.
89% of your hottest leads were describing a situation that no longer existed by the time a rep called.
Your scoring model isn't broken. It counts activity, and it counts activity well.
Activity became a stand-in for readiness. Twenty years of plumbing got built on that assumption. Nobody re-checked it.
The people who said no
We ran the same test in reverse, on the 2,481 people who ticked not in market and later took a call. 4.1% of them turned out to be shopping.
That's 101 real buyers, against 32 from the group who said yes. Suppress on that form field and you throw away three quarters of your actual demand.
So the tick isn't worthless. It carries 2.6 times the lift, which is a real signal worth having. It's also wrong nine times out of ten, and the people who told you no are where most of your buyers are sitting.
When this isn't your problem
If sales wants more names than you can physically produce, this isn't your first move. Conversation-based demand won't out-volume syndication and isn't trying to. Fix supply, then come back to quality.
It's the right move when your best-scoring leads keep dying at first contact, or when you're being asked for contribution margin while holding a lead count. You can't report influenced pipeline off form fills. You can off conversations.
Free guide for marketing leaders
300 leads in the queue. Your team is working 40.
The MQL model has a structural flaw, and better scoring won't fix it. This guide shows you how to audit whether it's hitting your program, and what the math looks like on the other side.
Three ways to buy demand
There are three different products sold under the word “demand,” and the difference between them is what your reps actually receive.
- Names off a form
- Contact details and a score. That's real data and it isn't nothing. It tells you somebody touched something you published.
- A model's guess
- An account and a probability, assembled from anonymous signals and sold to every vendor in your category on the same day. Closer to the buyer than the form. Still inference.
- What the buyer said out loud
- The problem in their words, what they're running now, when the contract ends, and who else has to sign.
The deepest signal is the scarcest and the broadest is the cheapest. Rover Insights sells both ends, and we'll tell you which one you're buying.
Stated, not inferred
A person on our team gets a member of the HRMorning or ResourcefulFinancePro community on the phone. Not a survey with a gift card attached. A conversation worth the person's time, in which five things get asked and answered.
The questions are the ones a good rep would ask if they could get twelve minutes: what they're running, how it's going, when the contract ends, who else weighs in, and what would have to change for them to move.
That produces roughly 50 fields per call, about 38 of them from the conversation itself rather than the contact record.
One term for the rest of this. A Context Qualified Lead (CQL) carries all of it: the problem in their words, the system they're running now, the date the contract ends, and the names of the other people who have to sign.
That's what you get to hand your sales team. Not a name, a firm and maybe a score, but the shape of the opportunity before a rep spends an hour finding it out.
Buying habits change weekly
March: this buyer is happy with what they have. July: their renewal lands in Q1. October: they're shopping.
Your funnel sees three different people there. If you bought two webinars and a lead cap, you met one of them, and probably not the one that mattered.
That's the argument for always-on over campaign-shaped, and it has less to do with budget than with how often somebody is asking. A buyer who tells us in March that they're staying put has just given you a date to come back on.
Why the declared number always runs high
This isn't only our finding. Employ Inc. surveyed more than 1,200 US talent acquisition professionals and found 82% satisfied with their core systems while 76% said they were planning major technology updates.
Both answers came from the same people. Declared intention runs well ahead of a real replacement, and a form field is the thinnest version of a declared intention there is.
For scale on the live market: across 10,203 conversations, 2.1% of HR buyers were actively looking to replace their system. Every vendor in your category is bidding on the same two percent.
I've built funnels on the MQL for most of my career, and the model isn't dumb. Neither are the people running it.
But eighty-nine out of a hundred people who told a form they were shopping weren't shopping when somebody asked them out loud. We were reading the tea leaves very carefully when we could have picked up the phone.