Beta
Digital Cartography is in beta, and here is exactly what that means.
We would rather you knew the limits before you relied on the answer. This page lists what the product can do today, what it cannot do yet, and what we are building next. If anything here stops being true, it is a bug in this page and we want to hear about it.
What works today
How our diagnostic tiers operate
The Free Scan reads your site’s plain text to see which of your category’s search filters it states. No account needed.
The Paid Scan captures the exact sources retrieved for a category and suburb, then reports whether your business was cited or passed over. Findings are confirmed by hand on ChatGPT and Gemini before they count. It is built and in testing, its retrieval surface is being rebuilt, and it is not open to customers yet.
What you get is a diagnosis with its workings shown. Every finding names where it came from, and anything we have not confirmed on a live assistant is labelled as research rather than presented as a reason.
Known limitations
Six things to know before you rely on it
Coverage
We cover a small number of business types, in Perth
Our research is built one business type at a time, and what we know about cafes tells us nothing reliable about dentists. If your type is not covered yet, the scan will say so rather than guess. We work in Perth, Western Australia, and the sources an AI reads are local, so results elsewhere would not carry over.
The surface we ask
Most of our research is our own search, not a live ChatGPT answer
There are many AI assistants and they read different pages, so a finding is only ever true of the one we asked. Today most of our category research comes from our own web search alongside smaller models running locally. Where a finding has been confirmed directly on ChatGPT or Gemini, the scan says so. Where it has not, the scan says that too.
Evidence status
Some findings are confirmed, most are still research
Every attribute we check carries a status. Confirmed means we watched a named AI read a page and answer from it. Research means we found the pattern in retrieved pages but have not yet confirmed it on a live assistant, so we show it as a suggestion worth acting on, never as the reason you are being skipped.
What we will not claim
We can show you a gap. We cannot promise a result
We can show that an AI read a particular page, that the page lists businesses like yours, and that your name is not on it. What nobody has yet measured is how much being added moves the answer. Until we have run that test properly, we will not put a number on it, and you should not trust anyone who does.
Freshness
The pages an AI reads are often years old
Several of the roundups that AI assistants lean on were published a while ago and never updated. That is a real finding rather than a fault in the scan, but it means a business that opened recently can be missing from the AI's picture through no fault of its own, and no quick fix will change that.
Speed
A market we have not seen before takes a few minutes
If nobody has scanned your suburb and business type before, we have to go and read the market first. The page will tell you it is gathering rather than show you a half answer. Once a market is built, every later scan in it is instant.
On the way
What we are building next
In rough order of how much they would improve the answer you get. We have deliberately not put dates against these, because a date we miss is just another claim we could not back up.
- More business typesEach type needs its own research pass, because the pages an AI reads for a hair salon share almost nothing with the pages it reads for a cafe. We are working through them one at a time and adding each to the scan as it is ready.
- Checking directly against ChatGPT and GeminiThe strongest evidence we have comes from watching a named assistant answer and expanding its sources. Doing that at scale means paying those assistants for automated access. It is the next real upgrade to how much we can stand behind.
- Form availability verificationThe Free Scan automatically identifies applicable free directories and delivers their direct submit URLs. We are developing an automated verification step to continuously confirm these third-party forms are active. Currently 9 of our 20 catalogued directories have had their listing path confirmed by hand. The rest carry a tag saying so, rather than being presented as free.
- Re-checking over timeA single scan is a snapshot. Re-running it on a schedule would show whether anything moved after a change, which is also how we would honestly learn whether our own advice works.
- More suburbsThe research is per suburb as well as per business type, because a roundup of one suburb rarely mentions the businesses in the next one. We are widening across the Perth metro area.
Tell us when we are wrong
If you run a scan and the result does not match what you see when you ask an AI yourself, that is the single most useful thing you can send us. It is how the research improves, and being corrected costs us nothing compared to being quietly wrong.
Related reading: how we stay honest and what the paid scan adds.