Services

This is not statistical attribution modeling, and it is not marketing mix modeling. Both need more data and more budget than a company your size should be spending on the question, and anybody selling you one at this scale is selling you a number with a confidence interval they have no intention of showing you.

What you get instead is less clever and, at this size, a great deal more useful. Every system counts the same things the same way. So when you look at the same revenue through first touch and then through last touch and get two different answers, which you will, both are built from the same facts, and the meeting is about the gap instead of about whose spreadsheet is right.

What you do get is attribution you can check. If someone clicks a Google ad and fills in your form, the reporting tells you which campaign and which search term brought them, rather than filing it under "Google" and leaving you to guess.

It also handles the awkward details that quietly misfile leads. HubSpot, for instance, records where a lead came from a moment after the form is submitted rather than at the same time. Read it at the wrong moment and a lead your campaign paid for gets credited to direct traffic. Getting that right is unglamorous and it is most of the job.

And the numbers are counted rather than estimated. Every figure is built from records that exist, so when one looks wrong there is somewhere to go and find out why. That is the difference between a number you can defend in a meeting and a number you have to take on faith.

Nothing gets built until the questions are written down. We list what you already ask every month and cannot answer quickly, in your words, with the decision each one would change. That list becomes the scope, and it is what the finished work gets measured against.

Then the less comfortable half: checking whether the data needed to answer those questions is being collected at all. Sometimes it is. Often a form is not tracked, a conversion fires twice, campaign names have drifted so far that nothing groups cleanly, or the CRM has three fields that all mean "where did this come from" and disagree.

Connecting systems does not fix any of that. It just moves it into one place where it is easier to see. So you find out in week one rather than week four, and if collection has to be repaired before the reporting means anything, you get told what that costs before anything gets built.

HubSpot is where this work is most straightforward, because HubSpot already collects most of what you need and hands it over willingly. Other systems vary more than their vendors suggest. Salesforce with Pardot in particular keeps a lot of what matters in places that do not export cleanly, and assembling a full picture there involves more manual work than anybody would like.

So if that is your setup, say so on the first call. I will tell you what is achievable and what it will cost in effort, which is a better conversation to have in week zero than in week two.

Everything runs in your Google Cloud account, on your billing. There is no license fee to me and nothing is marked up through me. It is worth knowing the number before you start rather than discovering it on a statement.

Two things cost money: running the pipeline that pulls your sources on a schedule, and BigQuery storage and queries. The pipeline software itself is open source, so what you pay for is the compute it runs on. Data Studio is free.

For a business your size this is usually tens of dollars a month rather than hundreds. It scales with how many sources you have and how often they refresh. You get the figure for your setup in writing with the quote, before you commit to anything.

A warehouse nobody maintains is worth less in month six than it was the day it was built, because by then it is wrong in ways nobody has noticed. Platforms change how they hand over data. Google Analytics changes. Somebody renames a campaign and it quietly stops being counted. Your business adds a channel, or sales changes what it calls a qualified opportunity, and the reporting carries on answering last quarter's question.

That is the unglamorous half of what a retainer buys, and it is not optional.

Some companies do not need another dashboard. They need one person who is accountable for what gets measured: someone who decides which questions matter this quarter, who will tell a founder that the attribution model everybody likes is not telling the truth, and who can brief an agency or a contractor without being oversold to.

That is a different engagement to a buildout, and it is priced by conversation because it depends entirely on how much of your measurement thinking you want to hand over. Only a small number of these run at a time.

These engagements are scoped and priced around the measurement work. They do not include media buying, creative, public relations, search engine optimization as a service, event work, recruiting, or building your website. If a project needs one of those, I will tell you what good looks like so you can hire it properly, and then get back to the measurement.

There is a reason to be blunt about that. When reporting is a line item inside a media retainer, it gets whatever attention is left at the end of the month, and it is being produced by the same people whose performance it measures. Here it is the whole engagement.

If you run an agency, the same work is available behind your brand.