Notes · The stack · June 15, 2026

No single tool tells you the whole story. So I stopped looking for one.

Every analytics platform sells you the same promise: this is the one that finally puts everything in a single place. None of them do it, because none can see the whole journey. They each see one slice, beautifully, and call it the picture. Once you accept that, you stop shopping for the perfect tool and start running a stack that covers the slices, and you learn which tool is for which job.

This is that reference. The tools I actually run, which hat each one is for, why I built my own dashboard over the top, and the build-versus-buy calls I made along the way. It is opinionated and it is dated the day I publish it, because this market moves every quarter. Use it as a map, not a gospel.

That's the trap. You're not short a tool, you're drowning in them, jumping between six in a single day, each built for a different stage and a different question. So the skill a solo operator actually needs isn't owning the right software. It's knowing which screen the moment in front of you needs, and that depends less on the stage than on the hat you happen to be wearing.

Wear all of these in a day and the tools stop being a list of products and start being a set of modes. Here's which one I open, and when.

Five tools that do the looking, two connectors that do the plumbing, and one dashboard I built to stop them disagreeing in different tabs.

The honest reason wasn't that the tools disagreed with each other. Unified dashboards already exist; Triple Whale is one. It was that none of them could answer the question I actually cared about: take a single Meta ad, and show me the real journey of the people who clicked it. Where they landed, what they clicked, where they fell out of the funnel, whether they bought, and what they went on to be worth.

Every off-the-shelf tool gives you one slice of that and stops. Triple Whale gives blended attribution. GA4 has the on-site behaviour but won't tie it cleanly to the specific ad. Lifetimely has lifetime value, but only site-wide, never split by campaign. Nobody joins the ad to the customer's actual steps to the conversion to the lifetime value. So for one of the brands I ran growth for, I built the thing that did: Meta campaign data, the session-by-session journey from GA4 filtered to that campaign, the Clarity replays, and LTV per campaign, in one view. One ad, the whole real journey behind it, click to repeat purchase.

Here's the part people get wrong when they fall in love with a unified view. Even with all of it on one screen, it's an attempt at the truth, not the truth. For any decision that actually matters I'm still drilling back into the underlying tools and cross-checking, because each one measures differently and each is wrong in its own particular way.

Think of it the way you should think about attribution windows. A seven-day click, a one-day view, and a last-click model will each tell you a different story about the same ad, and the operator's job isn't to pick the one that flatters the number. It's to look at several and triangulate. Dashboards are no different. A unified view tells you where to look. It doesn't get to tell you what's true. You confirm that by drilling in, every time.

Here's the part the all-in-one crowd gets wrong. You cannot read everything in the same frame of mind. At 8am in the ad accounts, killing yesterday's losers, I'm a media buyer and I need velocity. On a Sunday, looking at contribution margin against the quarter's forecast, I'm a CFO and I need to be nowhere near a flashing ROAS number, because it will pull me out of the only mode that matters right then.

So the journey data gets unified, because you have to read it together to read it at all. But the macro financial view stays its own room, on purpose. You don't do payroll in the break room.

These used to be separate jobs, and now they overlap, so don't pick by the feature list. Pick by the mode you live in most, and resist paying three vendors to tell you the same thing.

Every tool on that map moves a lever beautifully. Not one of them tells you which lever, or why, or when to stop. They show you what happened in extraordinary detail. They cannot tell you what it means, and they certainly cannot tell you that the thing you're staring at is the wrong thing.

That is the whole job, and it's the one part I never managed to automate. The stack gets me to the easy eighty percent faster than any team could. The read, the twenty percent that decides whether any of it makes money, is still a person looking at the same numbers everyone else can see and knowing which one actually matters. The tools got cheaper and better. That part got rarer and more valuable.