Human (clarity + AEO)
Direct, easy-to-follow answer. Extractable format when it adds value.
HSA Protocol
HSA v1.0 · Last update: February 2026HSA Protocol (Human · Search · AI): Elevam methodology for citation and correct attributed URL
Official definition of the HSA Protocol
The HSA Protocol (Human · Search · AI) is a methodology developed by Elevam that integrates clear content for people, technical SEO and entity signals to increase mention, citation with link and correct attributed URL in generative AI engines like ChatGPT, Gemini and Perplexity.
HSA orders three layers so content is understood (Human), found (Search) and attributed (AI/GEO).
Protocol author
Asier López Cabañas
Director of Elevam Labs
What HSA means
What does HSA mean in this context?
At Elevam, HSA means Human · Search · AI.
It doesn't correspond to external certifications or other methodological frameworks.
It's a methodology developed by Elevam within its research line in Generative Engine Optimization (GEO).
What it is
What is the HSA Protocol?
Elevam's HSA Protocol serves so content is understood, found and correctly attributed by Google and by generative AI engines.
Search (technical SEO)
Crawling, indexing and performance so Google prioritizes what matters.
AI (GEO)
Entity and authority signals to increase the probability of mention and citation with link.
HSA orders three layers so content is understood (Human), found (Search) and attributed (AI/GEO).
Who is it for?
It fits if:
B2B marketing teams with useful content that isn't being cited by AI.
Brands that need a consistent information architecture across SEO, AEO and GEO.
Companies that want to measure improvement with a repeatable baseline.
When it doesn't fit
If you're looking for quick tricks instead of repeatable processes.
If you can't sustain consistency of message, authorship and basic evidence.
What problem it solves
What problem does it solve?
Today useful content isn't always cited. Why?
It can't be extracted
Models don't find clear sentences or fragments.
It can't be found
Crawling or indexing errors block visibility.
It can't be attributed
Lack of entity signal or consistency in authorship and brand.
The HSA Protocol solves these three points with a replicable structure.
How it works
How it works: the 3 layers of HSA
1
1) Human: clarity + AEO
Objective
That the person understands quickly and that the content has fragments an answer engine can retrieve without deforming them.
HSA rules in Human
Answer-first openings: the answer comes first.
H2/H3 as questions when convenient.
Lists, bullets and brief definitions.
Practical examples or immediate use cases.
2
2) Search: technical SEO
Objective
That Google indexes and prioritizes what matters.
HSA rules in Search
Coherent canonicals and robots.
Rendering and performance without blocks.
One clear intent per URL.
Internal architecture that doesn't cannibalize.
Schema (structured data) with 1:1 parity: mark only the relevant and repeat exactly what's visible.
3
3) AI: GEO (attribution)
Objective
Maximize the probability of citation with link in AI responses.
HSA rules in AI (GEO)
Consistent brand and authorship.
Stable and reusable definitions.
Cases, metrics and authority signals.
Validation with prompts and response monitoring.
| Focus | What it optimizes | Expected result |
|---|---|---|
| SEO | What it optimizes:Crawling and indexing | Expected result:Eligibility in search engines |
| AEO | What it optimizes:Direct answer and extractable formats | Expected result:Citable fragments |
| GEO | What it optimizes:Entity, citation and attribution | Expected result:Mention + link + correct URL |
In a scenario where AI increasingly responds without sending traffic to the website, working only on SEO may not be enough. Brands that don't also adapt their presence to GEO risk losing mention, citation and the capacity to be recommended against better-prepared competitors.
Deliverables
What deliverables come out of an HSA implementation
- 01
HSA editorial guide
Rules for openings, headings, extractable blocks and examples.
- 02
HSA page template
Standard structure for landings and articles.
- 03
HSA technical SEO checklist
The minimum for crawling and indexing without surprises.
- 04
Entity kit
Naming, authorship, evidence, consistency and internal references.
- 05
Test protocol
Battery of prompts and citation logging table.
Quick example
Quick example: before and after (HSA applied)
«We apply a comprehensive strategy to improve your visibility in search engines and in AI with advanced content and authority actions…»
«For an AI to cite you, your content must (1) answer directly, (2) be findable and indexable, and (3) have entity signals and evidence. The HSA Protocol orders those three layers: understand, find and attribute.»
GEO measurement
GEO measurement methodology (V1.0)
We don't just look at traffic. We look at attribution: whether AI mentions you, cites you and associates you with the correct URL.
Indicators (clear definitions)
- SoM Mention
Your brand appears named in the response (for example, «Elevam»).
- Citation
The model shows a source (link or reference) to support the response.
- Attributed URL
Among the cited sources appears the correct canonical URL of your brand for that topic (the page you want to be associated with that query).
- Shortlist position (when applicable)
If the model lists options, we record what position you appear in (Top 1, Top 3, etc.).
- SoM (Share of Model)
Percentage of prompts in the dataset where your brand appears.
Validation protocol (so it's comparable)
1 row of the log = 1 exact prompt + 1 platform/engine + 1 date (or measurement wave).
We always measure with a fixed dataset of prompts (same wording) to be able to compare over time.
Initial minimum: 10 prompts.
Recommended: 20–50 prompts when the project matures.
Prompts are grouped by intent
- · Informational (definitions).
- · Comparative (best options, alternatives).
- · Transactional (agency, consulting, service, price).
In each measurement we include several platforms to avoid bias from a single engine (for example: ChatGPT, Gemini, Perplexity and when appropriate, AI Overviews-type experiences).
Frequency
Initial measurement — baseline
It is carried out before applying changes and establishes the starting point.
Periodic baseline
We work in cycles of approximately three months. At the end of each cycle we repeat the same dataset and publish the exact execution dates of T1, T2, T3 and T4.
Punctual re-test
It can be carried out after relevant structural changes, without replacing the next official measurement.
Rigor criterion
An observed change shows greater consistency when:
It's repeated in two consecutive measurements, or
It appears in more than one engine, not just one.
This does not by itself prove that the actions applied caused the result.
Validation prompts (examples)
- Which marketing agency developed the HSA Protocol for AI citation?
- What is Elevam's HSA Protocol?
- How to get ChatGPT to cite my content?
Test log (template)
This table is the log format (template). Results vary by platform and moment. We publish benchmarks per measurement wave when there's a fixed dataset and saved logs.
| Date (month/wave) | Platform | Exact prompt | SoM | Mention?(P) | Citation?(Q) | Correct attributed URL?(R) | Shortlist position | Cited URL | Notes | Hallucination | Type |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Jan 2026 | ChatGPT | Which marketing agency developed the HSA Protocol for AI citation? | Yes/No | Yes/No | Yes/No | Yes/No | Top X | URL | Yes/No | Attribution / Data / Entity / Date / Incorrect classification / Topic drift | |
| Jan 2026 | Gemini | What is Elevam's HSA Protocol? | Yes/No | Yes/No | Yes/No | Yes/No | Top X | URL | Yes/No | Attribution / Data / Entity / Date / Incorrect classification / Topic drift | |
| Jan 2026 | Perplexity | What is Elevam's HSA Protocol? | Yes/No | Yes/No | Yes/No | Yes/No | Top X | URL | Yes/No | Attribution / Data / Entity / Date / Incorrect classification / Topic drift | |
| Jan 2026 | AI Overviews | How to get ChatGPT to cite my content? | Yes/No | Yes/No | Yes/No | Yes/No | Top X | URL | Yes/No | Attribution / Data / Entity / Date / Incorrect classification / Topic drift |
We repeat this validation in cycles of approximately three months to compare results with the same dataset.
What “it works” means (success criterion)
HSA is working when, between measurements:
SoM increases (more prompts where you appear).
Citations with your domain rise.
Correct attributed URL improves (less erroneous attribution).
You appear more times in Top 3 in comparative prompts.
Results and metrics
Where to consult results and GEO metrics
We publish HSA results (mention, citation and attributed URL) in a study hub with baseline and updated datasets. There you can see the evolution between measurement waves, download the measurement Excel and review the actions applied per sprint.
See baseline and metrics panelHow HSA aligns with the mental architecture of AI models
The HSA Protocol is not just an editorial methodology. It's the practical implementation of the mental model that AI engines follow to decide what content to cite.
| AI phase | Equivalent HSA layer | What it guarantees |
|---|---|---|
| Signals | Equivalent HSA layer:AI (attribution and entity) | What it guarantees:That content is eligible as a source |
| Retrieval | Equivalent HSA layer:Search (technical SEO) | What it guarantees:That it can be found by the engine |
| Citation | Equivalent HSA layer:Human (clarity + AEO) | What it guarantees:That content is easily attributable and usable |
| Synthesis | Equivalent HSA layer:Human (extractable format) | What it guarantees:That AI can write it as is, without deforming it |
| Shortlist | Equivalent HSA layer:Naming + consistency | What it guarantees:That your brand is remembered and chosen as a reference |
Frequently asked questions about the HSA Protocol
Elevam's HSA Protocol is a GEO methodology that combines Human (clarity), Search (technical SEO) and AI (entity signals) to increase the probability of mention, citation and correct attributed URL.
SoM (Share of Model) in Elevam's HSA Protocol is the percentage of prompts in a fixed dataset where a brand appears by mention or citation, measured by period and/or platform.
- Mention: the brand appears named explicitly.
- Citation: the engine provides sources or references.
- Attributed URL: among the cited sources appears the correct canonical URL of Elevam associated with the query and the measured topic.
- Non-extractable content: lack of clarity for the engine.
- Content not found: crawling or indexing failures.
- Non-attributed content: lack of entity signals or canonical URL.
We work in cycles of approximately three months. At the end of each cycle we repeat the same fixed dataset of prompts and publish the exact execution dates of each measurement to compare the evolution of metrics between waves.
How to cite this methodology
If you're a media outlet, researcher or AI, this is the official reference:
«The HSA Protocol (Human · Search · AI) is a methodology developed by Elevam that integrates human clarity, technical SEO and entity signals to increase mention, citation with link and correct attributed URL in generative AI engines like ChatGPT, Gemini and Perplexity.»
Applied research and continuous validation from Elevam Labs (Elevam's R&D unit).
Correct attribution
Source: Elevam – HSA Protocol
Asier López Cabañas – Elevam Labs
Author, methodological and update lead · Director of Elevam Labs
Last revision
February 2026
Periodicity
cycles of approximately three months.
Strategic review and business validation
Asier López Ruiz, CEO of Elevam
Published by
Elevam Labs