Be the answer,
not just a result
Search stopped being a list of links. KraftLinks combines classic SEO with generative and answer engine optimization, so assistants can find you, describe you correctly, and name you when a buyer asks.
The problem
The click is disappearing
For twenty years the job was to rank. A buyer typed a question, saw ten links, and chose one. Every part of that chain was visible to you: the impression, the position, the click, the session.
That chain now breaks at the first link. Assistants compose an answer, name two or three sources, and the buyer acts on it. If your brand is not in that paragraph, there is no impression to lose and no click to miss — you simply were not part of the conversation. It does not appear in your analytics as a drop. It appears as nothing at all.
The uncomfortable part: a model's willingness to name you has very little to do with how pretty your site is, and a great deal to do with whether it can identify what you are, state it without hedging, and find corroboration elsewhere.
Three disciplines
Ranked, quoted, and shown
These get bundled together and sold as one thing. They are not one thing. They share a foundation and then diverge.
Search Engine Optimization
Crawlability, index coverage, site architecture, internal linking, page speed, keyword and topic coverage, backlink authority. Nothing above this layer works if this layer is broken — a model cannot cite a page it cannot reach.
Generative Engine Optimization
Making your brand legible to a language model. Consistent entity naming across the web, structured and verifiable facts, presence in the sources models draw on, and content written so a machine can extract a claim without misquoting you.
Answer Engine Optimization
Structuring content so it can be lifted cleanly into AI Overviews, featured snippets and voice results. Explicit questions, direct answers in the first sentence, correct schema markup, and none of the throat-clearing that stops an engine from using a passage.
Why they run together
GEO without SEO is a brand a model likes but cannot verify. AEO without GEO wins a snippet for a question nobody connects to you. The foundation is shared, so splitting the work across vendors mostly buys you three invoices and a coordination problem.
Measurement
How we prove it moved
AI visibility has a reputation for being unmeasurable, which suits agencies that would rather not be measured. It is measurable — it just is not a single number you can pull from a dashboard.
At the start of an engagement we fix a prompt set: the actual questions your buyers ask, in their words, not keyword-tool phrasing. We run that set across the major assistants and record the full response each time — whether you appear, how you are described, who is named alongside you, and which sources the answer leans on.
That becomes your baseline. We re-run the identical set monthly. You get the transcripts, not a summary of the transcripts, so you can read what changed and disagree with our reading of it.
What lands in your inbox
- The prompt set, and why each prompt is in it
- Full response transcripts per assistant, per month
- Share of answer: how often you are named, and in what position
- Description accuracy: what the models get wrong about you
- Competitor co-mentions, and which sources drive them
- Classic SEO metrics alongside, so nothing is hidden
- The next month's priorities, ranked by expected impact
The work
What we actually change
A rough map of where effort goes. The mix shifts depending on what the audit finds — a site with excellent content and broken schema needs a very different month one than the reverse.
- Technical SEO: crawl, index, speed, Core Web Vitals
- Information architecture and internal linking
- Organization, Service, FAQ and Breadcrumb schema
- Entity consistency across every property you own
- Knowledge panel and business listing accuracy
- Answer-shaped content: claim first, evidence second
- Topic clusters that establish depth, not just coverage
- Citation placement on sources assistants actually read
- Review and directory footprint where it carries weight
- llms.txt, robots and crawler access for AI agents
- Monthly prompt-set re-measurement and reporting
- Conversion paths so the visibility turns into calls
Questions
Straight answers
What is generative engine optimization, in plain terms?
It is the work of making a language model able to describe your business correctly and willing to name it. A model does not rank pages when it writes an answer — it draws on what it has absorbed about entities and what it can retrieve at the moment. GEO means giving it consistent, verifiable, easily extracted facts about you, in enough places that it is confident enough to say your name.
Is GEO just SEO with a new label?
The foundation overlaps almost completely, and anyone selling GEO while ignoring your crawl errors is selling nothing. Above that foundation they diverge. SEO optimises a page for a position in a list. GEO optimises an entity for inclusion in a sentence. Different unit of work, different success condition.
How do you measure something this fuzzy?
With a fixed prompt set, re-run monthly, transcripts included. You see the exact wording each assistant produced, in which month, with which sources cited. It is more work than pulling a rank-tracker export, and it is the only honest way to show movement.
How long before anything changes?
Schema and technical fixes can change how a model describes you within a few weeks, because retrieval picks up structured facts quickly. Getting named more often is slower — that depends on citation footprint and authority, which take three to six months to shift durably. Anyone promising a faster durable result is guessing.
Do you guarantee we will appear in ChatGPT or AI Overviews?
No, and neither can anyone else. Nobody controls the output of a model they do not own. What we control is whether you are findable, correctly described, structurally quotable and corroborated elsewhere — which is the entire set of levers that exists. We will show you the movement either way.
Can this hurt our existing Google rankings?
The work is additive: cleaner technical foundations, better structured data, clearer content. Those are the same things classic SEO rewards. We report traditional rankings and organic traffic alongside AI metrics precisely so you can watch for any regression rather than take our word for it.
Want to see where you stand?
We will run your prompt set through the major assistants and send back what they currently say about you — free, and yours to keep.
Request my free audit