SEO helps search engines discover, process and display relevant pages. GEO focuses on how a brand and its content are represented and interpreted in generative answers. It is not simply a new name for SEO and does not replace its technical or content foundations.
How I work with GEO and SEO evidence and measurement.
Here I explain the evidence I use in GEO and SEO work, how I check findings and how I evaluate changes.
Discuss your briefGEO and SEO: a shared foundation, different evaluation
For AI, we examine answers, sources and factual accuracy on an agreed set of questions. For SEO, we use indexation, query and organic traffic data. Changes in AI answers alone do not prove revenue growth; rankings and citations cannot be guaranteed.
GEO (Generative Engine Optimization), AEO (Answer Engine Optimization), LLMO and AI SEO describe overlapping approaches, not four separate products. Here the job is external discoverability and accurate representation of your business, not internal AI automation or producing a quota of articles. This work suits businesses able to verify facts and implement changes; it is not a fit if the requirement is a guaranteed recommendation or a fixed number of citations.
Shared foundation, different measures
- SEO
- Indexation · queries · organic traffic
- GEO
- Answer · source · factual accuracy
The specific system selects sources; a website change does not control selection.
From available evidence to delivery priorities
Audit and evidence
I first check whether your existing audit or strategy is still useful and current. I build on what we can use, fill the gaps and propose a new analysis when evidence is missing. The review covers your website and current activities: technical health, content, consistent brand information, authorship, links and mentions, and available AI visibility data. I use competitor comparisons to identify relevant gaps rather than apply a generic checklist.
Priorities and ownership
Findings lead to agreement on priorities, ownership and measurement. For example, a discrepancy between a business profile and the website needs a source of truth, an owner for the correction and a check afterwards. This illustrates the method, not a client result. Scope follows the actual situation, rather than a fixed package.
Audit inputs and outputs
For the audit, bring the website, target markets and languages, customer questions, competitors and any authorised read-only exports. I assess availability to crawlers, entity clarity, evidence and whether important passages answer the question in context. The output is a prioritised roadmap with owners, dependencies and a proposed measurement protocol. Existing raw AI answers can be analysed; a protocol alone is not a measured baseline.
Evidence into a priority map
For a finding: what · why · owner · verification. Missing evidence is not zero.
Delivery with or without an in-house team
My involvement
With an in-house team, I take ownership of agreed initiatives and work alongside marketing, content and engineering.
Without an in-house team
Without an in-house team, I develop the strategy and begin the agreed work.
Agreed delivery scope
We define my delivery scope and suppliers’ responsibilities beforehand.
Your facts and source material
Your team confirms business facts, subject expertise, rights to source material and access.
Delivery owners and capacity limits
Development, content production and any PR distribution need named owners; unlimited delivery capacity or secured media coverage are not included.
Agree roles before delivery
With an internal teamAgreed initiatives alongside marketing, content and engineering
Without an internal teamMy delivery and specialist work within agreed scope
- Business facts and rights
- Confirmed by your team
- Implementing a change
- An owner agreed beforehand
- Review and acceptance
- Agreed evidence and responsibility
Methods chosen for the problem, not separate packages
- Technical SEO: accessibility to crawlers, indexation, speed and structured data
- Content: topics and queries, content gaps, site structure, metadata and internal links
- Authority: relevant links, digital PR and credible mentions, distinguishing paid placements from independent editorial coverage
- Evaluation: agreed metrics, data limitations and next priorities
Four lenses on one page
A direct answer in context
Documented source and authorship
Relevant related content
- Availability
- Can content be fetched and processed?
- Content
- Does it answer with evidence?
- Authority
- What does an independent source support?
- Evaluation
- How do we verify the specific change?
Make business information consistent and verifiable
Fact sheet and discrepancy register
The review brings names, services, contact details, About information, authorship and source evidence into a fact sheet and discrepancy register.
Source, date and owner
Important facts need a source, date and an owner responsible for updates. Website text, relevant external profiles and structured data should describe the same entity.
Independent evidence and rights
An owned profile is not independent corroboration. Original evidence, credible mentions and relevant digital PR support trust, within verified scope and with clarified rights.
No platform guarantee
Neither a Wikipedia page nor a Knowledge Panel is guaranteed. My role, source evidence and its limitations are described in the Gen Digital case study. The published SEO overview is not evidence of increased AI visibility.
Fact, provenance and ownership
- Fact
- Source · date · update owner
- Discrepancy
- What differs · where to correct · how to verify
Solid connection: owned representation. Dashed: independent provenance, not automatic corroboration.
Six answer surfaces, six separate observations
Observed surfaces
ChatGPT, Gemini Apps, Google AI Overviews, Google AI Mode, Perplexity and Microsoft Copilot are assessed separately.
Questions and sources
We define relevant questions and observed sources for each selected surface rather than pooling them into one ranking or assuming coverage of every product.
Recorded conditions
Answers depend on the product, retrieval mode, available sources, language, location and observation time.
Limits of control
Improving access, clarity and evidence does not control whether a platform retrieves, cites or recommends a page.
Source list versus citation
A link in a source list alone does not prove an inline citation.
Training and guarantees
Search visibility is not proof of model training, and citations and positions cannot be guaranteed.
Separate surfaces, separate observations
- ChatGPT
- Mode · language · location · time
- Answer / sources: record separately
- Gemini Apps
- Mode · language · location · time
- Answer / sources: record separately
- Google AI Overviews
- Mode · language · location · time
- Answer / sources: record separately
- Google AI Mode
- Mode · language · location · time
- Answer / sources: record separately
- Perplexity
- Mode · language · location · time
- Answer / sources: record separately
- Microsoft Copilot
- Mode · language · location · time
- Answer / sources: record separately
Citation ≠ mention ≠ recommendation. An API is not the consumer interface.
Measure answers, not a universal AI ranking
Question panel
We agree a question panel around discovery, comparisons, branded facts and buying decisions.
Recorded conditions
The protocol records the exact prompt, surface, model or version where available, language, location, search mode, date and repetitions.
Traceable sources
Raw answers and their sources must remain traceable; missing observations stay unknown.
Separate measures
Brand mentions, linked citations, recommendations and factual errors are different measures.
Denominators
Each share needs an explicit denominator and repeatable collection conditions.
SEO and business metrics
We evaluate SEO indexation, organic queries and traffic separately, then connect available AI referrals and enquiries to business goals without claiming that answer visibility proves revenue.
Collection scope
Live collection, monitoring frequency and any service commitment depend on a verified collection method, permissions, capacity and an agreed budget.
Consumer-interface limitation
An API response is not evidence of what a customer saw in a consumer interface; no automatic baseline or consumer-interface monitoring SLA is included by default.
The same protocol, repeated observations
- Time and condition version
- Raw answer and sources
- Mentions, citations, factual errors
- Time and condition version
- Raw answer and sources
- Mentions, citations, factual errors
Local discovery: accurate facts, relevant questions
For location-dependent decisions, the audit checks consistency of business name, address and phone details, Google Business Profile information and genuinely local customer scenarios. This is a scoped review, not a promise to manage a location network or manufacture near-identical city pages.
Consistent facts for a real location
- Website
- Name: verify consistency
- Address: verify consistency
- Phone: verify consistency
- Business profile
- Name: verify consistency
- Address: verify consistency
- Phone: verify consistency
- Other relevant sources
- Name: verify consistency
- Address: verify consistency
- Phone: verify consistency
Correct discrepancies at the source. Do not create fictitious locations.
Product data: audit before integration
For ecommerce, the review covers product detail pages, feeds and structured data for consistency, accessibility and evidence. Names, specifications, prices and availability should come from the same current source. Recommendations identify what needs correction and who must implement it. Feed engineering, merchant integrations or checkout inside an assistant require a separately agreed technical scope and suitable delivery resources; they are not included as a ready-made integration.
One product, consistent data
Mismatch → source of truth → correct representations → recheck. Integration follows an audit and agreement.
Scope and price follow the brief
Agreement before work starts
We agree scope, budget, handover and review frequency beforehand.
Mentoring is not a minimum
The four-hour starting scope for mentoring is not a minimum for this engagement.
Collaboration conditions
We start by mapping your customers’ journey: how they find you, what information they lack, which feedback has no one to receive it and where their experience could improve. I call this initial review screening. Ideally, we combine it with an audit and use the findings to guide agreed follow-on work. Scope, responsibilities, budget and billing are confirmed in our agreement. Submitting the form does not itself start paid work.
All outputs are handed over to your team in the agreed format. Further sharing requires Martin Bangho’s consent. Rights to third-party tools and components remain subject to their licences; handing over outputs does not transfer those rights. Rankings, AI citations and the timing of business results are not guaranteed.
- Scope: Audit and agreed follow-on GEO and SEO initiatives, direct delivery or collaboration with your in-house team.
- Price: scope and budget are confirmed before work starts.
- Who it is for: Businesses that need to connect AI and search visibility with concrete priorities and delivery.
Let’s discuss your project.
Tell me what your company is working on and where you need help. You do not need a finished brief.