What they believed walking in
A profitable, family-run training company came to us in the middle of its biggest year ever, and it was still worried. A slow quarter for new leads had spooked the owners, and they had a clear read on why. They needed more leads, their search ranking was holding them back, and they had just hired a search team to fix it.
That was a reasonable place to land. When the leads slow down, more leads is the move every owner reaches for first. They could see traffic, and a dashboard showing the cost of a lead dropping by about 66 percent, which felt like proof the problem was volume and nothing deeper. And the salesperson kept losing deals the owners thought should have closed, so the natural read was that he needed to be better.
So the ask was simple and honest: get us more good leads, fix the search, and help the sales calls land.
What we saw
Before we touched anything, we looked at their own numbers, because you cannot fix what you cannot see. The first thing we found was that they could not see it either.
The dashboard was counting the wrong thing. A conversion was firing the moment a visitor opened a contact form, on the click, not when anyone actually sent it. So that flattering 66 percent drop in cost per lead was, in plain terms, not real. If what you are measuring is not accurate, any improvement on it is superficial. We told them the numbers would look worse before they looked true, because a broken count was teaching the ad system to go find more of the wrong people.
It went deeper than the ads. Four different forms across the site all fed an unknown mix of tools, and on the call the owners could not say which form fed which system. The customer records were run like a notepad: deals not linked to the people who bought, no pipeline, no record of where a lead came from. And the whole stack, the forms, the pages, the ad accounts, sat inside their old vendor’s accounts. They owned none of it.
Then we watched a recording of a real sales call with a large prospect. It went well until minute 35, when price came up. The moment money entered the room, before the buyer had agreed the problem was worth solving, the call fell apart. That was not a price problem or a product problem. The solution and the number had been shown too early, to a buyer who was not yet sold the pain was worth paying to fix.
What we decided, and what we chose not to do
The owners had already given a hard no to an expensive website rebuild. We did not argue. The site was old, but a new one was not going to make the phone ring, and spending on it then would have been spending on the wrong thing. That no was correct, so we let it stand.
What we did instead was work in order. Fix the tracking and take ownership of the tools first, then rebuild how leads were captured and routed, then fix the sales conversation. The rule we held to was earn the right to the big spend: prove the work shows up in the bank account, not the dashboard, before asking for another dollar. A few other tactics got proposed and killed along the way, and we let those noes stand too.
What we built
On tracking and ownership, we moved the accounts into the client’s own hands: analytics, search console, the tag manager, the automation tool. We rewired the conversions so one fired only on a real, sent submission, not on a click. We stood up an owned ad account and pixel and migrated the landing pages onto a subdomain the client controlled. If we ever got fired, they would keep everything.
On lead capture, we replaced the identical form on every page with one that branched by intent: book a consultation, send me information, or just looking. Each path created a deal automatically and recorded the exact page the lead came from. Low-quality sources went into their own stage so the data could tell the truth, and cold contacts stayed out of the customer records until they raised a hand, so it did not fill up with junk.
On the sales conversation, we resequenced it. The old order revealed the program and quoted the price first. The new order put the diagnosis first: validate that the pain is real and worth solving before showing the solution or the number. We graded the pain with the buyer, separating what they would trip over but get past from what actually stops them. We reframed budget from what they are willing to invest to what they are willing to lose. And we anchored price as an outcome range only after the problem was agreed. As one line in the rebuilt script put it, a doctor does not tell you the cost before they have looked at what is wrong.
What changed
We want to be straight about what this shows. These are figures from inside the engagement, not an audited result, and the early 66 percent number was flagged as superficial from the start, because the tracking behind it was broken.
With the tracking honest, the picture changed. By the scaling phase, cost per deal had roughly halved month over month, from about $320 to about $160. The scheduled-meeting rate was up about 300 percent, and leads were up about 26 percent month over month, alongside a cleaned-up fraud picture where a large share of one month’s leads turned out to be junk.
There is a fuller test of it, too. The company hit a scary stretch where leads dipped while it was spending about $40,000 a month, and the owners were ready to cut back. Instead of walking, they kept going, because the behavior data relocated the panic. Visitors were arriving and staying on the page for about 51 seconds on average, reading and comparing, not bouncing. The market had not dried up. In that same window the company reported landing about $150,000 in the prior two weeks. The pipeline was never empty. It had just never been measured honestly.
The lesson
A profitable, owner-led business will almost always name a tactic as its problem. We need more leads. Our search ranking is weak. Our salesperson keeps losing deals. It is almost never the tactic.
The first job is to make the numbers honest, because a business that cannot see what actually turns into revenue is not having a marketing problem. It is having a measurement problem wearing a marketing costume. Once the count was real here, the felt problem moved to where it had been hiding: a tracking problem, an ownership problem, and a sales conversation that showed its hand too early.
If your own dashboard looks good and you cannot quite say which of it is real, that is the thread worth pulling. Pull it, and the thing you named as the problem is almost never the thing you find at the other end.