The Analytics Stack You Will Actually Use
The data you collect is not the data you use
Open almost any analytics account running for more than a year and you find the same pattern. Dozens of events fire. Someone configured goals, funnels, and custom properties. Underneath all of that instrumentation, the person who owns the account can usually name one, maybe two numbers they actually look at.
Everything else is collected and unused. It is worth being honest about the difference between a system that captures data and a practice that uses it. The first is a technical achievement. The second is a habit, and only the habit changes anything.
The fastest way to see this clearly is to ask a simple question of any setup: if it stopped working tomorrow, what decision would go unmade. For most of what gets tracked, the honest answer is none. That is not a reason to remove it all, only a reason to stop treating volume of data as the goal itself.
Why most dashboards die within a few weeks
Dashboards tend to follow a predictable arc. Someone builds one, often with real enthusiasm, arranging charts until the page looks like proof that the business is legible.
Then the excitement that built it fades, because building the dashboard was itself the satisfying part. Nobody scheduled a recurring moment to revisit it. Without that ritual, the dashboard quietly becomes wallpaper: technically live, practically ignored, until someone asks whether the tracking still works.
This is less a discipline failure than a design failure. Dashboards built around what is possible to display, rather than a specific recurring decision, do not have a natural reason to be revisited. The setups that survive share one trait: they were built to answer a question someone keeps asking, on a rhythm that matches how often that question comes up.
Match tool complexity to your decision cadence
Decision cadence is a simple idea: how often does new information change what you do. A solo operator who reviews their site once a month has a monthly decision cadence. A small team running weekly experiments has a weekly one. Almost nobody has a genuinely daily one, even though most tools default to a daily view.
Choose your instrumentation to match that cadence, not the other way around. If you check in monthly, a single trend line and one supporting number will tell you almost everything you need, and a real-time behavioral suite will mostly generate noise between visits. If your team reviews weekly with several people in the room, a shared dashboard with two or three agreed numbers earns its keep, because the habit already exists to carry it.
The mismatch we see most often runs in one direction: light habits paired with heavy tools. A powerful analytics suite installed by a team that reviews numbers twice a year is not rigor, it is just weight. The tool did not fail. It was scaled to an ambition rather than a rhythm, and rhythm determines whether anything gets read.
The honest privacy and attention cost of tracking everything
It is worth naming the costs plainly, since they rarely appear on a feature comparison. Every script you add is a small tax on page performance and on the attention of whoever keeps the definitions straight across tools. Every additional event tracked is a small claim on the privacy of the person visiting your site, whether or not they notice.
None of this means tracking is wrong. It means tracking should be a deliberate trade, not a default setting. For most people it is the more rigorous choice, chosen on purpose rather than installed because it came bundled with a template.
We would rather see a site running one honest, minimal measurement tool its owner actually trusts and checks, than a site carrying five tracking scripts that slow the page down, complicate every redesign, and get consulted twice a year.
Pick one core metric before you add a second
If you are starting from nothing, resist the instinct to instrument everything at once. Pick one number: the one that, if it moved, would change a decision you make. For a small content site that might be returning readers. For a simple product it might be the share of new signups who return a second time.
Install for that one number first. Build the habit of checking it on a schedule that matches your real decision cadence, not the schedule the tool suggests by default. Only once that habit is genuinely established, meaning you have checked it consistently for a real stretch of time and it has changed at least one decision, does it make sense to add a second metric.
This is slower than installing a full suite on day one. It is also the only approach we have seen reliably survive past the first month, because each addition earns its place by proving the first already works, rather than being added on the hope that more visibility will translate into more clarity.
What good enough measurement looks like at each stage
A solo maker with no team and infrequent review needs almost nothing: one core number, checked monthly, ideally from a calendar reminder rather than anxious habit. A small team with a standing weekly review can support a shared dashboard of two or three numbers everyone trusts and understands the same way. A growing team with someone whose job includes interpreting data can support real depth, because a person is accountable for keeping those definitions honest over time.
None of these stages is better or worse than the others. They describe different amounts of attention available for measurement, and the right setup respects that amount rather than aspiring past it. The mistake is not choosing a simple tool. It is choosing a tool sized for a stage of attention you have not reached yet, and quietly resenting it for the months it sits there mostly unopened.
| Approach | Setup effort | What it's good at | Real risk |
|---|---|---|---|
| A single core metric tracker | Low. Running within an afternoon: one number, one view. | Answering one recurring question clearly, and staying checked for years rather than weeks. | Too simple to catch a problem outside the one number you chose to watch. |
| A full behavioral analytics suite | High. Requires planning events, naming conventions, and ongoing upkeep. | Deep questions: funnels, cohorts, segment-level behavior, once someone is actually asking them. | Too complex to ever fully open again once the person who configured it moves on. |
| A privacy-first lightweight analytics tool | Low to moderate. Simple install, minimal configuration. | An honest, trustworthy trend view without a heavy privacy or performance cost. | Too limited for genuinely deep questions once your decisions require that depth. |
Do I need more than one analytics tool?
Usually no. The common failure mode is not too little data, it is two or three overlapping tools that each show a slightly different number for the same thing, so you trust none of them fully. One tool, checked consistently, beats several tools checked rarely. Add a second only for a specific unanswered question, not a general feeling that more coverage is safer.
Is a simple tool good enough or am I missing something?
It depends on the decisions you actually make from the data, not on how the tool compares to more powerful options. If your real decisions are simple, such as whether traffic is growing or whether a change helped, a simple trend tool answers all of it. If you are regularly asking harder questions, such as where people drop off, you are missing something, but only once those are questions you are actually asking, not ones you might ask someday.
How often should I actually look at analytics?
Tie the frequency to your decision cadence, not to how often the data updates. If you make one real decision a month, checking daily just adds noise and a low-grade anxiety about fluctuations that do not mean anything yet. Pick the rhythm that matches how often new information could change what you do, then hold to it deliberately, rather than checking whenever the tab happens to be open.
If any of this sounds like the setup you have been meaning to simplify, you do not need to do it all at once. Pick the one number that would change a decision, watch it for a month, and let everything else earn its place after that. We write occasionally about measurement done well, and the sign-up is on the home page, whenever you want it.