Every agency report in Malta leads with the same two numbers: click-through rate and how many people watched the video. I wanted to know whether either of them had ever predicted a client for us. So I took every lead generation ad The Growth Bully has run for its own services since 2025, forty creatives, video and static, and matched every lead back to our CRM by the ad that produced it. This is the same audit I now run on client accounts, and I am publishing our own numbers because they are the only ones I can show in full.
The short version: click-through rate had a correlation of 0.04 with leads per euro. Hook rate was slightly negative. The numbers that predicted clients were not on the ads dashboard at all.
What did the audit actually measure?
For each of the forty creatives we pulled spend, leads, cost per lead, click-through rate, three-second hook rate, completion rate and CPM from Ads Manager. Then we did the part almost nobody does. Every contact in our CRM carries the identifier of the ad that captured it, so we could join the 394 platform leads to real contacts, follow each one through the pipeline, and count how many reached a qualified conversation and how many became clients. Seven did. Across the period the account spent a little under fifteen thousand euro, so the sample is small, but it is complete, and it is ours.
One honest caveat. For one summer a form change dropped the ad identifier, and 68 platform leads could only be matched to 8 contacts. Those weeks are excluded from the correlations. If your tracking has ever done that, you will recognise the feeling.
Did click-through rate predict leads?
No. The correlation between click-through rate and leads per euro was 0.04. Bucketed, the pattern was worse than random. Ads under 2% click-through were expensive. Ads between 2% and 3% held every winner. Ads above 4% were expensive again, because the extra clicks were curiosity clicks. People clicked to see what the fuss was about and left, and none of them booked anything.
Hook rate, the share of viewers who watch the first three seconds, correlated at minus 0.08. The nine creatives with hook rates of 12% and above cost EUR 78 a lead. The creatives sitting at a boring 5% to 8% cost EUR 29. Completion rate was the only attention metric pointing the right way, at 0.21, and even that is weak. CPM came in at minus 0.16.
The reason is simple once you see it. A hook that makes everyone watch makes the wrong people watch. The ad that closed a client for us had an ordinary hook rate. It said who it was for and what we guarantee, and the right person clicked.
What did the cheapest lead we ever bought produce?
Nothing. In February two video ads ran to the same audience with the same landing page. One produced leads at EUR 10, the other at EUR 9. On the report they were twins. The EUR 9 ad produced a client and a second deal behind it. The EUR 10 ad produced sixteen leads, zero conversations, and eight contacts marked not a fit in the CRM.
The difference was the first five seconds. The winner opened with a question and a guarantee. The loser opened by mocking the viewer, which earned the cheap click and the laugh and a list of people who were never going to buy. Cost per lead is not a number to scale on. It is a number to check against the CRM, which is the whole argument in our piece on what a qualified lead actually is.
What did predict clients?
Three things, none of them a platform metric.
The guarantee sentence
I transcribed every video. One sentence appeared in four of our seven closed deals and in none of the ads that failed: if after ninety days you do not have forty percent more qualified leads, you do not pay. No small print, no excuses. A number, a date, and who carries the risk. The ads that said "we guarantee results" produced nothing, which is a lesson in offer design more than copywriting.
The walkthrough ask
Ads that ended with "watch the walkthrough" produced leads at EUR 22, and five of the seven clients came through that door. Ads that ended with "book a call" cost EUR 54 a lead, to the same audience. Most people are not ready to book with a stranger, and when you ask them to, only the wrong ones say yes.
The CRM stage, not the lead count
The ad with the most leads was not the ad with the most clients. Judge on leads and you scale the wrong creative. Judge on qualified conversations and you scale the right one. That needs the ad identifier on every contact, which is one line of tracking most accounts are missing.
Why does a winning ad stop working?
A bonus finding, because it cost us money. One opening line, about what happens to next month's revenue if your top salesperson resigns and takes the contacts, produced leads at EUR 17 in April. Re-shot in June it cost EUR 66. In August, EUR 88. Malta is small, and the people who respond to an idea respond the first time. Re-running a winner is paying to show a question to people who already answered it. Keep the structure, change the opening.
What should you ask for in your own report?
If line one of your monthly report is reach, impressions, click-through rate or video views, ask for three numbers instead. Cost per lead where the lead exists in your CRM. How many of those reached a real stage, a booked call or a proposal. And the cost per one of those. If whoever runs your ads cannot produce them, they are reporting on the ad rather than on your business, and the place to start is where your leads actually die.
Then look at your last thirty leads and count how many reached a real conversation. That number is your marketing. If you want the same audit run on your account, with the CRM join and the correlation table, the pipeline scorecard is where we start. I write about the method in more depth at stephenellul.co.