Measuring AI visibility for field service software

The obvious page to write here would list eight field service vendors, put numbers beside them and rank them. Every number would be invented, because we have not run the questions.

The temptation is to borrow. We have real counts for five categories and it would be easy to reason that field service probably looks like dealership management, or probably looks like healthcare revenue cycle. The categories we have measured do not resemble each other closely enough for that to be safe. One had seven vendors sharing a crowded list. One had a single vendor taking a majority of every answer. One had a third of answers naming nobody at all. An estimate drawn across that spread would be a guess wearing a number.

What we have and have not run. The last row is the reason this page exists.

The method does not change by category. It is published in full at github.com/Broadcastwell/audit-method so a vendor can run it without commissioning anyone.

The five categories above were measured with the four-engine method under v1.0. A new sweep would run the five-engine method described on the methodology page.

The questions matter more than anything else in that list. They have to be the questions a service manager or an operations lead actually types, about dispatch, scheduling, mobile access offline, parts and inventory, contract and warranty handling, and integration with the systems they already run. Questions written in vendor language return vendor language and measure nothing.

This section is our reasoning rather than a result, and it is labelled that way deliberately.

Field service management sits close to categories where the citation layer has been dominated by vendor-owned domains, because there is limited independent analyst infrastructure and a lot of vendor-published operational content. If that holds, the lever would be the depth and specificity of a vendor's own documentation rather than third-party placement. It might not hold. That is what the measurement is for.

In healthcare revenue cycle management, 13 of 37 answers described selection criteria instead of naming a vendor. That observation belongs to that measured category. It does not establish the pattern in field service management or the cost of changing an answer.

It would tell you how often five engines named your company when a buyer asked a real question, how often your domain appeared in the cited sources, and which of the two is your actual problem. Those two numbers move independently, and we have measured a category where a domain was cited 25 times while the company behind it was named zero times in open questions.

It would not tell you whether your product is better, whether your technicians close more jobs, whether your pricing is right or whether a given buyer should choose you. No amount of AI answer scoring speaks to any of that, and a page that implies otherwise is selling.