I Chose a Scoreboard I Cannot See
The number I care about is how many real people near me get in touch because of something I published. I can't see that number. I picked it anyway, and I want to explain why that isn't as silly as it sounds.
The gap, stated plainly
My contact hierarchy is a text message first, then email. That's deliberate. It's how the people I want to work with actually get in touch, and it doesn't ask anybody to fill in a form. If your best enquiries arrive the same way, you have this problem too, whether or not you've named it.
A text message carries no information about what caused it. Somebody reads a post on a Tuesday, thinks about it for a week, and texts me. Nothing anywhere connects those two events. My analytics saw an anonymous session. My phone saw a number.
So the thing I'm trying to grow is invisible to every instrument I've built, and the instruments I've built measure things I've explicitly said I don't care much about.
What the instruments see instead
This is the part that turned out to be more interesting than the gap itself, and you can check all of it on your own site this afternoon.
Over one recent fortnight, my analytics counted 44 human visitors. My server logs counted 1,142 things asking for pages. Neither number is wrong. The first is people whose browsers ran JavaScript. The second is everything that touched the machine.
About 60% of that larger number was declared crawlers, meaning robots that say what they are. That share has been almost exactly flat since the summer, two independent windows five weeks apart, 59% and then 59.8%.
And the single most-requested address on my whole server wasn't a page at all. It was my own publishing job, checking every fifteen minutes whether a post was due. Fourteen percent of everything, roughly 1,300 requests, from me.
If I'd reported "traffic" without separating those, I'd have been reporting a scheduling script as an audience. Yours will have something equivalent in it, and it won't be labelled.
The two failures that look like each other
There's a version of this that's worse than not measuring, and I nearly published it.
One period showed a handful of requests for two page addresses that shouldn't exist. That looks like a broken link — like my own site was sending people somewhere real and getting a 404. I wrote it up as a defect.
It wasn't. Every one of those requests came from Googlebot. No human user agent anywhere near it. The site had never produced those addresses. A crawler had guessed them.
A robot hitting a dead page and a person hitting a dead page are completely different findings, and a count of requests cannot tell you which you have. Your dashboard will show you one number for both. I'd found a number, invented a cause, and the cause was wrong. What saved it was going back to the raw lines and looking at who asked.
Why keep a scoreboard you can't read
Three reasons, and I hold them in this order.
It keeps you honest about what the visible numbers mean. If your goal is "traffic," every one of those robots is good news. Because my goal is local people, a spike in server requests is probably just a new crawler finding me, and I don't get to feel good about it.
The invisible thing is measurable later, cheaply, if I start now. There's an obvious low-tech answer: ask people how they found me, and write it down by hand. It costs nothing and it isn't tracking anybody. I haven't started, which I'll say plainly rather than describe as a plan. Every month I don't is a month of "before" I can't get back.
Naming the gap is what makes the rest believable. If you're shown a chart and nobody tells you the chart can't see the actual goal, you're being invited to think the chart is the goal. Most reporting works that way, and it's why so many people distrust it without being able to say why.
The honest position
The instruments are good at telling you about machines and rough at telling you about people. The thing you're working toward is usually people. There's a real gap in the middle and I don't have a clever fix for it.
What I have instead is a number I write down every two weeks — Oregon-attributed visitors, currently nine — and a standing note about what it does and doesn't include.
It undercounts, because satellite internet and cell connections geolocate to the carrier's ground station rather than the town, and the coastal Oregon I'm trying to reach is exactly where people depend on those. My laptop browsers are kept out of the count deliberately. My phone isn't, yet, and at nine visitors one person's phone is not a rounding error.
Nine is nine. It isn't nine customers, it isn't nine locals for certain, and it's the honest state of a thing I've been measuring for about a month.
I'd rather show you that than a bigger number I can't defend. And if you take one thing from this, let it be the question rather than my answer: what does your scoreboard actually measure, and can anything you own see it?
If you want a hand working out what your own measurements can and can't see, text me at 503-664-0546 or email eog@ernestofgaia.xyz.
#ernestGoesToAI