How I work

The method, published in full.

This page is the whole audit method, written out step by step. It is the same sequence I run for every company, in the same order, whatever the product. You can read it before you hire me, question any part of it, or hand it to someone else and have them check my work.

  1. 1

    Map the buyers by clinical job

    I start from the clinical job, not the job title. Two people with the same title can be doing entirely different work, and two very different titles can be asking AI the same question. Each job is then crossed with the awareness stages, from a buyer describing a problem to a buyer raising an objection, so the map covers the whole route to a decision instead of only the end of it.

  2. 2

    Fix the query set before any capture

    The full list of questions is written and frozen before a single answer is collected, in the buyers' real words and in the languages they actually use. Fixing the set first is what stops the test drifting into confirming whatever I already believed. If a query is missing, it is missing for everyone, and it waits for the next quarter rather than being slipped in halfway through.

  3. 3

    Controlled capture across engines

    Every query is run in a fresh, logged out session, once per engine, so that no answer can contaminate the next one. Assistants carry context within a conversation and personalise to an account, and both of those quietly change the result. Controlling the capture is what makes the numbers comparable, this quarter and every quarter after it.

  4. 4

    Layered scoring

    Each answer is scored on three layers: was the solution category mentioned, was the product type mentioned, and was the brand named. A brand can drop out at any one of those layers, and each drop needs a different fix. A single yes or no hides which layer failed, and that is the one thing you need to know before spending anything.

  5. 5

    Source classification

    Every citation behind every answer is logged and sorted by source type, then split into two piles: the answers where the brand appeared, and the answers where it did not. The difference between those two piles is what explains the why. It is also what dictates the content plan, because it shows which kind of source you have to be in before the answer can change.

The method is published because a method you can inspect is worth more than a promise.

Work with me

Let's make your product impossible to miss.

Send me your product. I will show you where you are invisible in AI, and what it takes to fix it.

eva@evalorger.com

Eva Lorger, healthcare and medtech content and AI visibility.