पाठशाला Pathshala · विचार Vichār, The idea · Lesson 26 · Scale

The innovator’s dilemma inside your own roadmap

Spot the low-end entrant your best customers tell you to ignore, date the year it becomes good enough for them, and choose an answer before the date arrives.

Pathshala, The Founder Library · 11 October 2026 · 8 min read

Molten metal pouring from vats on wagons inside a steelworks. — Photograph: Bence Szemerey · Pexels

The company that will take your market is probably already selling. Its product is worse than yours, it is far cheaper, and your best customers have told your sales team it is not worth discussing. They are right today. This lesson is about the date on which they stop being right.

It explains why a well-run company keeps choosing the wrong way for good reasons, gives four signals that separate a disruptive entrant from an ordinary competitor, shows how to date its arrival with numbers you already have, and sets out the three answers open to a company at scale. The closing section turns all of it into a quarterly review that sits inside the roadmap rather than beside it.

The dilemma is a budgeting rule, not a blind spot

Clayton Christensen’s best-known case is American steel. The Christensen Institute’s account of the theory describes how mini-mills melted scrap in electric arc furnaces and could at first make only the lowest grade of steel, reinforcing bar. Rebar earned the integrated mills a gross margin of about 7 per cent, so they were happy to cede it to the mini-mills, which earned 20 per cent or more on the same product. Each year the mini-mills improved and took the next grade up. Each year it made more sense on the integrated mills’ spreadsheets to put capital into the higher-margin grades. That logic held until the integrated mills ran out of customers.

Nobody in the story was foolish. Both the incumbents and the entrants made decisions that were financially sound from where each stood. That is what makes the dilemma hard: it is produced by the same discipline that makes a company good. Allocate capital to the highest return, listen to the customers who pay most, ship what they ask for. Every one of those rules points away from the entrant.

The definition matters, because the word is used for every new competitor. Christensen, Michael Raynor and Rory McDonald returned to it in Harvard Business Review in December 2015: disruption is a small company targeting overlooked customers with a novel but modest offering and gradually moving upmarket to challenge the leaders. By that test they argued Uber, the most cited disrupter of its day, did not fit the pattern. They also warned incumbents against the opposite error: following the slogan of disrupt or be disrupted so faithfully that they put a healthy core business at risk to defend against entrants that were never going to reach it.

Why your best customers vote for the incumbent

At scale the roadmap has an author, and it is the account list. The top twenty customers produce most of the revenue, most of the feature requests and most of the escalations. Product managers rank work by revenue at risk. Sales asks for what closes the next large deal. All of it is sensible, and all of it pulls the product upmarket faster than most customers can use the improvement. The gap between what you ship and what the mainstream customer uses is the overshoot, and it is the space an entrant lives in.

Ask a top customer about the entrant and the answer will be accurate: it cannot do what we need. Ask your sales team about the deals it lost to the entrant and the answer will be accurate too: those were small customers we did not want. Both answers describe the present. Neither describes the trajectory, which is the only thing that matters. The entrant does not need to be as good as you. It needs to be good enough for the customer in the middle of your list, at a price that customer notices.

An Indian case: the flat fee per order

Indian broking has a clean recent example of the low-end pattern. Zerodha began operating on 15 August 2010 and describes itself as having pioneered the discount broking model in India. Its published charges today are zero brokerage on equity delivery and ₹20 or 0.03 per cent per executed order, whichever is lower, on intraday trades. The company says it has more than 1.8 crore clients and contributes over 15 per cent of all Indian retail trading volumes. Those are its own figures, checked in October 2026.

The shape is the one Christensen described. The first customers were self-directed traders who wanted none of what a full-service broker sold: no research desk, no relationship manager, no branch. A broker whose income rose with the size of each trade had little reason to chase a customer who wanted to pay a flat fee and be left alone. The entrant’s product then improved year after year on the measure those customers ranked, the trading platform, until it was good enough for a large part of the middle of the market. The lesson for any Indian company at scale is not about broking. It is that the customer you are happy to lose is often the entrant’s whole plan.

Four signals that separate an entrant from a competitor

Most new competitors are ordinary. They sell roughly what you sell to roughly whom you sell it, and the answer is to be better. A disruptive entrant shows four signals together. It is worse on the measure your customers rank you on. If it is better, it is a competitor and your normal process will notice it. It is cheaper by a multiple, not a percentage. A price at a fifth or a tenth of yours changes who can buy. It sells to customers you decline or neglect. The deals your team calls too small, too slow to close or not worth the support load. It is improving faster than your mainstream customer’s needs. Not faster than you; faster than what the middle of your list can absorb.

The fourth signal is the one that can be measured, and few companies measure it. Two numbers come from your own data. The first is overshoot: of the features you have shipped, what share does the median customer use in a month? The second is the rate at which that customer’s needs grow, which you can read from the year-on-year change in the features they adopt. The entrant’s rate you estimate from its release notes, its pricing pages and the deals you lose to it, each quarter.

A worked example. A Pune company sells a GST compliance suite to mid-sized manufacturers at ₹6 lakh a year. Its usage logs show the median customer uses about 70 per cent of the product. An entrant sells a ₹24,000 app to small traders and, by the company’s own scoring of its release notes, does about 40 per cent of what the suite does, improving by perhaps 35 per cent a year. The suite improves by 15 per cent a year; its customers’ needs grow by 6 per cent. The figure below is set to those numbers.

A customer does not switch when the cheaper product is better. They switch when it is good enough.

On those assumptions the entrant becomes good enough for the median customer in a little over two years, while the suite’s overshoot widens every year. Move the entrant’s yearly improvement down to 10 per cent and the crossing recedes past the edge of the chart. Move the share used down to 50 per cent and the crossing comes almost at once. The arithmetic is crude; its value is that it replaces an argument about whether the entrant matters with an argument about three numbers, each of which someone can go and check.

Three answers, and the one most companies pick by default

Ignore it, on evidence. If the entrant’s line never crosses the need line, because it is improving more slowly than your customers’ needs or along a measure your customers do not rank, it is not a threat to the core. Write that down with the numbers and look again next quarter. Most entrants belong here.

A magnet crane lifting scrap metal in a recycling yard. — Mini-mills began by melting scrap like this into the lowest grade of steel. The integrated mills answered by retreating upmarket and kept retreating. Photograph: Willians Huerta · Pexels

Retreat upmarket, on purpose. This is what the integrated mills did by default. Done deliberately it can be a sound choice: you concede the bottom of the market, raise prices at the top, and run the company for margin. It works for as long as there is a top to retreat to. Its cost is that each year the business is smaller and more exposed to the next step of the entrant’s climb, so it needs a stated floor: the revenue at which the retreat stops and something else starts.

Answer it with a separate unit. The remedy Christensen and Joseph Bower set out in 1995, covered in the [platform shifts lesson](/library/platform-shifts-when-to-bet-the-company), is an organisation independent of the mainstream business: its own price, its own customers, its own profit and loss, and a leader whose bonus does not depend on the core. In practice that means a second product at the entrant’s price point, sold through a channel the core sales team does not own, and allowed to take customers from the core. If the unit has to win an argument with the top twenty accounts to get engineering time, it will lose every one.

The default answer is none of these. It is a feature request: the core team adds a cheaper tier to the existing product, priced to protect the existing product, sold by the existing team. It satisfies the board for a quarter and changes nothing about who the entrant’s customers are.

A quarterly entrant review

Put the dilemma inside the roadmap by giving it a fixed slot. Once a quarter, before roadmap planning, spend an hour on three numbers per entrant: its price as a multiple of yours, its estimated performance against your 100, and its estimated yearly improvement. Add your own two: the share of the product the median customer uses, and the growth in their needs. Put them into the figure and record the crossing year. Add the count of deals lost to each entrant and the median size of those deals; a rising median is the climb, measured.

Then make one decision per entrant and write it in the [strategy memo](/library/strategy-memo-thesis-your-company-runs-on): ignore with the reason, retreat with the floor, or answer with a unit, a leader and a budget. When the crossing year moves inside three years, the decision cannot be ignore. When it moves inside one, the unit should already exist.

Figures were checked in October 2026 against the sources below; company figures are the company’s own. The thresholds in the figure are this library’s judgement. Nothing here is investment advice.