A blog post goes quietly viral overnight. You wake up, open your dashboard, and the traffic graph has a beautiful mountain on it. Ten thousand visitors. You feel great for about four minutes, until the obvious question lands: so what? Did any of them stick around? Will they come back? Did this actually help the business, or did you just rent a crowd for a day? Your analytics showed you the mountain in vivid detail and went completely silent on every question that mattered.
This is the thing nobody tells you when you install a tracker. Analytics is extraordinary at some questions and structurally incapable of answering others, and most bad decisions come from asking it the second kind. Let us draw the line clearly, because once you can see where the data stops, you get much better at using the part that is real.
What analytics is genuinely great at
Web analytics is a measurement instrument, and like any good instrument it is excellent within its range. Trust it on these:
- What happened. Which pages got viewed, how many times, in what order at the aggregate level. This is the bedrock, and it is reliable.
- Where people came from. Search, social, a newsletter, a link on another site, direct visits. Channel-level attribution is one of the most actionable things you own.
- Relative comparison. This post versus that post. This month versus last. Mobile versus desktop. Analytics is far more trustworthy at comparisons than at absolute counts, and comparisons are usually what you actually need.
- Trends over time. Is the line going up, flat, or down over weeks and months. Direction and slope are where the real signal lives.
- Spotting anomalies. A sudden spike or a cliff-drop is a flare telling you to go investigate. The number will not explain itself, but it points you at the right day.
Notice the shape of that list. Every item is about what and where and how much, measured in aggregate and read as a pattern. That is the instrument's range. Stay inside it and analytics is one of the most useful tools you have.
What analytics quietly cannot tell you
Now the blind spots, which are just as important and far less discussed.
It cannot tell you why
This is the big one. Your data can show that a page has a high bounce rate. It cannot tell you whether people bounced because the page was useless, or because it answered their question so completely and quickly that they had no reason to stay. Those are opposite outcomes that produce the identical metric. Analytics measures behavior, not motivation, and the gap between the two is where most misreadings happen.
Every number is a fact about what happened and a Rorschach test about why. The data is real; the story you hang on it is you.
It cannot measure what did not get counted
Your dashboard only knows about events that fired and reached the server. The person who read your article in a private tab, shared the link in a group chat, and sent five friends who each read it in their email app, that entire ripple can be nearly invisible. Word of mouth, screenshots, dark social, offline conversations: they move real value around and leave almost no trace in the graph. The map is not the territory, and a chunk of the territory was never mapped.
It cannot read the mind behind a click
Two people click the same button for opposite reasons, one delighted, one confused and looking for an exit. The click looks identical. Aggregate behavioral data smooths away all of that individual intent, which is exactly what makes it privacy-respecting and also exactly what makes it mute on motivation.
It cannot promise causation
Traffic went up the week you changed your homepage. Did the change cause it? Maybe. Or a big account shared you, or a holiday shifted behavior, or a competitor stumbled, or it was ordinary noise. Analytics shows correlation in time. It almost never proves cause on its own, and confusing the two is the single most expensive mistake teams make with their data.
The vanity-metric trap
All of this feeds one recurring failure: optimizing for numbers that go up without checking whether anything that matters went up with them. Raw pageviews are the classic offender. They feel like success, and they are trivial to inflate with a clickbait headline that draws a crowd who bounce in three seconds and never return. The graph looks fantastic. The business is no better off.
The fix is not to ignore the numbers. It is to pair every metric with a "so what." A useful metric connects, however loosely, to something you actually care about: return visits, time reading the pieces you worked hardest on, traffic to the pages that lead somewhere, subscribers, replies, a growing base of people who come back on purpose. When you catch yourself celebrating a number, ask what it would look like if that number went up while the underlying thing you want got worse. If that scenario is easy to imagine, you have found a vanity metric.
How to actually decide from aggregate data
So how do you make good calls with a tool that is confident about what and silent about why? A few habits do most of the work.
Use analytics to find the question, not the answer
The graph is a great flashlight and a poor judge. Let it point you at the surprising thing, the page that is quietly outperforming, the channel that dropped, the spike on Tuesday. Then go find out why through means analytics cannot provide: read the page yourself, look at what got shared, ask a customer, run a small test. Data raises its hand; you go do the interview.
Favor trends and comparisons over single numbers
Any one figure is noisy and easy to over-read. "Direction over four weeks" and "this against a fair comparison" are far more robust than "we got 8,412 visits." Ask whether things are getting better or worse and relative to what, and you will be wrong far less often.
Triangulate before you commit
No single metric should carry a real decision alone. When traffic, engagement on the pages that matter, and some downstream signal like return visits or sign-ups all lean the same way, you have something worth acting on. When they disagree, that disagreement is itself the finding, and it usually means the simple story you were about to tell is wrong.
Match the tool to the honest question
If your real questions are "what is working, where is my audience coming from, and which way are things trending," you want analytics that answers those cleanly without drowning you in configuration or padding the view with numbers you cannot act on. That clarity is the whole idea behind Gabden Analytics: privacy-first, cookieless, and deliberately focused on the handful of things aggregate data can genuinely tell you, with a free tier and thirty days of history to get a feel for it.
Analytics will never tell you why someone loved your writing, why a customer hesitated, or what your audience whispered about you in a group chat it could not see. That is fine. It was never that instrument. What it can do, honestly and well, is show you what happened, where it came from, and which way the line is moving, and hand you the sharp questions worth chasing down. Use it for that, stay humble about the rest, and you will make better decisions than the person with ten times the dashboards and no idea which ones are lying to them.
Gabden Analytics is built around this honest view of the data. Next, trim the noise with a minimalist guide to metrics that matter, or see why your numbers will not match Google. Get started free.
Join the discussion