The number is real. It just answers the wrong question.
A monthly report that says "47 leads this month, up 12%" is not wrong. It is a true number, usually counted carefully. The problem is that it answers a question you were not really asking. You did not want to know how many forms were filled in. You wanted to know whether the money you spent came back, and that is a different number the report quietly does not contain.
This is not a story about a dishonest agency. Most report what they can measure, and lead volume is the easiest thing to measure. The gap is structural: the figure that is simplest to count is not the figure that decides next month's budget. So the report leads with the one it has, and the one you need stays off the page.
What a single number cannot tell you
A lead count, on its own, hides the three things that actually move a budget decision.
How many became customers. Forty-seven leads and four customers is a very different month from forty-seven leads and none, and the headline reads identically. Cost per lead rewards volume; cost per customer is the figure a budget should rest on, and it is the one missing.
Where they came from. "47 leads" is a total with no breakdown you can act on. If you cannot see which source produced the leads that became customers, you cannot move money toward it. And the breakdown is often missing for a real reason, not laziness, which is the next section.
What they did before they asked. A lead is the end of a journey: a source, some pages, a decision. The count is the last frame of that film with the rest cut. Without the path, you cannot tell a high-intent enquiry from a tyre-kicker until you have already spent the time.
Why the source is usually the missing piece
The reason the source breakdown is so often vague is the same reason your CRM is full of "Direct" and "Web": when a system cannot see where a visitor came from, it files them under a catch-all. And systems cannot see far more often than the report implies. In a controlled experiment where the true source was always known, SparkToro found analytics logged 100% of visits from TikTok, Slack, Discord, Mastodon and WhatsApp, plus around 75% of Facebook Messenger visits, as Direct with no referral data (SparkToro, 2023).
So a monthly report that groups leads by source is often summarising data that was already broken upstream. It is not that someone is hiding the answer. The answer was never captured, because referrers get stripped, campaign parameters vanish after the first page, and third-party cookies are blocked by default in modern browsers. The fix is not a better-looking report. It is capturing the source accurately in the first place, which is what lead attribution is about.
It is worth saying the wider context plainly, with the one number that is actually sourced: LinkedIn's B2B Institute, working with System1, found that 71% of B2B ads are unlikely to ever generate a sale (2022). When a large share of spend is working that hard against itself, a report that cannot tell you which share is the working share is not a small gap. It is the whole question.
The fix is a number you own
The point is not to distrust the report. It is to have something to check it against: a source of truth that sits on your own site, that you own, and that answers the questions the monthly summary cannot. First-party, independent of whoever sends the report, and readable without a translator.
With that in place, the monthly number stops being the only word on the subject. You can see which source started the leads that became customers, the path each one took, and the page where the rest quietly left. That is the full journey to every lead, and the skill of reading it is here. If your reports still show most leads as "Direct," the groundwork is cookieless tracking and what a lead source is. None of this replaces the agency. It just means you are both looking at the same, complete picture, instead of one tidy number.