Defensibility and Moats

A moat protects returns when capable rivals respond. Test each claimed advantage for a financial benefit, a barrier to imitation and evidence that the barrier is strengthening.

Why this matters

A feature advantage lasts only until a rival can match it. Defensibility asks what keeps competitors from eroding your returns.

Warren Buffett’s 2007 shareholder letter favored durable cost and brand advantages and warned about rapid industry change. His annual test remains useful: did the moat widen or narrow?

Use that question to guide investment, pricing and the credibility of your strategy. Chapter 13 maps competitors. Chapter 14 tests platform and network effects. Here you assess which advantages survive a competitive response.

The frameworks: what makes an advantage last

Helmer’s 7 Powers: benefit plus barrier

Hamilton Helmer’s 7 Powers (2017) gives a precise test for durable advantage.

Helmer defines Power as the conditions that create the potential for persistent differential returns: returns above those of your competitors that last. Every Power has two parts, and you need both:

• A benefit: something that improves cash flow. You can charge more, your costs are lower, or you need less investment to deliver the same thing.
• A barrier: something that stops competitors from copying the benefit and competing it away.

A benefit without a barrier invites imitation. Helmer excludes operational excellence because best practice spreads. His valuation shorthand, market size multiplied by Power, shows why large markets can still produce weak returns.

Here are the seven Powers, each with its benefit and barrier:

Counter-positioning can be temporary. In a 2020 Acquired interview, Helmer explained that incumbents may eventually accept damage to their existing business and copy a superior model. Use the window to build a second Power.

Switching costs need repeat purchase. Helmer notes that they earn returns through renewals, expansions or new modules. Look for the benefit in renewal pricing, expansion revenue or net revenue retention.

Powers have timing. Helmer places the opportunity to establish switching costs in the takeoff phase. Adding them after rapid adoption can be difficult.

Most products have one or two. A business with one strong Power, deliberately deepened, is in better shape than one that claims five on a slide.

Porter: advantage comes from a system, not a feature

Michael Porter’s “What Is Strategy?” (1996) distinguishes operational effectiveness, doing similar activities better, from strategy, choosing different activities to deliver a unique mix of value. Best practices diffuse, making effectiveness alone insufficient.

Porter argues that two things make a strategic position sustainable:

• Trade-offs. You choose what not to do. A company that serves enterprise security reviewers with deep audit controls can’t also be the cheapest self-serve tool, and a competitor that tries to do both ends up worse at each.
• Fit. Your activities reinforce each other. Porter describes three orders of fit: simple consistency between activities and strategy, activities that reinforce one another, and optimization of effort across them. Fit makes a position hard to copy, because a rival has to match the whole interlocking system, and the odds of matching every activity compound against the imitator.

In “Strategy and the Internet” (2001), Porter argued that widely available technology weakens profitability unless integrated into a distinctive position. Apply the same test to generative AI.

Assess the system of data, workflow, pricing, channels and safety reputation. Their fit can force rivals to copy several interdependent activities.

Christensen’s disruption theory, stated correctly

Clayton Christensen and Joseph Bower introduced disruption in 1995. Christensen developed it in The Innovator’s Dilemma (1997), then clarified its meaning with Michael Raynor and Rory McDonald in 2015.

The theory, as the Christensen Institute states it, describes a process:

1. Incumbents improve their products for their most demanding, most profitable customers. Christensen calls this sustaining innovation. Eventually they overshoot what many mainstream customers need.
2. A smaller entrant with fewer resources gets a foothold in one of two places: the low end of the market (customers the incumbent is happy to give up, because they’re the least profitable) or a new market of nonconsumers (people who didn’t buy anything before, because existing products were too expensive or complicated).
3. The entrant’s product starts out worse on the measures of performance the mainstream market has always valued, but it’s cheaper, simpler or more convenient.
4. The entrant improves until it’s good enough for mainstream customers, who then switch.

The Christensen Institute is explicit that disruptive innovations aren’t breakthrough technologies that make good products better. A better product sold to the incumbent’s best customers is a sustaining innovation, and by Christensen’s own account incumbents usually win those fights.

This is why Christensen and his co-authors argued in 2015 that Uber didn’t fit the theory: it started in the mainstream taxi market, not at the low end or among nonconsumers. Uber changed an industry, but through a different process.

The theory has serious critics. Jill Lepore’s 2014 New Yorker essay challenged case selection and noted that some supposed victims remained dominant. Andrew King and Baljir Baatartogtokh’s 2015 MIT Sloan analysis found many cases failed one or more required conditions: sustaining trajectory, overshoot, ability to respond and actual decline. Treat disruption as a specific pattern with limits.

The disruption test. Use the correct version of the theory as a four-question check on any threat:

1. Is the threat worse on the measures our customers use to judge us today?
2. Is it cheaper or simpler, or does it serve people who don’t buy from anyone today?
3. Is it improving fast on the measures that matter to our mainstream customers?
4. Is it attached to a business model we’d find unattractive to copy?

Four yeses describe a disruptive threat. Christensen himself later observed that low-end entry tends to come with a business model the incumbent finds unattractive, which is the bridge to Helmer’s counter-positioning. If the answers are mostly no, you’re facing a sustaining competitor, and you fight it with product, sales and price.

Moats in AI-era software

Jerry Chen’s AI-era revision of “The New Moats” emphasizes familiar advantages: workflows, data integration, trust, networks, scale and go-to-market. Test which remain durable as building costs fall.

Model access gives an application builder no exclusive barrier. Stanford’s AI Index 2025 reported a more than 280-fold fall in GPT-3.5-level inference cost between November 2022 and October 2024. Guido Appenzeller’s “LLMflation” analysis estimated roughly tenfold annual declines. Epoch AI found task-dependent declines of about 9–900 times annually, with caution about persistence. The AI Index also reported benchmark gaps narrowing between open and closed models (8% to 1.7%) and first and tenth place (11.9% to 5.4%). An anonymous 2023 Google employee memo made a similar convergence argument, but represented no official position. David Cahn argues that high fixed costs and low marginal costs drive undifferentiated prices toward marginal cost. Your rivals can buy the same model access.

Test data advantages for diminishing returns. Casado and Lauten’s 2019 “The Empty Promise of Data Moats” argues that enterprise data often becomes costlier and less valuable at the margin. Their chatbot example plateaus around 40% automated intent coverage. NFX similarly treats data effects as weaker than commonly claimed. A defensible corpus needs hard-to-get access, freshness, hard-case coverage and accuracy that changes adoption. Chapter 14 provides the network-effect test.

Workflow depth can create switching costs and process power. Chen distinguishes systems of record, engagement and cross-system intelligence. Bain’s 2025 technology report emphasizes authoritative data structures, activity history and embedded regulation. Check whether customers have built processes, approvals and compliance evidence around your product.

Proprietary context can be a candidate barrier. Anthropic’s 2025 context guidance emphasizes retrieving high-signal information just in time. Agents need customer topology, history, runbooks, permissions and outcome labels. Classify this as a cornered resource only when contract, structure or position protects access. Context any connected rival can retrieve offers, at best, a scale advantage.

Interoperability lowers integration moats. “We have 200 integrations” used to be a real switching cost. Anthropic introduced the Model Context Protocol (MCP) in November 2024 as an open standard for connecting AI applications to tools and data. In December 2025 it was donated to the Agentic AI Foundation, a directed fund under the Linux Foundation co-founded by Anthropic, Block and OpenAI, with more than 10,000 active public servers and more than 97 million monthly SDK downloads at the time. Google launched the Agent2Agent (A2A) protocol in April 2025 with more than 50 technology partners. When connectors are standard, the breadth of your integration catalog stops being a barrier. What still counts is depth: how much a connector can do safely, how well it’s governed, and what you learn from the actions taken through it.

Distribution and the user relationship. Ben Thompson’s Aggregation Theory, set out on Stratechery in 2015, argues that when distribution becomes nearly free, power shifts to whoever owns the relationship with the user, and suppliers get commoditized. In 2026 the AI assistant with connectors is a new candidate aggregator. A SaaS product reached only as a connector inside someone else’s assistant risks becoming a supplier to that assistant. Some companies are responding by putting up data walls: Slack’s API terms, effective October 2025, bar external apps from using Slack data obtained through the API to train large language models and from building persistent indexes of other organizations’ data. Bain’s 2025 report also recommends that SaaS vendors limit other AI platforms from learning on their data. Data walls protect a position, but they cost you openness, and Chapter 14 explains why openness can be the stronger play.

Trust and brand. Helmer’s branding benefit includes reduced uncertainty, and that’s the enterprise AI case exactly. A buyer letting an agent act in production is paying for confidence that it won’t cause harm. The barrier is hysteresis, which cuts both ways: trust takes years of consistent behavior to build and one bad incident to dent. In practice, AI trust is built from visible mechanisms (cited evidence, approval gates, audit logs, documented risk controls) and a long record of using them.

Ecosystems and complements. Joel Spolsky’s 2002 essay “Strategy Letter V” put a classic platform strategy in one line: “Smart companies try to commoditize their products’ complements.” If the layers next to yours become cheap and plentiful, demand for your layer rises. For an AI product, that can mean opening schemas, publishing connectors and welcoming partner agents, so the ecosystem grows around the part you own. Chen makes a related point about open source: it commoditizes the layer it touches and pushes value to the layers beside it.

AI and the threat to SaaS seats

The sharpest strategic question for many B2B products in 2026 is what happens to per-seat pricing when software does the work people used to do. Bain’s 2025 report put it plainly: seat-based pricing may not fit when AI is doing the work. Bain scores each workflow on two dimensions, how far AI can automate it and how far third-party AI can reach into it. The resulting four scenarios are AI enhancing the existing product (“core strongholds,” where human judgment still matters and rivals struggle to copy the workflow), spending compressing (“open doors,” where third-party agents reach in through exposed APIs and capture value), AI outshining the existing product (“gold mines,” where the incumbent’s exclusive data and rules give it a head start on full automation) and AI cannibalizing it (“battlegrounds,” where the incumbent must replace its own product with AI or lose the work). Bain recommends pricing for work done (tasks completed, tickets resolved) rather than for access.

Pricing pages show the shift. As of October 2026, Intercom prices its Fin agent at $0.99 per outcome, and Salesforce prices Agentforce through a mix of per-conversation fees, per-action credits and per-user licenses. Chapter 15 covers how to design these models.

Horizontal agents are also reaching into jobs that role-specific SaaS used to own. In January 2026 Anthropic released 11 open-source starter plugins for Claude Cowork, covering roles such as legal, sales, finance, support and product management. When a general assistant can pick up a role’s workflow through a plugin and a few connectors, a seat-priced tool for that role has to show what it does that the assistant can’t.

Market sentiment has moved too. Inc. reported that roughly $300 billion of SaaS market value was lost in a selloff in early February 2026, dubbed the “SaaSpocalypse.” The causes of a market move are hard to pin down, so treat it as evidence of what your board is worried about, not evidence about your product.

Apply both lenses before you accept the headline. In Christensen’s terms, an agent sold per outcome can be a genuine low-end or new-market foothold: it serves teams that never bought seats, at a lower price, with lower quality at first. In Helmer’s terms, a seat-priced incumbent faces counter-positioning, because charging per outcome would cut its own seat revenue as agents reduce the number of people who need seats. But many “AI kills SaaS” claims describe sustaining innovations sold to the same buyers, which incumbents usually win. Run the four-question test on each threat rather than adopting the narrative wholesale.

How to apply it

Here’s a method for assessing your moat honestly and turning the result into decisions. Run it once properly, then repeat the review every year (Buffett’s cadence) and whenever a major competitor or platform shift appears. It takes about three weeks the first time.

Week 1: Gather evidence before opinions.

1. Collect the claims. Pull every statement of differentiation from the strategy deck, the board deck, the sales battlecards (Chapter 13) and the last few analyst briefings. List each claim in one line. There are usually more than anyone expects, and some contradict each other.
2. Collect the evidence. For each claim, find data that would show the benefit if it were real: win rates against named competitors, discount levels by segment, renewal and expansion rates, gross margin trends, time and cost to onboard, and churn reasons from win/loss notes. Ask your AI harness to assemble this from connected systems, with sources, and treat anything without a source as opinion.
3. Interview the people who see competitors. Talk to four or five account executives, two solutions engineers and two customer success managers. Ask about the last deal lost to a competitor, the last renewal at risk, and what the customer said they could get elsewhere.

Week 2: Run the assessments.

4. Fill in the 7 Powers assessment. For each Power, write the benefit in cash-flow terms, the barrier in competitor terms and the evidence for both. Mark each as present, emerging or absent. Be strict: a Power with an empty barrier cell is absent. Expect one or two to survive.
5. Score the moat sources. Use the moat scorecard to rate the AI-era sources (proprietary data, proprietary context, workflow depth, switching costs, distribution, trust, ecosystem, scale economics, counter-positioning, process power) from 0 to 3. Require evidence for any score above 1. Add Buffett’s column: did this widen or narrow in the past year?
6. Check for disruption. For the three most serious competitive threats, run the four-question disruption test. Separate sustaining threats (fight with product and sales) from disruptive ones (which may need a different business model or unit) and from aggregator threats (where someone else is taking the user relationship).
7. Find the commoditizing features. List the features where a model provider, a platform bundle, open source or an AI assistant now offers something good enough. For each one, decide whether it’s your product or a complement to your product.

Week 3: Decide and set the watch.

8. Write commoditization response plans for the features under threat. Choose among five responses: move up the stack, bundle the feature into a system with fit, reprice, open it up and commoditize it yourself, or harvest and exit.
9. Decide where to invest. Pick the one or two Powers you’ll deliberately deepen over the next year. Name the roadmap items that build the barrier (not just the benefit) and the metric that will show the barrier is widening.
10. Set up the competitive threat monitor. Agree on the signals, sources and triggers, and have the AI harness draft it monthly for you to validate.
11. Present the result. Show your manager and the leadership team the Powers you have, the ones you don’t, the features you’ll stop defending and the barriers you’ll build. Record the decisions in the decision log, with a review date a year out.

Choosing a response when a feature gets commoditized

Commoditization is normal. The mistake is defending a commodity feature as if it were a moat. Five responses cover most situations:

Rita McGrath’s 2013 HBR article “Transient Advantage” is a useful backstop here. She argues that in fast-moving markets advantages are temporary, so companies should manage a portfolio of them and get as good at exiting old advantages gracefully as at launching new ones. Chapter 30 covers the mechanics of retiring a product or feature.

If the commoditizing force is a disruptive entrant (it passes the four-question test), Christensen’s advice from The Innovator’s Dilemma is to consider responding with a separate unit or a separate business model rather than forcing the response through the core business’s priorities and margins. Chapters 20 and 22 cover how to set that up.

The AI lens

Building AI-powered products

When the product is AI-powered, five defensibility questions change.

• The model is a supplier, so treat it like one. Design your architecture so you can switch models without rebuilding the product, and track inference cost per outcome as a core economic metric. Falling model prices help you, but they help your competitors equally, so a cost advantage only counts if it comes from something they can’t copy, such as better routing, caching or smaller specialized models trained on your own outcome data. Chapter 31 covers the engineering.
• Your evaluation pipeline can be process power. A team that turns every production failure into a test case, and every test case into a measurable improvement, builds know-how that’s slow to copy. That’s Helmer’s process power if it compounds and stays opaque to rivals. Write down the loop and measure its speed.
• Context beats corpus. For most B2B agents, the advantage is less about how much data you’ve collected and more about what customer-specific context the agent can retrieve at run time, and whether that context is exclusive. Map which context sources only you can reach, and which any competitor could reach through the same standard connectors.
• Outcome labels are the scarce data. Logs and telemetry are plentiful. What’s rare is a reliable label saying which action actually fixed the problem, approved by a human and confirmed by the result. If your product sits where those labels are created, design to capture them, with consent and contracts that say so.
• Trust is a feature you can lose in a day. For agents that act, trust is the branding Power. Build the mechanisms that make it visible (cited evidence, approval gates, audit trails, kill switches) and keep a record of incidents and fixes. Chapter 33 covers the governance and regulatory side.

Doing the work with AI

AI can reduce the time spent gathering and sorting competitive evidence.

Use AI for evidence and challenge. A connected harness can assemble CRM, win/loss, support and competitor signals, classify threats, draft a 7 Powers assessment and argue against each proposed barrier.

Treat assessments as hypotheses. Models can repeat fashionable moat claims, miss unlogged deals and cannot know rivals’ future decisions.

Keep the evidence attached. Every claim in the moat scorecard should link to a source: a win/loss note, a renewal record, a margin report, a customer quote. Competitive notes and pricing data are confidential; use an approved tool, and keep customer identifiers out of any prompt that doesn’t need them.

The Beacon case: what actually protects Beacon

The evidence. The harness gathers four quarters of win/loss, renewal, discount and churn data. Interviews with five account executives, two solutions engineers and two customer success managers add context. Correlation-only losses increasingly cite bundled alternatives. No account using Beacon’s remediation approval workflow has churned.

The 7 Powers assessment. You go through the Powers one at a time.

“Proprietary AI” disappears: Beacon calls the same frontier models its rivals can, so model access gives it no barrier. “Our data” becomes two things. The resolved-incident history is a data scale effect, valuable but likely to plateau, in line with Casado and Lauten. The outcome labels (which approved action actually resolved the incident) are rarer, and they’re the asset worth designing around.

The moat scorecard. The widened-or-narrowed column carries the most information. Switching costs and workflow depth widened, because more customers routed approvals through Beacon. Integration breadth narrowed: with MCP connectors now standard, “200 integrations” no longer stops anyone. Proprietary context is strong for Northlight Observe customers, where Beacon has deep, governed access to telemetry and change history, and weak for tool-agnostic prospects. That’s the case’s long-running tension between building on Observe data and staying tool-agnostic, now seen as a moat trade-off. Chapter 14 resolved it with an open architecture that gives Observe customers the deepest integration; the scorecard confirms that the Observe advantage is real but shouldn’t be the whole strategy.

The disruption check. You run the four-question test on the two threats. The large observability platforms’ AI summaries are sustaining innovations: better products sold to the same buyers. They’re a serious threat to correlation, but they aren’t disruption, and Beacon fights them with diagnosis quality and workflow depth. The startup ops agent is different. It’s worse on enterprise measures (no approval gates, no audit logs), it’s cheap and serves mid-market teams without an SRE function who never bought an AIOps tool, it’s improving quickly, and its low monthly price is a model Beacon’s enterprise sales motion can’t serve profitably. That’s four yeses, and the exposed flank is Leo’s MSP customers. You recommend the MSP channel get its own packaging and price point rather than waiting for the enterprise product to trickle down.

The commoditization response. Correlation, Beacon’s original feature, is being commoditized from two directions: platform bundles, and AI assistants that can summarize alerts through standard connectors. You write the response plan. Correlation is a complement to diagnosis and action, so Beacon treats it as one: it’s included in the platform fee rather than sold as a feature, and the team considers opening the incident-grouping schema so partners build on it. Value moves up the stack to Diagnose and Act, priced per resolved incident (Chapter 15). The legacy Beacon Classic engine is harvested through the planned migration (Chapter 30).

The data wall question. One proposal is to change Beacon’s API terms, following the Slack precedent, to bar third-party agents from training on or indexing Beacon incident data. You weigh it against the open architecture and the partner ecosystem from Chapter 14. The decision is a narrow wall: partners can read and act on incidents for the customer who connected them, but outcome labels aren’t exposed for training outside that customer, and Sam reviews the terms.

The decision. Leadership approves three moves. Deepen switching costs and process power through approval workflows, allowlist management and a measured post-incident learning loop that captures outcome labels. Pursue counter-positioning through per-resolved-incident pricing. Rebuild trust as an explicit program after the wrong-rollback incident, with cited evidence on every diagnosis and audit logs Sam can export. Each move gets a metric: renewal rate for accounts with approval workflows, eval improvement per month from post-incident reviews, and diagnosis acceptance rate. You set up a monthly threat monitor with product marketing and book the next full moat review for a year out.

Product 360 artifacts

# 7 Powers assessment · [Product] · [Date]

**Owner:** [PM]   **Reviewed with:** [names]   **Next review:** [date, one year out]

Rule: a Power counts only if both the benefit and the barrier cells are filled with evidence. Expect one or two to survive.

| Power | Status (present / emerging / absent) | Benefit in cash-flow terms (price, cost, capital) | Barrier: what stops a capable rival, and why won't they copy? | Evidence (with source) | Action |
|---|---|---|---|---|---|
| Scale economies | | | | | |
| Network economies | | | | | |
| Counter-positioning | | | | | |
| Switching costs | | | | | |
| Branding (including trust) | | | | | |
| Cornered resource | | | | | |
| Process power | | | | | |

## Claims that did not survive
| Claimed advantage | Why it fails (no benefit / no barrier / no evidence) | What it really is (feature, strength, scale effect) |
|---|---|---|
| | | |

## Timing
- Stage of the business (early, rapid growth, mature):
- Powers still available to build at this stage:
- Window closing on (for example, switching costs during rapid growth):

## The one or two Powers we will deepen
1. [Power] · Barrier-building roadmap items: [items] · Metric that shows the barrier widening: [metric]
2.

## Decision
Agreed by [name] on [date]. Logged in [decision log].

# Moat scorecard · [Product] · [Year]

Scoring: 0 = none, 1 = claimed but unproven, 2 = real with evidence, 3 = real, evidenced and hard to copy. Any score above 1 needs a source.

| Moat source | Score (0 to 3) | Evidence (source) | This year: widened / stable / narrowed | What would erase it | Leading indicator to watch |
|---|---|---|---|---|---|
| Proprietary data (exclusive, refreshed by use, in the hard tail) | | | | | |
| Proprietary context (what the agent can know that rivals can't) | | | | | |
| Outcome labels (which action actually worked) | | | | | |
| Workflow depth / system of record or action | | | | | |
| Switching costs (configuration, approvals, compliance evidence) | | | | | |
| Distribution and channel (who owns the user relationship) | | | | | |
| Trust and brand | | | | | |
| Ecosystem (partners, connectors, complements) | | | | | |
| Scale economics (including inference cost per outcome) | | | | | |
| Counter-positioning (business model rivals won't copy) | | | | | |
| Process power (eval and learning loop speed) | | | | | |
| Model access | 0 | Rivals can buy the same models | | | Price and quality of alternative models |

## Annual test
- Widened this year:
- Narrowed this year:
- Biggest surprise:

## Implications
- Stop defending:
- Start building:
- Pricing and margin effect:

# Disruption test · [Threat] · [Date]

**Threat:** [competitor, product or category]   **Source of signal:** [where we saw it]

| Question | Yes / No | Evidence |
|---|---|---|
| 1. Is it worse on the measures our customers use to judge us today? | | |
| 2. Is it cheaper or simpler, or does it serve people who don't buy from anyone today? | | |
| 3. Is it improving fast on the measures our mainstream customers care about? | | |
| 4. Is it attached to a business model we'd find unattractive to copy? | | |

## Classification
Sustaining / Low-end disruption / New-market disruption / Counter-positioning / Aggregator (taking the user relationship)

## Which of our customers are exposed first
[Segment, channel or persona]

## Response
- If sustaining: product, sales and pricing response:
- If disruptive: separate offer, unit or business model considered? [yes/no, why]
- Owner:
- Review date:

# Commoditization response plan · [Feature] · [Date]

**Owner:** [PM]   **Decision maker:** [name]   **Decide by:** [date]

## 1. The feature and the force
- Feature:
- Commoditizing force (model provider / platform bundle / open source / AI assistant / new entrant):
- Evidence that "good enough" alternatives exist (source):
- Share of revenue, usage or deals that depend on it today:

## 2. Product or complement?
Is this feature what customers pay us for, or something that makes the thing they pay for more valuable?
[Product / Complement], because:

## 3. Disruption check
Result of the disruption test (attach): [sustaining / disruptive / aggregator]

## 4. Options
| Option | What we'd do | Revenue effect | Cost effect | Risk | Fit with our Powers |
|---|---|---|---|---|---|
| Move up the stack (sell the outcome) | | | | | |
| Bundle into a system with fit | | | | | |
| Reprice (include it, or price on outcomes) | | | | | |
| Commoditize it ourselves (open it, make it standard) | | | | | |
| Harvest and exit | | | | | |

## 5. Decision
- Chosen response:
- What we stop doing:
- Customer and sales communication needed:
- Metrics to watch (three at most):
- Review date:

# Competitive threat monitor · [Product] · [Month]

**Drafted by:** AI harness from [connected sources]   **Validated by:** [PM]   **Reviewed with:** [product marketing, sales leader]

Rule: nothing enters the monitor without a cited source and a human check.

| Threat | Type (sustaining / low-end / new-market / counter-positioning / aggregator) | Signal observed | Source (link or record) | Moat row it attacks | Severity (H/M/L) | Trigger for action | Owner | Status |
|---|---|---|---|---|---|---|---|---|
| | | | | | | | | |

## Changes since last month
- New threats:
- Threats that grew:
- Threats that faded:

## Triggers hit this month
| Trigger | What happened | Action taken or proposed | Decision maker |
|---|---|---|---|
| | | | |

## Sources scanned
Win/loss notes · CRM opportunity notes · support tickets · call summaries · competitor release notes and pricing pages · partner feedback · analyst notes

Prompts

Goal: Find out which of your claimed advantages are real Powers and which are features or hopes.
Paste in: the differentiation claims from your strategy and board decks, plus any evidence you have (win rates, discounts, renewal and expansion rates, margin trends). This is confidential; use an approved tool.

Act as a skeptical strategy advisor who knows Hamilton Helmer's 7 Powers framework. For [product]:

1. List each claimed advantage. Map it to one of the seven Powers (scale economies, network economies, counter-positioning, switching costs, branding, cornered resource, process power), or to "none".
2. For each, write the benefit in cash-flow terms (higher price, lower cost, less investment) and the barrier in competitor terms (what would it cost a capable rival to copy this, and why wouldn't they?).
3. Mark each as present, emerging or absent. A Power with no barrier, or no evidence, is absent.
4. Flag any claim that is really operational excellence, a data scale effect, or model access.
5. Name the one or two Powers most worth deepening, and the roadmap work that would build the barrier rather than the benefit.

Use the supplied evidence. Mark unsupported fields "No evidence."

[claims and evidence]

Check the output: for every Power marked present, read the barrier cell and ask whether your best-funded competitor would agree it stops them. If the model accepted "our data" or "our AI" as a Power without a barrier, push back.

Goal: Hear the strongest case for why your moat won't hold, before a competitor proves it.
Paste in: your moat scorecard or 7 Powers assessment, and a short profile of your most capable competitor from public sources.

You are the head of strategy at [competitor]. Your CEO has asked how to take [product]'s best customers within 18 months. Using the assessment and profile below:

1. Identify which of [product]'s claimed barriers you could get around, and how (copy, bundle, reprice, partner, use standard connectors or protocols, or go after a segment they ignore).
2. Identify which barriers you could not get around, and why.
3. Say whether your attack would be sustaining (a better product for their best customers) or disruptive (cheaper or simpler, for the low end or for nonconsumers, with a business model they'd find unattractive).
4. Name the first customers you'd target and the message you'd use.

Use specific facts from the material. Label each additional assumption about either company.

[assessment]
[competitor profile]

Check the output: the barriers the model couldn't get around are your candidate moats. Test each against real deal evidence. Treat the attack plan as a hypothesis, not a forecast.

Goal: Decide whether a new competitor is a sustaining threat or a disruptive one.
Paste in: the entrant's public material (website, pricing page, release notes, launch posts) and a description of your mainstream customers and the measures they judge you on.

Using Clayton Christensen's definition of disruption (a foothold at the low end or among nonconsumers, worse on the incumbent's traditional measures at first, improving until good enough for the mainstream), assess [entrant] against [product].

Answer each question with yes or no and the evidence from the material:
1. Is it worse on the measures our customers use to judge us today?
2. Is it cheaper or simpler, or does it serve people who don't buy from anyone today?
3. Is it improving fast on the measures our mainstream customers care about?
4. Is it attached to a business model we'd find unattractive to copy?

Then classify it: sustaining, low-end disruption, new-market disruption, counter-positioning or aggregator. Name which of our customer segments are exposed first.

Classify disruption using the four questions above.

[entrant material]
[our customers and measures]

Check the output: confirm every "yes" cites the entrant's actual material. If the model called a better product for your best customers "disruptive," correct it: that's sustaining.

Goal: Decide what to do with a feature that competitors, platforms or AI assistants now offer as good enough.
Paste in: a description of the feature, evidence that alternatives exist, the revenue or deals that depend on it, and your current 7 Powers assessment.

For [feature] in [product], draft a commoditization response plan.

1. Name the commoditizing force: model provider, platform bundle, open source, AI assistant or new entrant.
2. Decide whether the feature is our product (what customers pay for) or a complement (something that makes our product more valuable). Explain the reasoning.
3. Evaluate five responses: move up the stack and sell the outcome; bundle into a system with our other activities; reprice; commoditize it ourselves by opening or standardizing it; harvest and exit. For each, give the revenue effect, cost effect, main risk and fit with our Powers.
4. Recommend one response, what we'd stop doing, and three metrics to watch.

Support any recommendation to defend the feature with a named barrier to imitation.

[feature, evidence, assessment]

Check the output: if the recommendation is "add more features," ask what the barrier is. Check the revenue effects against your own numbers before sharing the plan.

Goal: Turn scattered competitive signals into a short, sourced monitor you can review in 30 minutes.
Paste in (or connect the harness to): win/loss notes, CRM opportunity notes, support tickets, call summaries, competitor release notes and pricing pages, and last month's monitor. These include customer data; use an approved tool and leave out personal identifiers.

From the sources provided for [product], build this month's competitive threat monitor.

For each signal:
- Describe what was observed, in one line, with the source record or link.
- Classify the threat type: sustaining, low-end, new-market, counter-positioning or aggregator (someone else taking the user relationship).
- Name which moat source it attacks, using these rows: [paste your moat scorecard rows].
- Rate severity (H/M/L) and say which trigger, if any, it hits: [paste your triggers].

Group related signals into one threat. List what's new, what grew and what faded since last month's monitor. Include only signals with a cited source.

Check the output: open three of the cited sources and confirm the signal is described accurately. Read every High-severity item yourself before it goes to the review meeting.

Goal: Find out whether "our data" is a real barrier or a scale effect that will plateau.
Paste in: a description of the data asset (what it is, how it's collected, who else could collect it, how it's used in the product), and any evidence that more data has improved the product.

Assess whether [data asset] gives [product] a defensible advantage. Use these tests:

1. Is the data proprietary or hard to get, or could a competitor assemble a "minimum viable corpus" from public sources, customers' own systems or standard connectors?
2. Is it refreshed by use, or does it go stale?
3. Is it concentrated in the hard, rare cases that matter most, or in common cases rivals already handle?
4. Does more data still improve the product, or has improvement flattened?
5. Does the improvement drive adoption, pricing or retention, with evidence?
6. Is access protected by contract or position (a cornered resource), or only by being first?

Conclude with one of: real barrier, data scale effect with diminishing returns, or not an advantage. Say what evidence would change the verdict.

[data asset description and evidence]

Check the output: if the model concluded "real barrier," confirm that the answer to question 1 names something a competitor truly can't get. Compare the verdict with your Chapter 14 data network effects test.

Anti-patterns

Checklist

• Every claimed advantage has been mapped to one of the 7 Powers or marked “none.”
• Each Power marked present has a benefit in cash-flow terms, a barrier in competitor terms and sourced evidence for both.
• Model access is scored as a supplier relationship, not a moat.
• Any data moat claim has been tested for exclusivity, freshness, coverage of hard cases, diminishing returns and effect on adoption.
• The moat scorecard includes a widened, stable or narrowed verdict for every row this year.
• The three most serious threats have been run through the four-question disruption test and classified.
• Every feature facing good-enough alternatives has a decision: product or complement, and a chosen response.
• The one or two Powers to deepen are named, with barrier-building roadmap items and a metric for each.
• Pricing decisions (Chapter 15) reflect where the product has real pricing power.
• For AI products, inference cost per outcome is tracked and the architecture allows switching models.
• Trust mechanisms (cited evidence, approval gates, audit logs) are treated as part of the moat, with an owner.
• A monthly competitive threat monitor runs, with every entry sourced and validated by a person.
• The assessment and decisions are in the decision log, with a review date a year out.