The DHM Model
The DHM model is the sharpest one-line test I know for whether a product strategy is actually a strategy and not just a plan to build good things. It comes from Gibson Biddle, former VP of Product at Netflix, and it asks three questions of any strategic bet: does it Delight customers, is it Hard-to-copy, and is it Margin-enhancing? If a bet can't claim all three, Biddle argues, it's not strategic. I use it constantly as a gut-check on strategy.
The DHM model is Gibson Biddle's framework, developed from his time leading product at Netflix. I'm presenting his model and applying it to my own context.
#The three tests
| Test | The question | Why it matters |
|---|---|---|
| Delight | Does this create genuine, lasting customer joy? | Without delight there's no demand. But delight alone is copyable. |
| Hard-to-copy | Can competitors easily replicate this? | Delight a rival can clone in a quarter is a temporary lead, not a moat. |
| Margin-enhancing | Does this improve the business's economics over time? | A delightful, defensible product that loses money on every customer is a charity. |
The model's genius is the combination. Lots of frameworks (and most PMs) optimize for Delight alone, meaning build something users love. Biddle's insight is that customer love is necessary but radically insufficient. A feature users adore that any competitor can copy next quarter gives you a brief lead and a price war. A defensible, profitable product nobody loves doesn't get adopted. Strategy lives at the intersection of all three.
#Hard-to-copy is the part everyone underrates
Of the three, the one PMs neglect most is defensibility, so it's worth Biddle's own list of where "hard-to-copy" actually comes from:
- Brand: the trust and meaning a name carries (Netflix, Disney).
- Network effects: the product gets better as more people use it (marketplaces, social).
- Economies of scale: cost advantages that compound with size.
- Unique technology / IP: genuinely hard-to-build capability (and increasingly rare as a durable moat).
- Proprietary data: accumulated data that improves the product and can't be bought off the shelf.
When I evaluate a bet, I force the question: which of these does this strengthen? If the honest answer is "none, it's just a nice feature," it might still be worth building (table stakes matter, see The Kano Model), but I won't call it strategic, and I won't expect it to build a lasting advantage. This is the discipline that keeps a roadmap from being a pile of copyable niceties.
The most seductive trap in product is believing that building something users love is your strategy. It isn't. Delight without defensibility is a treadmill: you delight, they copy, you delight again, nobody pulls ahead, and margins erode for everyone. DHM exists to break that reflex by forcing the "hard-to-copy" and "margin" questions you'd rather skip.
#How I use it alongside my other tools
DHM is a qualitative strategy filter, and it pairs well with the quantitative tools elsewhere in this domain. I run it earlier in the funnel than scoring: a bet that fails DHM rarely deserves a RICE score at all, because it's strategically hollow no matter how it scores. It's also a direct complement to my own Strategic Impact behaviors. "Business Understanding & Clarity" is the margin test, and "Product Strategy Identification" is largely the hunt for hard-to-copy leverage. And the Margin-enhancing leg ties straight to Business Acumen and Models: you can't judge margin impact without reading the unit economics.
Biddle also pairs DHM with a "proxy metric" idea (a measurable near-term stand-in for long-term value) and the GLEe model (Get big, Lead, Expand) for sequencing, but if you take one thing, take the three-letter test. Run every strategic bet through D, H, and M before you fund it.
#How AI is changing it
AI doesn't change the model, but it reshuffles the moat sources behind the "hard-to-copy" test, and that's worth saying out loud. Unique technology is decaying fastest as a moat: when a competitor can vibe-code a feature clone in a weekend, "we built something hard" buys you less time than it used to. The moats that appreciate are the ones AI can't trivially replicate: proprietary data (it feeds the flywheels that make AI products improve with use), brand and trust (which matter more when probabilistic systems can fail), and network effects. So the "hard-to-copy" question is more urgent, not less. As a tool, a model is genuinely useful for stress-testing defensibility ("how would a well-funded rival replicate this, and how fast?") and catching wishful thinking. What stays human is the judgment on whether a moat is real and the conviction to invest behind a thesis that won't pay off for years.
#Continue Reading
- Strategic Impact for DHM as a filter for the strategy-identification behavior.
- Business Acumen and Models is the financial literacy behind the "margin-enhancing" test.
- The Kano Model for table-stakes vs delight, complementary to the Delight leg of DHM.
- Competitive Landscape Analysis is the input that makes the "hard-to-copy" assessment honest.
- The North Star Framework for pairing a defensible strategy with a durable value metric.