How to “become the business AI recommends” in 2026 and beyond

Type “best accountant near me” into ChatGPT, Gemini or Perplexity and you get three or four names with a sentence of justification each. The question every business owner now asks is obvious: how do I become one of those names?

The short answer, and the thesis of the RepOtz Framework manifesto (published in French, October 2026), is that nobody can force an AI assistant to recommend them. What a business can do is remove every reason for an assistant, or a person, to pass it over: be easy to find, easy to understand and worthy of trust, then measure and keep going. The manifesto calls this reputation optimization. This article summarizes it in plain terms and turns it into a plan you can start this quarter.

What changed: being found is no longer the finish line

For twenty years the question was “can people find my website?”. Search engine optimization (SEO) answered it with indexing, technical quality, content and relevance. That layer still matters. But the manifesto draws a line that most marketing advice skips: being shown in a result does not mean being chosen. People compare reviews, business details, public profiles, past work and how complaints were handled. A very visible business can still raise doubts, and a great experience nobody wrote about stays invisible to the next customer.

AI assistants add a third layer. They read sources, synthesize them, and sometimes name businesses. In the manifesto’s words, “it would be misleading to claim that a reputation mechanically commands an AI recommendation. The relevant question is rather: is the public information about an organization accurate, consistent, accessible and substantiated?”

The trend behind this is documented, with the caveats the manifesto insists on. Eurostat reports that 32.7% of EU residents aged 16 to 74 used generative AI tools in the three months before its 2025 survey. Ofcom found that 54% of UK adults said they use AI tools, up from 31% a year earlier. Pew Research Center reports that 49% of US adults have used an AI chatbot and 42% use one to look for information. Closest to our topic, BrightLocal found that 45% of US consumers had used an AI tool for local business recommendations in the previous twelve months. These surveys cover different populations, periods and questions. They cannot be added up or compared directly. They confirm a direction, not a universal rate.

Why “the business AI recommends” is the wrong target

An AI answer depends on the system, the wording of the question, the user’s location, the date and the sources the model could reach. Ask the same question twice and you may get two different lists. The manifesto is blunt about this: reputation optimization “does not allow anyone to control opinions, delete all criticism, or guarantee a rating, a revenue figure, a position in Google or a recommendation by an AI. The results are probabilistic, partial and subject to change.”

So the manifesto replaces the target with a word it defines carefully: recommendability, “the ability of an organization to assemble the elements that can justify an informed recommendation”. It is not an official score and not a promise that any engine or assistant will pick you. It is the state in which, whoever or whatever is doing the recommending, your business gives them every reason to include you and no reason to doubt you.

That reframing changes the work. Instead of chasing an algorithm, you fix what is actually yours to fix: the accuracy of your public information, the quality of the experience you deliver, the freedom your customers have to talk about it, the proof you make visible, and the discipline of measuring whether it changed anything.

What reputation optimization means

The manifesto’s working definition is worth quoting in full, because it is the backbone of everything that follows:

Reputation optimization is a strategic and continuous approach that aims to understand, measure and improve the conditions in which an organization is perceived, understood and judged trustworthy, based on its customer experience and its public information, in the digital environments used by people and by AI systems.

Three things in that sentence matter for a small business. It is continuous, not a campaign. It rests on customer experience as much as on public data, which means a reputation can start improving with an internal decision, like fixing a reply delay, before any communication happens. And it treats people and AI systems as two audiences reading the same public record, not as two separate projects.

The manifesto also separates three layers that are often confused: the discipline (reputation optimization), the method (the RepOtz Framework, six steps) and the tools (the RepOtz platform, which supports some of the steps). You can apply the discipline and the method with a spreadsheet and your own time. The reputation optimization page covers the discipline in more depth.

Six principles that rule out the shortcuts

Before the method, the manifesto sets principles. They are short, and each one closes a door that tempts a lot of businesses:

• Authenticity: never fabricate experiences or proof. No bought reviews, no invented testimonials.
• Fairness and respect: let every customer speak, protect their freedom and their privacy.
• Consistency: close the gap between what you promise and what customers live, and remove public contradictions.
• Transparency and verifiability: say where information comes from and what its limits are.
• Measurability and continuity: keep comparable reference points instead of one-off pushes with no follow-up.
• Proportionality and improvement: choose actions that fit your resources and risks, and change them when the results say so.

On the second principle, the manifesto is specific: a private channel to resolve a problem must never be used to keep an unhappy customer away from a public platform. Offering a private path is good practice. Reserving the public path for happy customers is not, and it breaks the policies of Google and other platforms.

The RepOtz Framework: six steps, one cycle

The RepOtz Framework turns the discipline into a loop: observe, act, observe again. The manifesto presents the steps in order to make them easier to learn, and notes that in daily practice they overlap. Here is each step with the question it answers and the work it implies.

1. Diagnose: what do you actually know about your reputation today?

Gather what is publicly visible: your Google Business Profile, your reviews, your website, your social profiles, a few authority signals, and what AI assistants answer when asked about you. Write down what you observed, what is missing and what remains uncertain. To get a dated starting point, RepOtz offers a free online reputation audit across six pillars (Google profile, reviews and sentiment, social media, AI visibility, website and SEO, trust and authority), scored out of 100. The manifesto is careful here too: the score reflects the tool’s own grid and a dated snapshot, not a universal measure of reputation. If you want to run your own quick test first, the two-minute AI test in our October article is a good place to start.

2. Structure: are your basics present, accurate and consistent?

Identity, address, phone, hours, categories, descriptions, website, public profiles, and who inside the company owns each of them. This step usually produces a checklist of unglamorous corrections: a stale opening time on Google, a different phone number in a directory, a profile that was never claimed. Some of this is done by hand on Google, in a directory or on your site; some of it by other tools. The manifesto notes that the method covers these tasks even when no software does them for you. It also includes designing the path a customer uses to give feedback and the path your team uses to reply.

3. Route: can every customer tell you, and tell others, how it went?

Routing organizes the ways a customer can share an experience. A private comment lets someone explain a problem and get an answer. A public review helps other people decide. The manifesto’s rule: both must be available to everyone, and neither may be conditioned on a favourable opinion. Solicitation must follow each platform’s policies, consumer law and the customer’s own choice. And a technical setup does not make a journey compliant by itself; how it really behaves has to be checked.

4. Amplify: is your quality observable, or only claimed?

Amplifying means encouraging the voluntary expression of authentic experiences and making real proof accessible: reviews given freely, testimonials with consent, documented work, useful information, and only the accreditations you actually hold. A small business publishing a case study with the client’s problem, what was done and the verifiable result has stronger proof than a page saying “best in town”. The manifesto’s line: “the goal is to make quality observable, not to manufacture an artificial consensus.”

5. Monitor: what are you measuring, and who owns each number?

Choose indicators before you act: reviews received, satisfaction trends, profile completeness, website improvements, inbound requests, and structured tests of AI visibility. Give each indicator an owner, a source, a collection frequency and a stated limit of interpretation. A swing can come from a platform change, a small sample or an outside event, so no decision should hang on one isolated figure. The manifesto also warns against reading a dashboard as surveillance of the whole web: no signal in one source does not mean no reputation.

For AI answers in particular, it asks for a protocol: record the model and version when known, the date, the country or language, the exact wording, the number of repetitions, whether web search was on, the sources cited and the results, including contradictory ones. Being named, being described favourably and being explicitly recommended are three different events. A good showing in three queries does not prove general visibility, and it does not prove that your reviews caused it.

6. Optimize: did the results change what you do?

Optimization begins when monitoring changes behaviour. Pick a correction, name a person, set a deadline, verify the effect. Drop what did not help, make what worked part of operations. Then the cycle returns to step one. A new audit is not there to produce another score; it checks whether the initial gaps were closed, whether new risks appeared and what comes next. The manifesto adds that a higher score alone is not enough: service quality and trust indicators have to be read alongside it.

What this looks like in one real diagnosis

The manifesto includes one anonymized example: a RepOtz report dated 28 October 2026 for a Montreal massage therapy business. Overall score 67 out of 100, with a perfect 15 out of 15 on website and SEO, 14 out of 15 on AI visibility, and 3 out of 15 on trust and authority. Two priorities came out as high impact: increase review volume and build case studies.

The point is not that 67 is good or bad. It is that a business can be technically excellent and almost invisible on proof at the same time, and that different pillars lead to different actions. The manifesto is explicit that this illustrates a diagnosis, not a commercial result: improvement will only be demonstrated after the work is done and measured again.

What none of this promises

This is the part most “get recommended by AI” content leaves out, and the part the manifesto spends a whole chapter on. Reputation optimization does not control opinions, cannot remove all criticism, and cannot guarantee a rating, a ranking, a revenue number or an AI recommendation. Audit tools can rely on incomplete or outdated data. AI models can confuse two businesses, cite the wrong address or name competitors that are not comparable. A score summarizes what its grid measures and can miss what is hard to quantify: a business with a thin online presence is not necessarily unreliable, and a flawless presence does not guarantee a good service.

If more requests come in after you fix your profiles and ask for reviews, that is encouraging. It is not, on its own, proof of cause. Season, advertising, word of mouth, competitors and algorithm changes all move at the same time. The manifesto recommends documenting the association and stating the limits, rather than promising guaranteed results. We hold ourselves to that standard too.

Where to start this month

1. Get a dated baseline. Run an audit or do your own inventory, and keep it. You will compare against it later.
2. Fix the contradictions. Same name, address, phone and hours everywhere. Claim what is unclaimed.
3. Open both doors. Make it easy for every customer to write to you privately and to review you publicly, with no sorting by mood.
4. Publish one real proof. A case study, a testimonial with consent, or a certification you hold, on your own site, visible to a human reader.
5. Pick three indicators and an owner for each. Decide now how often you will look and what would count as a change.

Then, next quarter, go back to step one. That loop, more than any single tactic, is what the manifesto means by becoming recommendable.

Frequently asked questions

Can a business guarantee that ChatGPT or Gemini will recommend it?

No. AI answers vary with the system, the question, the location, the date and the available sources. What a business controls is whether its public information is accurate, consistent, accessible and backed by proof. That raises the odds of being included; it never guarantees it.

Is reputation optimization just a new name for collecting reviews?

No. Reviews are one input. The discipline also covers the accuracy of public information, the quality of the experience itself, how complaints are resolved, the proof you make visible, and the measurement that tells you whether anything changed.

Do I need the RepOtz platform to apply the RepOtz Framework?

No. The manifesto separates the discipline, the method and the tools on purpose. The six steps can be run with your own time and a spreadsheet. The platform supports some of the steps, such as the audit and review collection; profile corrections, directory listings and content remain human work.

Is it acceptable to send only happy customers to Google?

No. Routing unhappy customers to a private channel while reserving public review requests for satisfied ones is selective solicitation. It goes against the manifesto’s principles and against the policies of Google and other platforms. Offer both paths to everyone.

Where can I read the full manifesto?

The manifesto is published in French as a free PDF, with no form to fill in, at repotz.com/manifeste. The edition is versioned and revised at least once a year.

Sources

• Hilaire Douanla, L’optimisation de la réputation à l’ère de l’intelligence artificielle. Fondements, principes et méthodologie du RepOtz Framework™, manifeste fondateur, édition 2026, version 1.1 (1 November 2026). repotz.com/manifeste. Quotations in this article are translated from the French.
• Eurostat (16 December 2025), “32.7% of EU people used generative AI tools in 2025”. ec.europa.eu
• Ofcom (2 April 2026), Adults’ Media Use and Attitudes 2026, 7,533 UK adults, fieldwork 29 September to 28 November 2025. ofcom.org.uk
• Pew Research Center (17 June 2026), Americans and AI 2026: Chatbots, Smart Devices and Views on Impact, 5,119 US adults, fieldwork 17 to 23 February 2026. pewresearch.org
• BrightLocal (10 March 2026), “Nearly Half of Consumers are Asking AI for Business Recommendations”, Local Consumer Review Survey 2026, 1,002 US adults. brightlocal.com
• Google, Maps user-generated content policy. support.google.com