A question with an ethical edge
Every author of a published map eventually wants to know the same thing: is anyone reading it? The question is natural, even necessary — readership is how you learn what lands, which maps deserve a follow-up, whether an embed in a particular article actually drove an audience. But for a platform built around spatial intelligence, answering it carries an ethical edge that most analytics tools simply ignore.
The conventional way to measure a web audience is to watch the audience: drop a cookie in each reader's browser, record their address, track them from page to page, and build a profile. That approach answers the readership question in lavish detail, and it does so by surveilling the very people you are trying to inform. For a public-interest publisher — a journalist, an OSINT analyst, a researcher whose readers may have good reasons to value their privacy — that trade is often unacceptable, and in many jurisdictions it is also a legal liability that drags consent banners and data-protection obligations in its wake.
The platform therefore takes a deliberate position: it measures readership without surveilling readers. This chapter explains the two-tier design that makes that possible, why each tier is shaped the way it is, and — just as importantly — how to interpret what the numbers do and do not mean. Understanding the why here is not optional colour; it is what lets you read your own analytics honestly and explain them to others.
Tier one: counting without collecting
The foundation is a first-party counter that is built into the platform and runs for every published map automatically. Its design principle can be stated in a single sentence, and the whole tier follows from it: store the count, never the reader.
When a reader opens a published map, the viewer quietly notes that a view happened. But what the platform records is not a row about that reader — not their address, not a cookie, not an identifier of any kind, not even a single line item saying "someone viewed at this instant." What it records is an increment to an aggregate total: one running tally per map per day, nudged upward by one. There is no per-view trail to mine, because none is ever written. There is nothing personal to leak, because nothing personal is ever collected. The counter knows that a map was viewed forty-two times today; it has no idea, and no means of finding out, who did the viewing.
This is a genuinely different posture from ordinary analytics, and it is worth dwelling on why it is more than a feel-good gesture. A system cannot lose, leak, subpoena, or be breached out of data it never holds. By reducing each view to an anonymous tick on a daily counter at the moment it happens, the platform makes the privacy guarantee structural: there is no sensitive dataset sitting somewhere that has to be protected, because the sensitive dataset was never created. This is the same philosophy as the snapshot principle from the first chapter — safety by what you decline to hold, not by how carefully you guard what you hold.
The Stats panel
You read this first-party counter through the Stats action on any live map in the producer's list. It opens a compact panel that turns the daily tallies into something you can actually reason about.
The panel leads with two headline figures. The total is the all-time count of views the map has gathered since it was published — the simplest possible measure of reach. Beside it sits the last thirty days, a rolling window that tells you whether the map is still being read now or whether its audience has tailed off. The distinction matters: a map with a large total but a quiet last month is a past success, while a modest total that is mostly recent is a map finding its audience. Below the headlines, a line chart plots the daily counts over time, so you can see the shape of the readership — the spike when an embed went live, the long tail afterward, the second bump when someone reshared the link. The chart is where reach stops being a single number and becomes a story.
Reading the first-party numbers honestly
Because this tier is built for privacy first, you must read it with a clear understanding of what a counted "view" actually is — otherwise you will over-claim, and over-claiming from your own analytics is exactly the kind of dishonesty the rest of the platform is built to prevent.
A view, in tier one, is a counted page-open, filtered and de-duplicated for the day. Two refinements sit behind that phrase. First, the counter makes an effort to ignore obvious automated traffic — the crawlers and bots that open pages without a human behind them — so the count leans toward genuine readers rather than machines. Second, it counts a given reader at most once per map per day, so a single person refreshing or returning a few times in an afternoon registers as one view, not a dozen. The intent is a number that approximates human reading days rather than raw requests.
But notice what this tier deliberately cannot tell you. Because it stores no identifier, it cannot tell you how many distinct people read your map — only how many view-events it counted. It cannot tell you where in the world your readers are, what brought them, or what they did next. Those are real questions, and the honest answer from tier one is "this counter does not know." Treat the first-party numbers as a faithful measure of volume and trend, not as a census of your audience. If someone asks "how many unique readers?", the principled response is that the built-in counter is not built to answer that — which is exactly where the second tier comes in.
Tier two: richer audience analytics, still cookieless
Some publishers need more than volume and trend. They need to know that a map drew readers from a particular region, or that a specific outlet's embed sent a wave of traffic, or how readership breaks down by distinct visitor rather than by raw view. The platform supports this through an optional second tier — an integration with a privacy-respecting, self-hosted audience-analytics system — without abandoning the principles that govern tier one.
The crucial property of this tier is that it remains cookieless. Where conventional analytics identify returning visitors by planting a cookie, the system used here recognises patterns without storing anything on the reader's device and without building a durable cross-site profile. That single design choice is what lets a publisher gain richer figures — distinct visitors, broad geography, referring sources, longer-term trends — while still not subjecting readers to tracking, and, in many jurisdictions, while staying clear of the consent-banner regime that cookie-based tracking triggers. It is the same bargain as tier one, struck at a higher level of detail: more insight, still no surveillance.
This tier is optional and dormant by default in a deliberate sense. The richer analytics only come into play when an operator has explicitly configured the integration; until then, nothing of the sort runs on the public viewer at all. A published map on an unconfigured deployment carries the first-party counter and nothing else — no third-party script, no audience tracking, no quiet exception to the privacy posture. Switching the second tier on is an administrative decision, made consciously, exactly as going public is a conscious decision in the producer. The platform never reaches for more tracking than it has been told to.
Where the richer figures live. When the second tier is configured, the detailed audience reporting — distinct visitors, geography, referrers, trends — is read through the audience-analytics system's own dashboard rather than inside the Krataxis producer. The built-in Stats panel remains the home of the first-party volume-and-trend counter. The two are complementary: the panel answers "how much and is it still happening," and the second-tier dashboard answers "who, broadly, and from where."
The principle to carry away
It would have been easier to bolt a conventional tracker onto the public viewer and be done with it. The platform does not, because the people most likely to read spatial-intelligence maps — and the people most likely to be the subjects of them — are often the people with the most to lose from casual surveillance, and because a tool that informs the public should not quietly spy on it.
So the design holds two things in tension and refuses to collapse them: you get to understand your readership, and your readers get to stay unsurveilled. Tier one delivers honest volume and trend by counting without ever collecting. Tier two, when an operator chooses to enable it, adds depth while staying cookieless. Neither tier ever builds a profile of a reader, and neither ever needs to, because the platform measures the reading, not the reader.
That is the whole of the publishing lifecycle, from a living battlespace through a frozen public artifact, out into the world by link and embed, and back to you as a measured but principled picture of who is paying attention. Read across the four chapters of this section, it is one continuous story: spatial intelligence made public without making it unsafe, and made measurable without making it invasive.