Professor Dawes Was Right About the Problem. The Solution Has Changed.
How the Ehrenberg-Bass Institute’s 95/5 Rule proves that AI optimization is no longer optional, and why algorithmic memory is the structural successor to brand advertising.
By Jason Barnard
In 2021, Professor John Dawes of the Ehrenberg-Bass Institute wrote down an observation so simple that he had never bothered to formalise it. At any given time, only about 5 percent of your potential buyers are in-market. The other 95 percent are not ready to buy today, will not be ready next month, and cannot be persuaded otherwise.
This is not a theory. It is an observable pattern confirmed by purchase frequency data across industries. Companies change business banks roughly once every five years. Corporations replace their principal law firm, their CRM, their logistics partner on cycles measured in years, not weeks. At any given quarter, approximately 95 percent of the addressable market is simply not buying.
Dawes called this the 95/5 Rule, and its implications are profound. If 95 percent of the people who see your marketing are not going to buy for months or years, then the primary function of marketing cannot be to generate immediate sales. Marketing works, Dawes argued, by building memory links to the brand that activate later, when the buyer does eventually enter the market.
This insight was correct. It remains correct. And it exposes a structural problem that traditional marketing has never adequately solved.
The Expensive Fragility of Human Memory
The Ehrenberg-Bass prescription for the 95 percent was brand advertising designed to build “mental availability.” Create distinct impressions about your brand in people’s minds, and those impressions will be activated when buyers come into the market. The mechanism is human memory: plant a seed through advertising impressions accumulated over time, and hope that seed survives the months or years between first exposure and purchase trigger.
The logic is sound. The execution is brutally expensive and structurally fragile.
Human memory decays. The impressions you paid to create fade within weeks without reinforcement. The carefully crafted brand associations compete with thousands of other messages for a finite slice of cognitive real estate. And the timing problem is, in practice, insoluble: you cannot know which of the 95 percent will trigger into market next quarter, so you must advertise to all of them, continuously, hoping that when the moment arrives, your brand is the one they remember.
Dawes himself acknowledged this tension. His recommendation was to shift budget from short-term activation (targeting the 5 percent in-market) toward long-term brand building (priming the 95 percent for future purchase). Binet and Field’s research suggested a 46/54 split for B2B: 46 percent brand building, 54 percent short-term activation. Peter Weinberg and Jon Lombardo went further, arguing that the 95:5 rule is the new 60:40 rule. The principle was right. The medium was the constraint.
Because the medium was human memory, and human memory has limits that no amount of advertising spend can overcome.
What Changed: The Gatekeeper Is No Longer Human
Consider what actually happens when someone from the 95 percent triggers into market. A new project gets approved, a key system fails, a budget allocation lands, and they are suddenly, urgently in-market.
Do they sit back and try to recall every marketing message they have passively absorbed over the preceding months? They do not. They turn to an AI assistive engine. They go to Google, ChatGPT, Perplexity, Claude, or Gemini and begin their research. They ask the machine to help them understand the landscape, evaluate options, and identify credible providers.
The gatekeeper to the 95 percent is no longer fragile human memory. It is durable algorithmic memory. And this changes the economics of the problem completely.
Algorithmic memory does not decay. A machine that has learned your brand’s positioning, verified your claims through corroboration, and indexed your expertise does not forget after six weeks. It does not need continuous reinforcement through paid impressions. It does not compete with thousands of other messages for attention. When the buyer triggers into market and asks the AI for guidance, the machine serves its recommendation based on accumulated understanding, not on whether you happened to run a display ad last Tuesday.
This is not an incremental improvement on the Ehrenberg-Bass model. It is a structural replacement of the medium through which the 95 percent find their solution.
From Memory Links to Algorithmic Confidence
Dawes framed the marketer’s job as building “a memory link for the brand in buyers’ minds.” The AI equivalent is building cascading confidence through the algorithmic processing pipeline. The mechanism is different, but the goal is identical: ensure that when the 95 percent trigger into market, your brand is the one that surfaces.
The algorithmic pipeline has ten stages across three acts. Retrieval: content is discovered, selected, crawled, and rendered. Storage: it is indexed, annotated, and recruited into the system’s knowledge representations. Execution: it grounds AI responses, is displayed to the person, and either wins the conversation or does not. The first eight stages are machine-facing: bots fetch content, algorithms classify and absorb it, confidence accumulates at each gate. At Stage 9 (Display), the person appears for the first time. At Stage 10, the brand either wins the action (The Perfect Click) or does not (silence). Zero sum.
Within Display, there are three layers of trust depth. The person may encounter your brand at the level of awareness (the algorithm mentions you), credibility (the algorithm recommends you in comparison), or understanding (the algorithm trusts you enough to present you as the definitive answer). These three layers map to the customer journey that Ehrenberg-Bass researchers would recognise: awareness corresponds to top-of-funnel, credibility to mid-funnel, understanding to bottom-of-funnel.
The 95/5 Rule maps onto this model as the temporal axis. The 5 percent who are in-market today arrive at Display already deep in the trust funnel. They know you, they are evaluating, they need a final confirmation. The 95 percent will arrive at Display at some future trigger moment, entering at the widest point (awareness) and descending through credibility to understanding.
In Dawes’s model, you prime the 95 percent with advertising so they remember you when they trigger. In the algorithmic model, you prime the pipeline so the machine recommends you when the person asks. The destination is the same. The vehicle is fundamentally more durable.
Why Algorithmic Memory Solves Dawes’s Timing Problem
The most expensive problem in Dawes’s framework is timing. You know the 95 percent will eventually trigger into market, but you cannot predict when. So you must sustain advertising pressure across the entire addressable market, indefinitely, to ensure your brand is mentally available at the unpredictable moment of need.
This is why Dawes, Binet, and Field all emphasise long-term brand building: not because it is efficient, but because it is the only strategy that covers the timing uncertainty. You cannot target the moment, so you must carpet the months and years before it.
Algorithmic memory eliminates the timing problem. When you build cascading confidence through the pipeline (a clear entity home, consistent corroboration across your digital ecosystem, structured claims verified by independent sources), that confidence persists in the machine’s representation of your brand. Knowledge Graphs store entities indefinitely. LLM training data retains patterns across model generations. Search indexes update continuously but do not forget.
When someone from the 95 percent triggers into market six months from now and asks an AI assistive engine for guidance, the machine does not need to recall your advertising. It needs to recall your brand. And if you have systematically educated the algorithms through consistent corroboration, the machine’s recall is effectively perfect.
The implication for marketing spend is significant. The Ehrenberg-Bass recommendation was to allocate roughly half your budget to long-term brand building because human memory is the bottleneck. With algorithmic memory, the “long-term brand building” investment shifts from repeated advertising impressions (which decay) to systematic digital presence engineering (which persists). You invest once in building the algorithmic understanding, then maintain it. The maintenance cost is a fraction of the ongoing advertising cost required to sustain equivalent mental availability in human minds.
The One-Two Punch: Where Human and Algorithmic Memory Converge
This does not mean human memory is irrelevant. It means the two forms of memory now work in concert, and the combination is more powerful than either alone.
You still make a positive first impression on prospects through traditional channels: a conference appearance, a podcast interview, a LinkedIn post, a referral from a colleague. That impression plants a seed in human memory, just as Dawes describes.
The difference is what happens next. In the old model, you spent the next eleven months frantically trying to keep that seed alive through remarketing, email sequences, and repeated brand touchpoints. In the algorithmic model, you let the seed do what seeds do (sit quietly until conditions are right) and focus instead on ensuring the AI assistive engines are primed to validate that impression when the trigger moment arrives.
The prospect’s internal experience becomes: “I remember them, they seemed sharp. And now the AI is telling me they are the best solution.” The faint human memory is instantly confirmed by an apparently impartial third party with access to the entire digital corpus. The AI bridges the gap that was previously the most expensive and unreliable chasm in marketing: the months or years between first impression and purchase decision.
Dawes’s research proves that gap exists for 95 percent of your market. The algorithmic pipeline provides the bridge.
What This Means for Marketers
The 95/5 Rule is not a problem to be solved. It is a structural reality of how markets work, and no amount of marketing spend will change the ratio. What can change is the cost and reliability of reaching the 95 percent when they trigger.
Dawes and the Ehrenberg-Bass Institute gave us the diagnosis: most of your market is not buying today, and your job is to ensure they think of you when they eventually do. That diagnosis stands. The prescription, however, has evolved.
Building memory links in human minds remains valuable, but it is no longer sufficient. The primary gatekeeper to the 95 percent is now an algorithm, and algorithms do not respond to advertising impressions. They respond to consistent, corroborated, structured information that builds confidence in their understanding of your brand.
The marketer’s job in 2026 is to build both forms of memory: human (through distinctive brand experiences that create lasting impressions) and algorithmic (through systematic digital presence engineering that creates lasting machine understanding). The first is expensive and fragile. The second is investable and durable. Together, they ensure that when anyone from the 95 percent triggers into market, you are the answer that both the person and the machine converge upon.
Professor Dawes proved that 95 percent of your market is not buying today. The question was always what to do about it. For decades, the answer was to spend heavily on brand advertising and hope human memory held. Today, the answer is to educate the algorithms, and let algorithmic memory deliver you to the right person at exactly the right time, every time they ask.
The 95 percent are coming. They always were. The only question is whether the machine they turn to for advice will recommend you, or your competitor.
Jason Barnard is the founder and CEO of Kalicube®, a Digital Brand Intelligence™ company that has spent a decade building the methodology and platform for training AI systems to understand, trust, and recommend brands. He has tracked the intersection of brand strategy and algorithmic behavior since 1998. His previous article on the 95/5 Rule, “Most Businesses Lose 95 Percent of Their Market By Fighting the Wrong Battle,” was published in Rolling Stone Culture Council.
Professor John Dawes is a Senior Researcher at the Ehrenberg-Bass Institute for Marketing Science at the University of South Australia.
References
Dawes, J. (2021). “Advertising effectiveness and the 95-5 rule: most B2B buyers are not in the market right now.” Ehrenberg-Bass Institute / LinkedIn B2B Institute.
Dawes, J. (2021). “The 95:5 Rule.” Personal site / original LinkedIn publication.
Weinberg, P. & Lombardo, J. (2022). “The 95:5 rule is the new 60:40 rule.” Marketing Week.
LinkedIn B2B Institute. (2022). “95-5 Rule.” LinkedIn Marketing Solutions.
Ehrenberg-Bass Institute. (2025). “The 95:5 Rule: Why B2B Growth Starts Long Before the Purchase.” marketingscience.info.
Binet, L. & Field, P. (2019). “The 5 Principles of Growth in B2B Marketing.” LinkedIn B2B Institute / IPA. Full PDF.
Barnard, J. (2025). “Most Businesses Lose 95 Percent of Their Market By Fighting the Wrong Battle.” Rolling Stone Culture Council.
Barnard, J. (2015-2026). “The Kalicube Process™.” Kalicube®.
Barnard, J. (2015). “What is an Entity Home?” Kalicube Learning Space.
