What happens when you feed a lifetime of marketing theory into a large language model and ask it whether your brand strategy is any good? Evidenza thinks it knows the answer. The rest of us are still working out the question.
The Audacious Pitch
Somewhere in London, there exists a product that allows a mid-level brand manager at a B2B software company to book a strategy session with Byron Sharp, Mark Ritson, and Les Binet simultaneously. The session costs a fraction of what any of those individuals would charge for a single speaking engagement. It runs in minutes rather than months. And none of the three thinkers need to be in the room, or even aware it is happening.
This is Evidenza's Synthetic CMO product, and it is either the most interesting development in marketing technology this decade or the most elaborate exercise in intellectual karaoke ever attempted. Possibly both.
The concept is straightforward enough to describe and complex enough to interrogate. Evidenza has built AI representations of prominent marketing thinkers - trained on their published works, public frameworks, conference speeches, articles, and stated positions - and packaged them as interactive advisors. You present your brand strategy, positioning, or creative work, and a synthetic version of Sharp, Ritson, or Binet & Field critiques it through the lens of their known intellectual frameworks.
Peter Weinberg and Jon Lombardo, Evidenza's co-founders and former heads of the LinkedIn B2B Institute, have assembled an advisory board that lends the concept considerable credibility. Mark Ritson himself sits on it, alongside Linda Boff (former CMO of GE), Tomer Cohen (Chief Product Officer at LinkedIn), and Stefano Puntoni (Professor of Marketing at Wharton). These are not people who lend their names to frivolous ventures. Whatever one thinks of the Synthetic CMO concept, its backers have thought about it seriously.
The question worth asking is not whether this is clever - it plainly is - but whether cloning the advisor is the right thing to clone. That question has implications far beyond Evidenza, and it sits at the heart of an emerging divergence in synthetic research philosophy.
The Intellectual Lineage
To understand what Evidenza is attempting to replicate, one must first understand what each of these thinkers actually argues. The frameworks are not interchangeable, and the tensions between them are part of what makes marketing theory interesting.
Byron Sharp and the Laws of Growth
Sharp's contribution, distilled primarily in How Brands Grow (2010) and its sequel, rests on empirical generalisations about how brands actually grow versus how marketers believe they grow. The core argument is that brands grow primarily by acquiring new and light buyers rather than by deepening loyalty among existing ones. This leads to several operational principles: maximise mental and physical availability, invest in distinctive brand assets rather than differentiated positioning, target the entire category rather than narrow segments, and maintain continuous advertising rather than pulsed campaigns.
Sharp's framework has been validated across more than 50 product categories and multiple countries through the Ehrenberg-Bass Institute's research programme. The data set is enormous. The laws - the duplication of purchase law, the natural monopoly law, the double jeopardy law - are empirical regularities, not theoretical constructs. This matters for AI replication: the underlying logic is systematic and rules-based, which makes it, at least in principle, amenable to computational modelling.
Mark Ritson and Strategic Diagnosis
Ritson's contribution is different in kind. Where Sharp is a researcher who derives principles from data, Ritson is a strategist who synthesises principles into practice. His "Brand Strategy in Three Steps" framework - diagnosis, strategy, tactics - emphasises rigorous market research as the foundation for all subsequent decisions. His Mini MBA in Marketing, completed by over 10,000 marketers worldwide, is essentially a curriculum for how to think about brands before deciding what to do about them.
Ritson's intellectual value is not in a set of universal laws but in his capacity for contextual judgment. He is famous for his Marketing Week columns precisely because they apply frameworks to specific situations with specificity and (frequently savage) critique. Replicating the frameworks is possible. Replicating the judgment is harder. A synthetic Ritson that can recite the three-step model is merely a textbook. A synthetic Ritson that can look at your particular brand in your particular category and tell you that your segmentation is lazy and your targeting is delusional - that would be something.
Binet & Field and the Effectiveness Curve
Les Binet and Peter Field's contribution, principally through The Long and the Short of It (2013) and their subsequent IPA Databank analyses, established the empirical case for balancing long-term brand building with short-term activation. Their famous 60:40 rule - approximately 60% of budget to brand, 40% to activation - became one of the most cited figures in modern marketing, even as they themselves revised it to vary by category and competitive context in later work.
The framework's power lies in its simplicity and its empirical grounding: analysis of over 1,000 IPA effectiveness case studies spanning three decades. Like Sharp, Binet & Field's work is systematic and data-derived. Unlike Sharp, their recommendations are explicitly contingent on category and brand maturity, which introduces ambiguity that a computational model must navigate.
What "Trained On" Actually Means
The phrase "AI clones" does rather a lot of heavy lifting in Evidenza's marketing. It is worth being precise about what this means technically, because the gap between what the phrase implies and what the technology does is where the interesting questions live.
Evidenza's Synthetic CMOs are large language models that have been fine-tuned or prompted with the published corpus of each thinker's work. This includes books, articles, conference presentations, podcast transcripts, and publicly available interviews. The model learns to produce outputs that are stylistically and substantively consistent with how each thinker writes and argues. When you ask synthetic Byron Sharp about your brand's segmentation strategy, the model generates a response that reflects Sharp's known positions on segmentation - essentially, that it is less important than most marketers think.
What the model does not have access to is the private intellectual life of each thinker. It cannot replicate the unpublished data analyses Sharp has conducted at the Ehrenberg-Bass Institute. It cannot reproduce the specific client engagements Ritson has advised on. It cannot access the proprietary IPA data that underpins Binet & Field's effectiveness conclusions. The model knows what these thinkers have said publicly. It does not know what they think privately, what they would say about scenarios they have never addressed, or how their thinking has evolved in ways not yet published.
This distinction is not a criticism - it is a description of the fundamental boundary of any AI system trained on published output. And for many use cases, it is sufficient. If a brand manager wants to pressure-test their media budget allocation against the Binet & Field framework, a model that faithfully represents the published 60:40 principle (and its category-specific variations) delivers genuine value. The brand manager does not need the real Les Binet on speed-dial to benefit from the framework.
The trouble arrives at the edges. Marketing, like most applied disciplines, is full of questions that existing frameworks do not neatly answer. What should a DTC brand that sells exclusively through TikTok Shop allocate to brand building? How should a B2B SaaS company with a two-person marketing team prioritise between mental availability and physical availability when both require investment they cannot afford? What does the 60:40 split look like for a category that did not exist when the IPA data was collected?
These are the questions where the real Byron Sharp earns his speaking fee - not by reciting How Brands Grow, but by applying judgment formed through decades of research, debate, and observation to novel situations. A synthetic version, however well-trained, is interpolating from known positions rather than reasoning from first principles. The distinction matters more in some contexts than others, but it always matters.
The Philosophical Question: Can Frameworks Be Cloned?
There is something philosophically interesting about the attempt to replicate a thinker's intellectual framework in computational form. It forces a question that marketing practitioners rarely confront directly: to what extent is marketing expertise a set of transferable rules, and to what extent is it irreducibly contextual judgment?
Sharp would likely argue - and has, in various formulations - that the laws of growth are generalisable precisely because they are empirical regularities, not situational heuristics. If the double jeopardy law holds across 50 categories, then a computational model trained on the law should apply it as reliably as a human who has memorised the same data. The framework is the expertise, and if the framework can be encoded, then the expertise can be encoded.
Ritson would likely argue the opposite. His entire professional identity is built on the proposition that strategy is diagnosis before prescription, and that diagnosis requires the kind of contextual sensitivity that resists systematisation. The Mini MBA does not teach rules so much as it teaches a way of looking at markets. Whether that way of looking can be computationally replicated is, at minimum, an open question.
There is a parallel worth noting from another domain. In chess, computational engines long ago surpassed the strongest human players. But chess is a closed system with perfect information and finite states. Marketing is an open system with imperfect information and infinite states. The analogy breaks down precisely where it would be most useful - at the point of application to novel, ambiguous, real-world decisions.
None of which is to say that Synthetic CMOs are without value. A model that consistently applies Sharp's frameworks would outperform the median brand manager who has never read How Brands Grow. A model that structures strategic thinking along Ritson's three-step process would improve the quality of briefs at most organisations. The question is whether the improvement comes from the AI being a good clone of the thinker, or simply a good organiser of publicly available knowledge. The answer is probably the latter, and that is still worth paying for.
A Different Clone: The Customer Instead of the Advisor
At Ditto, we have taken a fundamentally different approach to the question of what synthetic research should replicate. Rather than cloning the advisor - the person who tells you what should work based on theory and framework - we clone the customer, the person who tells you what does work based on their actual preferences, needs, and behaviour.
This is not a philosophical disagreement for the sake of product differentiation. It reflects a genuine conviction about where the bottleneck sits in most marketing decisions.
The typical brand manager does not lack access to marketing theory. They have read Sharp, or at least read the summaries. They have attended webinars on Binet & Field's effectiveness research. They may have completed Ritson's Mini MBA. What they lack is reliable data about their specific customers in their specific market. They know, in principle, that they should maximise mental availability. They do not know which of four tagline options their target audience finds most memorable. They know that the 60:40 split is a useful starting point. They do not know whether their particular category and competitive context warrants 70:30 or 50:50.
Ditto's synthetic respondents are AI-generated personas calibrated to match real demographic, psychographic, and behavioural profiles. When a CPG brand asks 500 synthetic consumers to evaluate their packaging redesign, the responses reflect how people matching those profiles actually think about packaging - not how Byron Sharp thinks they should think about packaging. When a political campaign tests voter messaging across swing-state demographics, the responses reflect the concerns and language of those specific voter segments, not the theoretical framework of any single strategist.
The distinction matters operationally. An advisor-clone is best suited for strategic validation: "Is my brand strategy theoretically sound?" A customer-clone is best suited for tactical execution: "Which of these three executions will resonate with my target audience?" Both questions are important. They are not the same question.
To put it in Ritson's own framework: diagnosis requires understanding the market (the customer), strategy requires choosing where to play and how to win (potentially with advisory input), and tactics require testing specific executions against real preferences (the customer again). The advisor is most useful in the strategy phase. The customer is most useful in the diagnosis and tactics phases. If you had to choose one, you would probably choose the customer, because diagnosis and tactics together represent roughly two-thirds of the process.
But you do not have to choose one. And this is perhaps the most honest conclusion available.
The Ideal Workflow: Both, Not Either
It would be convenient for competitive positioning purposes to argue that Synthetic CMOs are intellectual theatre and that customer clones are the only thing that matters. It would also be wrong.
The most rigorous marketing process would use both approaches at different stages of the decision cycle. A plausible workflow might look like this:
Stage 1: Market Diagnosis - Use synthetic customer research (Ditto or similar) to understand how your target audience currently perceives your category, your brand, and your competitors. What do they associate with you? What needs are unmet? Where are the gaps between what you think you offer and what they think you offer?
Stage 2: Strategic Review - Present the diagnostic findings to a Synthetic CMO (Evidenza) for framework-based critique. Does your proposed strategy align with the empirical laws of growth? Is your budget split consistent with the effectiveness evidence? Is your segmentation defensible or have you carved the market into artificially small niches?
Stage 3: Execution Testing - Take the strategy back to synthetic customers (Ditto) and test specific executions - messaging, creative, pricing, packaging - against the audience segments identified in Stage 1.
Stage 4: Ongoing Tracking - Use synthetic customer panels for continuous brand tracking, with periodic Synthetic CMO reviews to ensure strategic drift has not set in.
This workflow is, I suspect, prohibitively expensive for most organisations today. Evidenza's pricing places it in the $50,000-$100,000 annual range, and adding a second platform doubles the research budget. But prices in synthetic research are falling rapidly. What costs $150,000 today may cost $30,000 in two years. The workflow, even if aspirational today, points toward how marketing research will actually be conducted in the near future.
The Pricing Question
Evidenza has not disclosed whether Synthetic CMO access is included in all enterprise engagements or available as a standalone add-on. This ambiguity matters because the Synthetic CMO feature is the most distinctive element of their offering - the thing that no other platform currently replicates.
If Synthetic CMOs are included as part of a broader enterprise engagement, then the pricing question folds into the larger calculation of whether $50,000-$100,000 per year delivers sufficient value across all six of Evidenza's research modules. For organisations that would use segmentation, positioning, creative testing, and Synthetic CMO advisory together, the bundled model is likely efficient.
If Synthetic CMOs are available as a standalone product - or planned for eventual self-serve release - the pricing dynamics shift considerably. A strategy team that wants only the advisory function, without the broader research suite, would face a different value calculation. And the "self-serve" label that Evidenza has listed as "coming soon" on its website raises the possibility that a more accessible version of the Synthetic CMO product may eventually reach the market at a lower price point.
Weinberg and Lombardo are former LinkedIn B2B Institute leaders who built their reputations on the argument that B2B marketing is plagued by short-termism and tactical thinking at the expense of brand building. It would be consistent with their intellectual commitments to offer Synthetic CMOs as a way to democratise access to strategic frameworks - to give every brand manager access to a simulated version of the thinking that was previously available only to those who could afford Mark Ritson's consulting fees or a seat in the Ehrenberg-Bass research programme.
Whether they democratise the access or gate-keep it behind enterprise pricing will say something about the kind of company Evidenza intends to become.
What the Thinkers Think About Being Cloned
Ritson's position on the advisory board implies endorsement, or at least tolerance, of the concept. He has written positively about Evidenza in Marketing Week, though his columns have focused more on the platform's research capabilities than on the experience of being computationally replicated. One imagines that a marketing professor who has spent a career arguing for the irreducible importance of contextual judgment might have complicated feelings about being reduced to a set of computational weights. But he has not, to date, expressed public reservations.
Sharp has been characteristically less forthcoming. The Ehrenberg-Bass Institute's relationship with Evidenza is unclear from public sources, and Sharp's personal position on being AI-modelled has not been publicly articulated. Given that Sharp has spent decades arguing that marketing is more science than art - that empirical generalisations should replace practitioner intuition - one might expect him to be philosophically sympathetic to the project. Whether he is practically comfortable with it is another matter.
Binet and Field have likewise not commented publicly on their synthetic representations. Their framework, being explicitly quantitative and rules-based, is arguably the most amenable to AI replication. The 60:40 split (and its category-specific variations) can be encoded without much loss of nuance. Whether the subtlety of their later work - which increasingly emphasised that the ratio varies by context, maturity, and competitive dynamics - survives the encoding process is the kind of question that would require testing to answer.
The Deeper Implication
The Synthetic CMO concept, whether it succeeds commercially or remains a niche curiosity, introduces a question that the marketing profession will need to confront: if the frameworks that underpin marketing strategy can be computationally replicated, what exactly is the strategist selling?
The answer, I think, is judgment under uncertainty. Frameworks are necessary but not sufficient. Knowing the 60:40 split exists tells you nothing about whether your specific situation calls for 80:20 or 50:50. Knowing that mental availability matters tells you nothing about which of six creative routes best builds mental availability for your brand in your category at this moment. Knowing that segmentation is overrated (Sharp's view) or essential (Ritson's view) does not resolve the contradiction - it merely presents you with two expert opinions that disagree.
This is where the synthetic customer has an advantage that no synthetic advisor can replicate. The customer does not have a framework. The customer has a reaction. When 500 synthetic consumers look at your packaging and 63% choose the green variant over the blue, that is not a theoretical position - it is a data point. When 800 synthetic voters in a swing state rank "healthcare costs" above "immigration" as their primary concern, that is not a framework-derived prediction - it is a simulated measurement.
The best marketing decisions, of course, combine both: frameworks to structure the problem, data to resolve it. Evidenza has made a bold bet that the framework side of this equation can be productised. At Ditto, we have made the complementary bet that the data side can be. The market will ultimately decide which bet was better placed, though I suspect the answer, as with most interesting questions, is that both were right about different things.
Disclosure: Phillip Gales is co-founder at Ditto, a synthetic research platform that competes with Evidenza in the broader synthetic market research category. This article attempts to be fair in its assessment of Evidenza's Synthetic CMO product, but the reader should factor in the author's competitive interest when evaluating claims and comparisons. All information about Evidenza is drawn from publicly available sources including their website, press coverage in Adweek, The Drum, and Marketing Week, and public statements by their founders and advisors.
Phillip Gales is co-founder at Ditto. He writes about synthetic research, competitive intelligence, and the occasionally absurd intersection of AI and marketing theory. Previously: strategy consulting, too many spreadsheets, and a persistent conviction that the customer knows more than the consultant.