Concept

DHM Model

conceptproduct-strategynetflixframework

DHM Model

The DHM model is a product strategy framework: every strategy should aim to Delight customers in Hard-to-copy, Margin-enhancing ways. Originated at Netflix under Reed Hastings; articulated and taught extensively by Gibson Biddle (VP Product at Netflix 2005–2010).

Primary source: Gibson Biddle on Product Strategy.


The three dimensions

DimensionQuestionNetflix examples
DelightIs this 10X better than the alternative?Personalised recommendations; streaming; original content
Hard-to-copyCan competitors replicate this quickly?Taste data (profiles at scale); brand trust; content licensing relationships; economies of scale
Margin-enhancingDoes this build a better business?Personalisation enables right-sizing content investment; subscription model provides predictable CAC

Why all three simultaneously

Any two legs without the third produces a weak strategy:

  • Delight + hard-to-copy without margin: delightful and defensible, but the business doesn’t work (e.g., dropping Netflix to $5/month).
  • Delight + margin without hard-to-copy: grows the business short-term, but competitors copy immediately (e.g., Netflix’s happy-family hero image on the sign-up page — Blockbuster could copy in a week).
  • Hard-to-copy + margin without delight: builds moat and revenue, but users don’t love it. Defensible incumbency without genuine loyalty.

Strategies are hypotheses

Gibson’s core operating principle: product strategies are not conclusions — they are high-level hypotheses about how to satisfy all three dimensions. Most fail.

Netflix experimented with ~10 strategies; ~6 failed. Failed: social movie recommendations from friends (tested twice); Xbox Party watch-together (barely reached 5% usage). Worked: personalisation, streaming, original content.

The DHM model frames the hypothesis correctly so that failures are informative. A strategy that fails on “delight” teaches something different than one that fails on “margin.”


The delight–margin trade-off

The hardest tension in the model. Resolving it requires quantifying the value of customer retention improvements.

Gibson’s formula at Netflix:

Value = retained customers × LTV × word-of-mouth factor

Netflix used a 2X word-of-mouth factor (each retained customer brings one additional customer). Amazon used 10X. The choice of this multiplier is both empirically difficult to isolate and politically loaded — it determines how much product investment is financially justified.

Perfect New Release Test (Netflix ~2005): A/B test of next-day DVD delivery vs. average 2-week wait. Customers said they wanted faster delivery. The test showed: churn from 4.5% → 4.45%. Value at 2X: $1M. Inventory cost: $5M. Decision: do not roll out.

Lesson: customer stated preference reliably overestimates the actual retention impact of improvements. Measure observed behaviour.


Hard-to-copy taxonomy

Sources of hard-to-copy advantage commonly relevant to product companies:

TypeDescriptionExample
Proprietary dataUnique dataset that compounds with usageNetflix taste graph (1B+ profiles)
Network effectsValue increases as more users joinFacebook friend graph; Figma multiplayer
Economies of scaleCost advantages at volumeNetflix content amortised across 220M subscribers
Brand trustYears of credibility in sensitive contextsNetflix credit card trust; Stripe payments trust
Original IPExclusive content or technologyNetflix originals; Stripe’s reliability engineering
Switching costsHigh cost to leave (B2B especially)Enterprise SaaS with deep integrations

Relationship to adjacent frameworks

FrameworkRelationship
Product Positioning (April Dunford)Positioning answers “for whom and against what.” DHM answers “how do we win once positioned.” Complementary: positioning clarifies the competitive context; DHM evaluates whether strategies within that context are defensible.
Jobs to Be Done (Bob Moesta)JTBD identifies the causal mechanism of switching and purchase. DHM uses JTBD insights to design strategies: a JTBD defines the delight target; DHM adds the defensibility and margin tests.
Reference Customer (Christian Idiodi)Reference customers validate that delight is real (they publicly recommend). DHM provides the strategic frame for evaluating what to build next once delight is established.

Where mainstream views differ

On “delight” as a realistic standard: many practitioners argue “10X better” is a startup/consumer standard not applicable to B2B or enterprise software, where complexity and power matter more than elegance. Gibson acknowledges this: B2B switching costs are themselves a form of hard-to-copy advantage; the delight bar is different from consumer.

On the word-of-mouth factor: the multiplier is genuinely difficult to measure and subject to motivated reasoning (product teams want a higher number; CFOs want a lower one). Gibson’s honest admission: “It was really frustrating. We could never isolate it.”

On the model’s prescriptive limits: DHM tells you how to evaluate strategies, not how to generate them. The creative work of finding novel hard-to-copy advantages requires additional tools (ethnographic research, JTBD, competitive analysis). DHM is a filter, not a generator.