Audit the plumbing before you touch the faucet

Most marketing problems are reporting problems wearing a campaign costume.

A lender asked me to fix their ads recently, so naturally the ads were the last thing I opened.

That sounds like a bit, and it is a little bit, but it's also the most useful habit I have. The request was reasonable. Clicks were down, leads felt thin, and the room had already settled on the story every team tells when a number drops. The creative got tired, the targeting drifted, and the agency had stopped trying.

That story might have been true. There was no way to know yet, because nobody had checked whether the reporting could tell a good month from a bad one.

The faucet is where everyone looks

When revenue softens, the visible parts of the machine get the blame first. Ads, landing pages, subject lines, the new SDR who seems a little quiet on calls. Those are the faucet. You can see the water come out, you can see that there's less of it, and it's very tempting to start unscrewing things, ideally in a meeting where leadership can watch.

The trouble is that every decision about the faucet sits on top of a decision about the plumbing that somebody made years ago and forgot about. How a lead gets tagged. Which system counts it. What happens to the person who calls instead of filling out the form. Whether a renewal shows up as new revenue or disappears into a spreadsheet the finance team keeps on a shared drive.

Nobody revisits those choices, because they were made once, they worked well enough at the time, and they don't show up on any dashboard as a choice. They show up as the number.

So when the number moves, the team argues about the faucet with total confidence, using a meter nobody has checked since the last reorg.

Three questions before you touch anything

Here's where I start on every engagement now, before I look at a single ad. None of these are clever. They're the questions that get skipped because they feel too basic to ask out loud in a meeting with a VP in it, which is exactly why nobody has.

1. 1

   Where does a lead get its source, and who decided that?

   Find the exact moment a record gets tagged with where it came from. Is it a form field, a UTM parameter, a rep picking from a dropdown at the end of a call? If it's the dropdown, go look at how often "Other" gets picked. That one tells you a lot about how the team feels about the dropdown.

2. 2

   What happens to the buyer who never touches the website?

   Phone calls, referrals, the person who emailed the founder directly because they met at a conference. In a lot of businesses these are the best leads in the building, and in a lot of CRMs they either arrive with no source at all or get credited to whatever the last click happened to be. If your highest-intent channel is invisible, every report is going to undervalue it, and the budget will slowly drift away from it.

3. 3

   How much spend comes back with no source attached?

   Pull the leads and the revenue that can't be traced to anything at all. Badly attributed records are a separate headache, so set those aside for now. Then compare that against the budget.

   At the lender, the answer to that last question was "enough to matter." A real share of the spend was going out the door and coming back as leads nobody could trace, which meant every conversation about which channel was working had been an argument about a spreadsheet with a hole in it.

   That's the moment the engagement changed shape. The ads were still going to get looked at, but they'd been demoted from prime suspect to witness.

Sometimes the gap is the finding

There's a reflex every analyst has, and I have it too. When the data can't answer a question, you estimate. You split the decline into the parts you think it's made of, you footnote the assumption, and you hand over a clean chart.

On the lender work, there was a decline in clicks that couldn't be cleanly broken apart into the people searching for the brand by name and the people searching for the category. The tooling just wasn't there. The natural move was to estimate the split and keep going, and it was suggested.

My answer was "No, that drives urgency."

That one surprised the room a little. Leaving a number unexplained felt like an unfinished deliverable. But an estimated split would have quietly told the client that the problem was understood, and it wasn't. The honest version was that the decline couldn't be decomposed with the infrastructure they had, and that sentence did more work than any chart could. It turned "your ads are down" into "you can't currently see why your ads are down," which is a different problem with a different owner and a much shorter list of next steps.

If you're making the case for fixing measurement, the places where you can't measure are the case. Papering over them with a careful estimate makes the report look better and the decision worse.

This is uncomfortable for people who are paid to have answers. I get it. I'm one of them, and charts look great on an invoice. But a confident wrong number travels further in an organization than an honest blank, and it gets harder to retract every week it sits in a board deck.

Where I learned this

The first time I ran into this was years ago at an agency, on an airline account. The argument at the time was whether social media was a revenue channel at all or just a place where the brand team spent money to feel current. Making that case meant you had to be able to count. If the conversion paths from social couldn't be traced, the channel looked like a cost center, and no amount of good creative was going to change that conversation.

The budget was big enough that a counting error was a real line item, which is a great way to learn that counting matters.

The lender confirmed something I'd suspected for a while, which is that this is the same problem at every size. A national airline and a regional lender both had teams arguing confidently about channels using numbers that couldn't support the argument. The airline just had more decimal places.

What it looks like when the plumbing is fine

To be fair to the faucet, sometimes the plumbing checks out. The sources are captured, the phone calls land in the right place, the unattributed bucket is small and boring. When that happens, great, you've spent a week confirming that the reports are trustworthy, and now every creative and targeting conversation that follows can actually be settled with data instead of seniority.

That's the part people miss about this order of operations. Checking the plumbing first doesn't delay the campaign work. It's what makes the campaign work mean anything. A better ad on broken measurement just produces more miscounted leads, faster, and the team gets more confident about the wrong channel.

The boring rule

So the rule I run is boring on purpose: audit the plumbing before you touch the faucet.

It's the least exciting week of any engagement. Nobody posts a screenshot of a lead-source field mapping. Nobody gets promoted for discovering that the phone system never talked to the CRM. But it decides whether the rest of the engagement is real or just very well formatted.

If your team is in a heated debate about creative right now, try one thing before the next meeting. Ask someone to pull up last month's best ten deals, record by record, and show you where each one came from. The dashboard can wait. If that takes more than ten minutes, or if three people give you three different answers, the argument is happening one floor too high.

Fix the faucet later. It's usually fine.