The six, side by side
Each row below is a discipline, what it covers in practice, and the measure that tells you whether it worked. The measures matter more than the labels: every agency uses the same six words, and the useful question is what number moves.
| Discipline | What it covers | How it is measured |
|---|---|---|
| Full-stack SEO | Technical foundation, intent-first keyword architecture, content engine and authority work, run as one programme | Organic impressions, clicks and assisted pipeline in Search Console Google Search Console Help, performance report |
| AI SEO | Visibility across AI-driven surfaces, including Google AI Overviews and assistant answers | Citation share across a fixed prompt panel, reported as the prompts and the answers, not a score |
| Answer engine optimisation | Structuring content so a specific passage can be lifted as a direct answer, with supporting structured data where it reflects visible content Google Search Central, FAQ structured data | Presence as the extracted answer, and the share of target questions answered on the page |
| Generative engine optimisation | Being named and cited inside a generated answer, through quotations, cited statistics and entity clarity Aggarwal et al., GEO: Generative Engine Optimization, KDD 2024 | Whether the brand is named, and whether the citation links back |
| Technical SEO | Crawlability, indexation, rendering, Core Web Vitals, structured data, internal architecture Google Search Central, understanding Core Web Vitals | Indexed-to-submitted ratio, render-blocking issues resolved, Core Web Vitals field data |
| SEO content engine | Briefs mapped to buying intent, human-approved, AI-accelerated research, human-written argument | Pages ranking in the top 20 within weeks 5 to 7, and conversion from those pages |
Why these are one engagement rather than six
The four search-facing disciplines share almost all of their inputs. An assistant that cannot reach a page cannot cite it, which makes crawlability a GEO problem as much as a technical one. A page with no sourced figures is rarely quoted, which makes citation discipline a content problem rather than a schema problem. And an entity described differently on five surfaces is harder for any system to resolve, which makes consistency a prerequisite for all of it.
This is also why a separate AI SEO retainer, sold alongside an existing SEO retainer, usually produces duplicated work. The published study on generative engines found the largest gains came from adding cited sources and quotations to the content itself, with cited sources producing roughly a 115% improvement for a page at position five Aggarwal et al., GEO: Generative Engine Optimization, KDD 2024. That is an editorial change to a page the SEO retainer already owns.
Who each discipline is for, and who it is not for
- Technical SEO first, always, where a site is large, JavaScript-rendered, or has an indexed-to-submitted ratio well below one. Content published onto an unindexable site is a sunk cost Google Search Central, JavaScript SEO basics.
- The content engine where there is existing demand to capture. If nobody is searching for the problem the product solves, content will not create that demand and a different channel is the honest recommendation.
- AEO and GEO where buyers research in assistants before shortlisting, which is most B2B software and a growing share of professional services. Where buyers arrive through a map pack or a physical catchment, local work matters more.
- None of it where the business is pre-product, has no search demand, or cannot give a decision-maker roughly 90 minutes a week. That last constraint is the one that most often decides whether a programme works.
Limitations worth knowing before an engagement
Capacity is four founder reviews a month, and an engagement cannot be widened by adding people next week. Timelines are honest rather than flattering: a pipeline live in week two, first indexed pages around week three, first top-20 positions around weeks five to seven, and a compounding curve in months three to six, with assistant citations typically first appearing inside 30 to 45 days.
Outside B2B SaaS there is no measured result to point to. That is not a reason to avoid the work, but it is a reason a procurement process requiring three comparable references will not clear. Several agencies on our own published shortlists are a better fit in that case.
Frequently asked questions
Can I buy only technical SEO?
A technical-only engagement is possible where that is genuinely the bottleneck, and the free audit will say so if it is. What is not offered is a content retainer on a site that cannot be crawled properly.
Is AI SEO a separate retainer?
No. It is part of the same programme, because the work that produces assistant citations is mostly work the content and technical streams already own.
What is actually delivered each month?
Written deliverables: briefs, published pages, technical change specifications or shipped changes, and a measurement report covering both search and assistant citations. Not an activity log.
Do you do link building?
Not as a purchased product. Links that come from published research and from being cited are the ones that survive, and buying placements is against Google spam policies Google Search Central, spam policies for Google web search.
What happens to the work if the engagement ends?
Pages, briefs, maps, workflows and documentation stay with the client, on client infrastructure. There is no proprietary layer that stops working when the invoices do.