totaldesign.rs / ai-discovery-playbook / en
https ✓ indexed
SR EN
+381 69 202 27 17
PLAYBOOK · LLMO METHODOLOGY

How AI models find, understand, and cite your website

This is our internal methodology, open for anyone to read — five phases we go through when doing AI Discovery (LLMO) optimization, and a realistic view of what to expect if you hire us.

5 phasesaudit → measurement
~ 8–12 weeksto first measurable movement
1 reportmonthly, during tracking

A five-phase methodology

The phases run in this order because each one depends on findings from the last — we don't skip steps, even when a client is in a hurry.

01audit

AI visibility audit

First we check whether AI models can even access the content, and whether you're currently mentioned when someone asks a question in your field.

checking robots.txt rules for GPTBot, ClaudeBot, PerplexityBot, Google-Extended
checking for an existing llms.txt / llms-full.txt
testing 15–20 real prompts in ChatGPT, Perplexity, and Gemini — is the brand mentioned
checking structured data (JSON-LD) and baseline technical SEO
robots.txtllms.txtschema.orgprompt testing
02structure

Structuring content for machine understanding

AI models don't "read" a site the way a person does — they look for a clear answer near the top of the page, unambiguous definitions, and clearly named entities (who you are, what you do, who it's for).

a direct answer in the first 2–3 sentences of every key page
FAQ format for questions people actually ask AI models
clear naming of services, location, and area of work, with no ambiguity
FAQ structureentitiesclarity
03implementation

Technical implementation

Whatever the audit flagged as a blocker gets fixed here — including things a person would never notice, but a bot would.

creating/updating the llms.txt file
JSON-LD markup for the organization, services, and FAQ
load speed and mobile-friendliness (PageSpeed + Mobile-Friendly Test)
content accessible without relying on JavaScript rendering
llms.txtJSON-LDspeedmobile-first
04authority

Distribution and authority

AI models more often cite sources they already "know" from elsewhere on the internet — so part of the work happens off your own site.

presence on directories and platforms relevant to your industry
consistent brand data (name, location, services) across every source
this phase depends most heavily on industry and competition
E-E-A-T signalsconsistency
05measurement

Measurement and iteration

The same prompts we tested during the audit phase get repeated on a schedule — that's how we see whether and when the brand starts showing up in AI answers.

monthly re-run of test prompts in ChatGPT, Perplexity, Gemini
tracking classic Google rankings in parallel, for comparison
a report with concrete examples of where the brand appeared (or didn't)
trackingmonthly report

What you can realistically expect

No promises of guaranteed results — LLMO is a fairly new discipline and AI models change fast. These are rough timelines based on experience so far.

week 1–2

Audit findings

You get a clear picture: whether AI models mention you today, where the technical blockers are, and what has the highest priority.

week 2–4

Technical changes

Implementation of the llms.txt file, structured data, and core technical fixes. This is the most concrete and fastest part of the process.

week 4–8

First movement

Some prompts start returning your brand as a mention or a source — results vary from client to client at this stage.

week 8–12+

Stabilization and tracking

Ongoing testing shows the trend — whether visibility in AI answers holds and spreads to new prompts.

An honest warning No one can guarantee that a specific AI model will mention you for a specific question — models update constantly and don't disclose exactly how they choose sources. What we can do is remove the blockers, improve the odds, and measure that transparently, month by month.

What you get

/ audit-report

AI visibility audit report

An overview of the current state, the prompts we tested, and a prioritized list — before we change anything.

/ implementation

Completed technical changes

llms.txt, structured data, speed, and mobile-friendliness — documented so you know exactly what was done.

/ monthly-report

Monthly tracking

Repeated test prompts, concrete examples of mentions, and recommendations for the next step.

What working together looks like

step 1

Intro call

A short conversation about your site, your field, and where you show up today — or don't.

step 2

Proposal and scope

Based on that conversation, we propose which phases you need and a realistic schedule.

step 3

Working through the phases

We move through audit, implementation, and tracking with regular reports — no skipped steps.

NEXT STEP

Book an intro call and let's check your AI visibility

Call +381 69 202 27 17 Send an inquiry