29:25
positioning
April Dunford argues that AI has not changed the fundamentals of positioning — the same five-part framework and process from her 2019 book still hold after working with 60-70 companies through the AI disruption — but volatile, AI-driven markets create new ways to execute positioning badly, especially around articulating a defensible point of view on the future, choosing how far ahead to position, and systematizing customer understanding.
AI hasn't changed the fundamentals of positioning; the same repeatable process from Dunford's 2019 book still applies after working with roughly 60-70 companies over the last 2.5-3 years of AI disruption.
Positioning is defined by five elements: competitive alternatives, distinct capabilities, differentiated value, best-fit customers, and market category.
Step one of the process is mapping competitive alternatives (status quo plus actual shortlist competitors) from the customer's point of view by asking 'if we didn't exist, what would a customer do?'
Step two lists all distinct capabilities (often 50-100 items from a whiteboard exercise), then asks 'so what?' of each to surface customer value, compressing the result to just 1-3 value themes because 'nobody can keep all that in their heads.'
Best-fit customers (ICP) are reverse-engineered from which target accounts care most about that distilled value; only then is a market category chosen.
Even though fundamentals don't change, AI-era markets create 'a hundred new ways to mess it up.'
Four new problem areas: (1) needing a point of view about the future, (2) messiness in the competitive-alternative landscape, (3) how far ahead to position ('the biggie'), (4) operationalizing customer understanding.
In volatile markets customers need vendors to state a point of view on the whole market's future (not just the product) because typical B2B buyers lack the insider exposure to AI discourse that industry insiders have.
About half of B2B purchase processes end in 'no decision' when options look similar and the risk of a bad choice feels high.
Companies' stated AI points of view diverge sharply by business model: Anthropic (AI takes over developer tasks), OpenAI (intelligence as a metered utility, 'like electricity or water'), Replit (agents will spontaneously generate bespoke tools, replacing SaaS subscriptions), Microsoft (models are commodity; owned data surfaces like email/Office/SharePoint plus 'context engineering' are the differentiator), Salesforce/Benioff ('our API is the UI' — expose everything and let customers build however they want), IBM/Krishna (real-time, open, sovereign AI built on mainframes and orchestration), ServiceNow/McDermott (workflow and governance, not models, are 'the whole ballgame').
A credible point of view must be rooted in genuine differentiated capability, tested by asking 'why can only we claim this?' — illustrated via IBM's claimed 70% share of transactional data through its systems, its standards-body openness, and its global physical presence.
Distinguishes product vision (the long-term, investor-facing story), the current product ('the piece of junk we actually have right now'), and product strategy (the sequenced steps bridging the two) — positioning should describe today's product against today's competitors, not the vision.
Client anecdote (Feb 2025, IT-management company): the CPO believed vibe-coding tools (Replit/Lovable, then only ~2 months old) were already a threat ('not yet'), while the head of sales had never seen them on a real short list — dramatizing the tension between positioning for now versus for anticipated future competitors.
Rule of thumb: compete against who you actually compete with right now while building product for the competitors coming later; naming an alternative the customer wasn't already considering risks sending them off to research it and stalling the deal.
When pitching forward-looking capability (e.g., AI agents) to risk-averse enterprise buyers, success comes from 'meeting the customer where they are' via a maturity model: map processes, apply agents to low-risk tasks with human oversight, expand, then reach full agent-human orchestration.
'Operationalizing customer understanding': product, marketing, sales, and the CEO each define 'who we compete with' differently (horizon competitor, spend-driven 'scary competitor', the real short-list minus status quo, and an outdated/investor-skewed view), so no single internal function has the full picture.
Since internal opinions conflict and nobody wins an argument against the CEO or against sales, direct customer data is 'kind of all we got' to settle disputes.
Cautionary historical example: Jack Dorsey renamed Square to Block for an all-in blockchain bet, and weeks later Mark Zuckerberg renamed Facebook to Meta for the metaverse; four years later the metaverse team 'barely exists,' used as evidence that even major leaders are 'lousy at predicting the future.'
Recommended systematic customer-intelligence practices: a customer advisory board; an executive customer-sponsorship program (each executive personally owns about four accounts with quarterly check-ins); win-loss analysis (with win analysis specifically flagged as undervalued); and a recurring 'regular positioning check-in' every quarter or two that explicitly solicits sales's daily customer contact as an early-warning system for positioning drift.
Anecdote: a personal, multi-quarter relationship with Tiffany's CIO let the speaker resolve an internal product-manager dispute by getting the actual customer on the phone rather than guessing.
Speaker released an updated edition of the positioning book a couple of months before the talk, incorporating six to seven years of learning since the 2019 original.
Five-part positioning framework — Dunford's definition of positioning as five elements — competitive alternatives, distinct capabilities, differentiated value, best-fit customers, and market category — that together set the context for how a product is understood. Apply: Work through the five elements in order for any product to establish or audit its positioning.
Competitive-alternatives mapping (status quo vs. shortlist) — A technique for identifying what a customer would do without your product by distinguishing status-quo alternatives (habits, spreadsheets, legacy tools) from the actual shortlisted competitors they compare you to. Apply: Ask 'if we didn't exist, what would a customer do?' in customer/sales interviews, then position against whichever alternative the customer is actually weighing, not every theoretical option.
Capability-to-value 'so what?' exercise — A whiteboard process that lists all distinct capabilities (often 50-100 items) and repeatedly asks 'so what?' of each until arriving at the underlying customer-facing business value, then compresses the results into 1-3 value themes. Apply: Run a workshop to generate the full capability list, iterate 'so what' until reaching value (money, time, risk), and bucket the output into no more than three value themes since customers can't retain more.
Best-fit customer / ICP reverse-engineering — A bottoms-up method for defining the ideal customer profile by identifying which target-account characteristics correlate with caring most about the distilled value themes. Apply: Identify the accounts that most value your differentiated value bundle and use their shared traits to define targeting criteria.
Point of view on the future — An articulated belief about where the whole market — not just your product — is heading, used to help customers evaluate a vendor's roadmap choices during periods of high market change. Apply: Build and communicate a market-level point of view grounded in genuine differentiated capability (tested by asking 'why can only we credibly claim this?') so customers can use it to judge your roadmap and reduce their own indecision.
Product vision vs. current product vs. product strategy — A three-way distinction between the long-term, investor-facing aspirational story, the actual shippable product today, and the sequenced roadmap of steps bridging the two. Apply: Base marketing and sales messaging on the current product and today's competitors, not the vision, and use product strategy internally to explain the path forward.
'Position for now, build for the future' rule — A principle that a company should compete against its actual current alternatives in messaging while its product team builds toward anticipated future competitors. Apply: Keep customer-facing positioning focused on the real, current shortlist (e.g., don't mention vibe-coding tools if customers aren't yet comparing you to them) even while R&D targets future threats.
'Meet the customer where they are' — A technique for pitching forward-looking or futuristic capability by anchoring it to the customer's present maturity level rather than leading with the far-future end state, to avoid scaring risk-averse buyers. Apply: Pair an ambitious market point of view with a concrete near-term entry point so customers see an immediate, low-risk step rather than an intimidating leap.
Adoption maturity model (for agentic AI) — A staged sales framework moving customers from process-mapping to low-risk agent application with human oversight, then expansion, and finally full agent-human orchestration. Apply: Use it to give enterprise customers who fear agents a safe, incremental entry point instead of pitching the full future-state vision directly.
Operationalizing customer understanding (competitor-lens taxonomy) — A recognition that different internal functions hold different, biased views of 'who we compete with': product sees horizon/roadmap competitors, marketing sees whoever spends big on marketing ('scary competitor'), sales sees the real short-list but never counts status-quo/no-decision losses as competition, and CEOs carry an outdated view frozen from when they last sold personally, skewed further by investor-facing narratives. Apply: Reconcile these internally conflicting views with direct, systematic customer data rather than deferring to any single function's opinion (including the CEO's or sales').
Customer advisory board — A standing group of customers a company can consult directly to settle internal disagreements about customer needs, priorities, or perceptions. Apply: Convene one and use it as a source of truth when internal teams disagree about what customers actually want.
Executive customer sponsorship program — A practice where each executive is personally assigned a small number of accounts (about four) to check in with quarterly as their executive sponsor. Apply: Assign every executive roughly four accounts for regular quarterly relationship calls, maintaining firsthand customer insight usable in internal debates.
Win-loss analysis (with emphasis on win analysis) — Systematic post-decision research into why deals were won or lost, with win analysis specifically revealing what's working, why the company wins, and against whom. Apply: Run ongoing win-loss interviews in-house or via an agency, and weight win analysis as heavily as loss analysis to surface real positioning and competitive signal.
Regular positioning check-in — A recurring (roughly quarterly or bi-quarterly) session that reassembles the go-to-market team to 'speed run' the positioning and check whether competitors, capabilities, or differentiation have shifted, explicitly soliciting sales's input. Apply: Schedule this check-in every one to two quarters as an early-warning system for positioning drift, using sales's daily customer contact as a key input.
The worked contrast of divergent AI 'points of view' across major vendors suggests a company's stated market narrative functions as a rationalization built backward from its existing structural advantages, not a neutral forecast.
Avoiding mention of a competitor the customer hasn't yet considered is framed as a tactical necessity, not just a matter of confidence — naming it can actively prompt the buyer to go research that alternative and stall the deal.
Sales systematically undercounts status-quo/no-decision losses as competition because reps reframe every stall as 'not yet,' which distorts the whole company's shared picture of the competitive landscape.
CEOs are singled out as carrying two compounding distortions — a competitive view frozen at the point they last sold personally, plus contamination from investor-pitch futurism — making them the least reliable internal voice on current competition even though no one internally can out-argue them.
The Block/Meta rename anecdote is used to support a general epistemic claim: because even well-resourced leaders demonstrably fail at predicting technology's near future within a few years, companies should weight direct, systematic customer data over confident internal narrative-building.
Win analysis is explicitly framed as more neglected than loss analysis, inverting the usual emphasis in competitive intelligence practice.
«Positioning defines how our product is the best in the world at delivering something, some value that a well-defined set of customers cares a lot about.»
— 02:17
«If we didn't exist, what would a customer do?»
— 03:42
«Nobody can keep all that in their heads.»
— 05:03
«I can confidently say, having worked with, I don't know, 60, 70 companies in the last 2 and 1/2, 3 years, no, it doesn't»
— 06:18
«The fundamentals of it don't change, but there's like a hundred new ways to mess it up.»
— 06:39
«About half the time when they start a purchase process, they actually end up in no decision.»
— 08:53
«We see a future where intelligence is a utility. It's like electricity or water, and people are going to buy it from us on a meter.»
— 10:28
«Microsoft is the sun and every little dude is putting those open AIs just baby in the corner.»
— 12:16
«Our API is the UI.»
— 12:38
«My dudes, the mainframes are coming back.»
— 14:37
«I can only sell what I got on the truck right now.»
— 18:49
«Close the book. Come back in 2 years. Do this. Do this, man, in 2 years.»
— 19:07
«Whoa, whoa, vibe coding? I wasn't even looking at that. Stop. Oh, I got to go research that.»
— 19:59
«you have to compete against who you actually compete with right now, but be building for the thing in the future.»
— 20:15
«That's not a no. That's a not yet, buddy. I'm going to get you next year.»
— 23:04
«Dude, we haven't seen Oracle on our short list in like 5 years. What are you talking about?»
— 23:36
«We are lousy at predicting the future. Terrible at it. Terrible.»
— 25:41
«Win is where we find out what's working.»
— 27:41
Reception
No comments are available to assess audience reception.
As a practitioner talk, it succeeds in reframing a fashionable AI-disruption angle into a reusable checklist of positioning tactics grounded in named company examples and client anecdotes, though the specific vendor-stance details (model counts, vibe-coding threat level) are the parts most likely to date quickly.
29:25