Recommendation Marketing works by governing five things:
- the buyer questions a business must answer,
- the evidence behind those answers,
- the sources that shape them,
- the boundaries of an honest recommendation,
- and the accountability that connects specialized work to one buyer decision.
That is the direct answer. The better explanation came from this newsletter’s readers.
Across ten issues, readers have pushed the discipline past its definitions. They have asked where the fix belongs, what counts as evidence, how measurement survives model drift, and who owns the result when several specialists touch one buyer journey.
Five of those questions deserve fuller answers.
One reader contribution also changes the method itself.
TL;DR
- Recommendation Marketing is not a content tactic. It is a discipline for becoming easier for buyers and AI systems to understand, verify, and recommend for the right reasons.
- Recommendation Measurement cannot eliminate model drift. It can characterize normal variability, require sustained movement beyond it, and compare how touched and untouched questions move.
- Your own website controls clarity. Third-party sources supply corroboration. The citations reveal which problem you have.
- Named leaders and specialists can be evidence when their expertise, authorship, history, and outcomes are specific and verifiable.
- One owner must govern the buyer-question set across every channel and specialist.
- A Frozen Question Set should include negative-space questions: questions the business should not win.
How Do You Separate Real Movement From Model Drift?
“Once you declare the questions and measurement period, how do you separate true brand visibility shifts from organic model update fluctuations?”
- Lucy Cox
You cannot fully separate them.
AI systems update on their own schedules. Retrieval changes. Sources age. Any single observation can contain both a change in your standing and a change in the system observing it. An honest method begins by admitting that limitation rather than pretending it can be engineered away.
Recommendation Measurement bounds the problem through three disciplines.
- The envelope. Before improvement work begins, the baseline characterizes normal variability across the Frozen Question Set. Movement inside that envelope is not claimed as progress.
- The threshold. Movement becomes meaningful only when it exceeds the normal envelope and remains visible across repeated observations. One favorable answer is not a result.
- The signature. Evidence work should affect the questions that evidence touches while other questions remain comparatively stable. A system-wide shift across touched and untouched questions is more consistent with drift than improved standing.

Date the record and flag known platform changes, but do not claim certainty the evidence cannot support. The Recommendation Marketing principle is simple: measurement does not remove uncertainty. It governs what you are allowed to say despite it.
Is the Fix on Your Site or on the Sources AI Cites?
“Brand accuracy is huge. When you find it, is the fix usually on the brand’s own site or on the third-party pages the AI is pulling from?”
- Preston B.
Follow the citations. They tell you where the problem lives.
If an inaccurate answer cites your own pages, you have a clarity problem inside your control. Correct the facts, remove contradictions, strengthen the structure, and make the evidence easier to locate.
If the answer cites third-party pages, you have a corroboration problem. You cannot edit an independent source as if it were your website. You can correct what is legitimately correctable, publish stronger evidence, and earn enough independent agreement that systems have better material to retrieve.

The Recommendation Marketing principle: owned sources establish clarity; independent sources establish corroboration. A responsible diagnosis distinguishes the two before prescribing the fix.
Does Leadership Credibility Count as Evidence?
“If buyers are asking AI who can I trust, then hiring and leadership credibility matters earlier than the website. Are you seeing companies adjust their proof points, like leadership track record, for that first AI conversation?”
- Rick Kloete
Yes. People can be evidence, and many businesses make their expertise difficult to verify.
A named leader with a consistent professional identity, a specific track record, real authorship, and corroborating references gives buyers and AI systems something to check. “Seasoned leadership team” does not. It is an adjective without a record.
That does not mean a founder biography should become a parade of claims. It means the people responsible for the work should be legible: who they are, what they know, what they have done, and where the evidence lives.
The Recommendation Marketing principle: expertise becomes useful when it is attributable, specific, consistent, and verifiable.
Proof Reputation or Plain Language Which Matters Most?
“What do you think will matter most there: proof (case studies, data), brand reputation, or how consistently teams answer in plain language?”
- June Collins
The three do different jobs, and the order matters.
Plain language makes the business understandable. Proof makes its claims believable. Independent corroboration makes those claims durable when someone checks beyond the company’s own materials.
If forced to sequence them, plain language is the entry condition, proof is the deciding condition, and corroboration is the compounding condition. A business missing the first is hard to classify. Missing the second, it is hard to trust. Missing the third, its credibility remains fragile.
The Recommendation Marketing principle: clarity earns consideration; evidence earns belief; corroboration helps the recommendation survive scrutiny.
Who Owns the Outcome When Several Specialists Touch One Buyer Journey?
“How does Recommendation Marketing change accountability in practice when several specialists touch the same buyer journey?”
- The Folson Group
The specialists can keep their lanes. The lanes cannot remain the organizing idea.
One person or function must own the governed set of buyer questions the business needs to answer. Search, content, public relations, social, technical, and sales-enablement work should all be briefed against that set and evaluated by how each contribution improves the evidence available to the buyer’s evaluation.
The SEO firm still improves search performance. The PR firm still earns credible coverage. The content team still explains the business. But one question set connects the work, and one scorecard prevents channel reports from becoming substitutes for buyer progress.
You do not need a new commercial agreement to begin. Appoint an owner of the question set tomorrow.
The Recommendation Marketing principle: capabilities may be distributed, but accountability for the buyer’s evaluation must be integrated.
What These Five Questions Reveal About Recommendation Marketing
Together, the questions define the discipline more clearly than another abstract definition could.
- It is governed, not certain. It makes bounded claims instead of pretending AI systems are stable or controllable.
- It diagnoses before it prescribes. The source of an answer determines whether the work begins with clarity, evidence, or corroboration.
- It treats people as part of the evidence system. Expertise must be attributable and verifiable.
- It builds in sequence. Clarity, proof, and corroboration reinforce one another but cannot substitute for one another.
- It organizes specialists around the buyer’s questions. The buyer experiences one evaluation even when the work crosses many channels.
That is how Recommendation Marketing moves from an emerging idea to an operating discipline: not by claiming it explains everything, but by establishing better rules for what businesses should make clear, what they should prove, what they should measure, and who should own the result.
A Reader Contribution Now Part of the Method
“I’d want the baseline to include questions where we shouldn’t be recommended, too. If AI recommends Reveting for everything, it hasn’t understood our positioning.”
- Jessie Lizak
Jessie’s point improves the method, and I am adopting it into the question-set criteria with her name attached.
A Frozen Question Set should include negative-space questions: questions where a well-positioned business should not appear, or should appear only with the correct qualification. Showing up everywhere is not proof of strong positioning. It can be evidence that the market, and the systems synthesizing it, do not understand the boundary of what the business is for.

The questions you should not win may be the clearest test of whether your positioning is legible.
Ask the Next Question
The questions a market asks are intelligence. These five improved the discipline because they forced it to become more precise.
So ask the next one in the comments. If it touches your competitive strategy, send it directly. The questions that deserve full answers should receive them, and the people who improve the method should receive the credit.
Coming next: What is marketing actually for? Not the tactics or channels, but the underlying job. That first principle explains why Recommendation Marketing exists at all.
Frequently Asked Questions
What is Recommendation Marketing?
Recommendation Marketing is the discipline of making a business easier for buyers, and the AI systems assisting them, to understand, verify, appropriately recommend, and engage. It connects buyer questions, clear positioning, credible evidence, independent corroboration, and governed measurement.
How does Recommendation Measurement account for model drift?
It characterizes normal variability during a baseline, treats only sustained movement beyond that envelope as meaningful, and compares changes across touched and untouched questions in the Frozen Question Set. It reduces overclaiming; it does not create certainty.
Should a business fix its website or third-party citations first?
Follow the citations in the recorded answers. Your own pages reveal clarity problems you can correct directly. Independent sources reveal corroboration problems that usually require stronger evidence and time.
What is a negative-space question?
A negative-space question is one a well-positioned business should not win, or should answer only with an important qualifier. Including it in a Frozen Question Set tests whether the market and AI systems understand the boundary of the business’s positioning.
Who should own Recommendation Marketing?
One accountable owner should govern the buyer-question set and the shared measurement record. Specialists can own channels and capabilities, but their work should connect to the same buyer evaluation.
About this newsletter:
Recommendation Marketing is published by B Randall Willis, Founder of Right Angle. It documents an emerging discipline for a world in which buyers increasingly use AI systems to discover, evaluate, and compare businesses.