How does the ReBo method work?

The ReBo method is a methodology of six delivery workstreams: AI Visibility diagnostic, technical optimization, data transformation, authority content, strategic distribution and continuous monitoring. All six are engaged together and run in the same order, because each one depends on the last. Progress is measured across the four AI Visibility levels, from Level 1 (Invisible to AI) to Level 4 (AI Reference).

What are the six workstreams of the ReBo method?

Each workstream solves a different problem, and none of them moves citations on its own. Without a diagnostic there is no target, without a technical base the content is never read, and without transformation the knowledge stays trapped. Without original content there is nothing to cite, without distribution the page earns no external signal, and without monitoring there is no proof of result. What varies from client to client is the sizing, volume, channels and timeline, not the composition.

Source: ReBo method, 2026.

WORKSTREAM 01What does the AI Visibility diagnostic deliver?

The first workstream measures the starting point before anything is published. We define the set of target questions the firm’s clients actually ask AI assistants, run that set through ChatGPT, Perplexity, Gemini and Claude, and record which answers name the firm, who shows up in its place, and what the models claim about it.

That question set is frozen and versioned: it is the same ruler re-applied every month. Alongside the baseline, we inventory the existing archive, investor letters, quarterly reports, theses, market commentary, in PDF, Word or anything else, and classify the firm on one of the four levels, with the plan to move up.

Nothing new has to be written in this workstream. The thinking is already done; the problem is the format it is trapped in.

WORKSTREAM 02What does technical optimization change on the site?

AI crawlers do not execute JavaScript. If the text is not in the initial HTML, it does not exist for the model. So the second workstream delivers static HTML with a semantic hierarchy: one H1, no skipped levels, most H2s and H3s phrased as questions. It also delivers structured data in JSON-LD, llms.txt, a sitemap, and a robots.txt that explicitly allows GPTBot, ClaudeBot, PerplexityBot, Googlebot, Bingbot and the rest.

The acceptance test is "view source": if the text is not there, the page is not done. And publishing robots.txt is not enough, if the CDN or WAF returns 403 to those user agents, none of it works. That verification is part of the delivery, and it is the number-one cause of silent invisibility. More on why AI models can’t read PDFs.

WORKSTREAM 03How does content trapped in PDF become a citable page?

The third workstream is the transformation. Each document is broken into semantic blocks: a heading hierarchy, a 40–80 word executive summary, numbered theses, an FAQ built around the questions allocators actually ask, and structured data in JSON-LD. The denser the facts, the more citable the content, so every block keeps its figures, dates and named entities next to a declared source.

The result is a page with a permanent URL on the firm’s own domain, linked into the rest of the archive. The PDF is not replaced: it keeps going out to clients. The structured version exists for AI models and search engines. The same mechanics are applied to the historical archive, which stops being dead storage and becomes a library.

WORKSTREAM 04What authority content does ReBo produce?

The fourth workstream covers the questions the firm’s archive does not answer. These are articles and pillar pages built from the target questions that have no matching page yet, what the allocator asks AI before deciding, written in the vocabulary of the market.

Every piece is signed by a partner at the firm, not by the agency: named authorship, with verifiable credentials, marked up as Person in the JSON-LD. It is the tiebreaker models use when choosing between two equally readable sources, and it is what sustains the firm’s E-E-A-T.

WORKSTREAM 05How does strategic distribution work?

A good page on its own is rarely cited; a good page with a mesh of external signals pointing at it is. The fifth workstream produces the derivative pieces that circulate on LinkedIn, X, Instagram, Reddit and the newsletter, always with an anchor link back to the page on the firm’s domain.

The channels are not the asset; they are what makes retrieval systems judge the domain as a current, trustworthy source. The current calibration is around 25 published pieces per cycle, derived from one or two source pieces, on an editorial calendar set at the diagnostic.

WORKSTREAM 06How does continuous monitoring prove the result?

Every month, the same target-question set frozen at the diagnostic is re-tested on the same engines, in an anonymous session, with the date and engine on record. The report compares the result against the baseline on citation rate, share of voice, coverage by engine, accuracy and sentiment of what the model claims, AI referral traffic and crawler hits.

The sixth workstream closes the loop: where we still do not appear, why, and what the next cycle of production has to attack. It is what turns the work into a continuous operation rather than a one-off project.

What we do not promise: a guaranteed position, a deadline for citations, or a guaranteed citation in an AI answer. Nobody controls what a model says. We control the variables that move citations, and we measure their effect on the same ruler, every month.

How does hub-and-spoke work in practice?

The AI-friendly page is the hub and holds what you want cited. Channels are spokes and exist to point back to it. A page with no mesh of signal around it is rarely retrieved, and a post with no hub at the end takes the reader nowhere.

The hub is the only piece built to be read and cut up by a model: server-rendered HTML, one H1, H2s in question form and an anchor answer right below each one. The spoke has a different job, which is generating the external signal retrieval systems use to judge authority and freshness.

The operating rule that follows is easy to verify: no social piece goes live pointing at a page that has not been published or reworked yet.

One source, many assets: how does one piece of material become a multi-format package?

A parent piece, usually material the client already produced, becomes the hub page, the script for a 6 to 10 minute video with a published transcript, the posts for each channel and the newsletter edition. The input is the same across all of them.

The principle exists because the raw material already sits inside the client, produced by the people whose credentials matter. AI Visibility work repackages and redistributes that material into the formats each reader consumes, including the reader that is a machine. An end-to-end example, applied to an asset manager's letter, is in from PDF to AI-friendly page.

Why are the six workstreams inseparable?

Because a citation by an AI model is the product of a chain, and a chain is worth what its weakest link is worth. A technically flawless site with thin content is not cited. Excellent content in PDF is not read. A perfect page with no external signal is not retrieved. And any of the three, without a baseline and without measurement, is opinion, not evidence.

That is why ReBo does not sell a single workstream on its own. What varies per client is the sizing of each one, set at the diagnostic. See what we deliver in each workstream and the two engagement formats.

On which channels does a company need to publish for AI to cite the brand as a source?

Across four layers at once: the page on its own domain, which is the citable asset; the brand's own channels, which create repetition; third-party domains, which carry the most weight; and video with a published transcript. A single channel rarely sustains citation.

The layers are not worth the same, and the difference comes from who owns the domain. A mention on a third-party publication or site weighs most, because the source is not the brand itself. Communities and forums come next, with high retrieval weight and a real risk of backfiring when the presence is promotional. Owned channels weigh least and are what guarantee volume and consistency.

An Ahrefs measurement across 75,000 brands, published in 2026, indicates that brand mentions on the web correlate more with presence in AI Overviews than backlinks or domain rating. That is one reason distribution does not stop at the company's own site.

Where should you publish to get an AI to cite your brand?

The starting point is a page on your own domain, built to be retrieved. Then come LinkedIn, YouTube, a newsletter with a public URL and the communities where the sector talks. Order matters: a social post pointing at a page that does not exist yet wastes the signal.

Channel selection shifts by sector. In an asset manager, LinkedIn and the newsletter carry the operation. In a clinic, YouTube and Instagram come in and LinkedIn goes out. What does not change is the anchor link back to the hub, with UTM tags, on every published piece.

Does content in another language need its own strategy, or is translation enough?

It needs its own strategy. The question reaches the model in the local market's vocabulary, and a translated piece answers the question another market asks. Translation solves reach; citation in a given language depends on local questions, sources and examples.

There is a practical advantage here for anyone publishing outside English. The volume of structured content in Portuguese is still smaller than the English equivalent, which keeps the window open wider for whoever arrives now. In exchange, crawler checks matter more: a silent CDN block erases the whole source across every engine at once. See AI crawler.

How is progress measured?

By the four AI Visibility levels, which are the method’s ruler and not a second methodology. Level 1, Invisible to AI: models do not know the firm exists. Level 2, Partially Discoverable: they find fragments, without the depth to cite. Level 3, AI Friendly: they read, understand and begin to cite. Level 4, AI Reference: the firm is cited ahead of its competitors.

Running workstreams 01 to 03 is what reaches Level 3; Level 4 is the compound effect of workstreams 04 to 06 applied on a steady cadence. The criteria for each level live in the AI Visibility framework, and the full measurement protocol is in how AI Visibility is measured.

For the scope of each deliverable and the engagement formats, see services. To start with the diagnostic, book a call.