Field Notes 04 · SIGNAL

The silence

Your topic report ranks what customers said. The most expensive finding is usually the thing none of them mentioned.

Magicalia7 min read2026

How many electric-vehicle charging points does this company operate? It is a simple question, and its own public record gives four different answers.

The website said over four hundred. Press coverage said thirteen hundred. Corporate communications said three thousand five hundred. Its own interoperability announcement, made with a charging partner, said nine hundred and sixty.

And the public charge-point map that EV drivers actually navigate with — the one that determines whether a driver with 12% battery ever finds you — listed twenty-one.

So we went looking for what customers had to say about the charging. We read 372 reviews on the main public review platform, more than 760 app-store reviews, and the complaints filed on the national consumer portal. Mentions of charging: none. Not a complaint. Not a compliment. Not one.

The blind spot

Open any voice-of-customer platform and you will find a topic report. It is almost always a list, ranked by volume: here is what customers talk about, most to least. It is genuinely useful. It is also, in a very specific way, structurally incapable of showing you this finding.

A topic list is a ranking of things that occurred. A thing that occurred zero times has no row. It cannot be ranked, so it is not rendered, so it does not exist. The most consequential fact about this company's flagship investment was guaranteed to be invisible to the instrument it had bought to see its customers with.

A ranked list of what was said cannot, by construction, contain the thing that was never said. The report was working perfectly. That was the problem.

The principle

Four numbers, and the one that counts

Look again at those five figures, because they are not five estimates of the same quantity. Four of them are self-reported — the company describing its own asset to the market. One of them is observed from the demand side: what a driver, standing in a car park with a dying battery, can actually find.

That one is twenty-one.

The distance between three thousand five hundred claimed and twenty-one discoverable is not a rounding error or a data-hygiene issue. It is the finding. An asset that cannot be found does not exist to the customer, and the customer is the only party whose opinion converts into revenue.

Figure 4 — The same asset, five numbers

Four of these the company published about itself. One is what its customers can actually see. Any of the top four could be defended in a board meeting; all of them are probably true of something — points installed, points contracted, points in a joint network. But only the bottom bar describes the asset as it exists in the world, and it is smaller than all of them by two orders of magnitude. The number a buyer sees depends entirely on which page they land on. The number a customer sees does not.

The reframe

To find the silence you have to arrive with an expectation. You need a model of what the topic distribution should look like — given what this company sells, what it has invested in, what its competitors' customers talk about — and then you hunt for the difference between expected and observed.

Absence, once you are looking for it, resolves into three very different diagnoses, and they demand three opposite responses:

Invisible — customers do not know the capability exists, or cannot find it. That is a discovery and communication failure, and it is the cheapest of the three to fix. Unused — they know, and they are not using it. That is a product, pricing, or positioning failure. Frictionless — they use it constantly and it works so well it generates no comment. That is the best outcome in commerce and it looks, on a volume-ranked topic list, completely identical to the other two.

Here, the map settled it. Twenty-one discoverable stations against a claim of thousands is not seamlessness and it is not indifference. It is invisibility, and invisibility is the one that is cheap to fix — which is why finding the silence was worth more than any of the topics that made the list.

Figure 5 — What the report shows, and what it cannot

Every bar above the line is real, correct, and useless for this particular question. The finding is the dashed outline below it: a capability that should have generated conversation and generated none. The three explanations for that void — undiscoverable, unused, or so seamless it needs no comment — are separated by enormous strategic consequence, and a volume-ranked list cannot distinguish between them. It cannot even tell you to ask.

What this means for everyday decisions

Three consequences

Bring an expected distribution. If you only rank what you observed, you can only ever learn about things that already happened. Model what should be there, then look for the holes.
Separate self-reported from observed. When your own numbers disagree, the customer-side number is not one opinion among several. It is the only one the customer can act on.
Audit your strategic bets against the corpus. Take the three things your investor deck is proudest of. Search your customers' words for them. The result is occasionally a very bad morning, and always worth it.

The takeaway

The asset was real. Twenty-two-kilowatt chargers, competitive per-kilowatt-hour pricing, a bilateral interoperability deal, a public endorsement from the national EV industry association. None of that was ever in doubt.

What was missing was any evidence that a single customer had noticed — and the reporting the company had bought was, by its own construction, the one instrument that could never have told them.

The loudest thing in that dataset was a silence, and there was no row on the report where it could have appeared.

Method & scope

Grounding Drawn from a real Mode A diagnostic — built entirely from public signals: 372 reviews on a public review platform, 764+ app-store reviews, a national complaints portal, the public charge-point map, the company's own website, press coverage and a partnership announcement. No transcripts, no NDA, no client data.

Anonymised A European parking operator with an EV charging network. The operator is not named and identifying details have been generalised. The structural finding is ours to describe; the company's identity is not ours to publish.

Measured vs inferred The five point-counts and the review counts are measured from public sources at the time of the diagnostic. The reading of the silence as invisibility rather than seamlessness is an inference, supported by the discoverability gap — not a measurement.