Bruce Henderson and the Boston Consulting Group built the growth-share matrix in the early 1970s to answer one specific question. A diversified conglomerate owns a dozen unrelated businesses. Which should get the next dollar of capital, and which should fund it? The matrix plots every business unit on two axes. Relative market share runs along the x-axis on a log scale. Almost everyone gets one detail about it wrong. You measure it against the largest other competitor. You never measure it against the total market, and never against yourself, even when you're the leader. Market growth rate runs up the y-axis. Four quadrants follow: Stars (high share, high growth), Cash Cows (high share, low growth), Question Marks (low share, high growth), and Dogs (low share, low growth).
How does the BCG matrix work?
The driver logic is a causal chain you can say in one breath. Higher relative share means higher cumulative volume than rivals, which lowers unit cost through the experience curve, which raises margin, which throws off more cash per unit of revenue. Market growth works separately. The faster a market grows, the more capital a unit needs just to hold share, since a unit growing with its market needs new capacity and working capital to stand still competitively. Put the two together and each quadrant stops being a label. It becomes a claim about net cash flow. A Cash Cow is a net cash generator. A Star is roughly cash-neutral, self-funding. A Question Mark is a net cash consumer, with an open question about whether it can win the share to become a Star. A Dog typically just sustains itself. So the portfolio logic runs like this: milk the Cash Cows, fund the Stars, selectively fund the best Question Marks, harvest or divest the Dogs. The exception is a Dog that's propping up another part of the business, and the framework's biggest blind spot lives right there. Worked Example 1 shows it biting for real.
Advertisement
How do you calculate relative market share?
Building it correctly means dodging two traps. The first is using absolute share instead of relative share. A 25% share sounds strong on its own. Put it against a rival holding 55% and relative share is 25/55 ≈ 0.45, on the low-share side of the matrix. Relative share is always your share ÷ the largest other competitor's share, whether or not you're that largest competitor. The second trap is treating the growth-rate threshold as a universal constant. It's an analyst's explicit choice. Analysts commonly use a portfolio-weighted average, or a flat reference like nominal GDP-plus. A different, equally defensible threshold can flip a unit's classification. So state your threshold out loud. If the classification depends on a threshold nobody stated, nobody else can reproduce your matrix.
What are the limits of the BCG matrix?
Three blind spots show up at the mechanism level. First, the cash-generation claim rests on the experience curve actually being the industry's cost driver. Where brand equity, regulation, network effects, or a scarce input drives cost instead, high relative share doesn't reliably produce a cash advantage. Second, market definition is a choice, and it changes the share numbers directly. The same business can look like a Dog nationally and a Star regionally. Worked Example 1 and Practice item P3 put real numbers on this. Third, the matrix scores every unit as financially independent. That erases interdependencies like shared distribution, or a "Dog" that's really the channel infrastructure a "Star" depends on, and a purely cash-driven review will happily divest it into oblivion. There's a quieter fourth spot too. The matrix says nothing about the cost of capital. It classifies by growth and share and never asks whether the unit's return clears its hurdle rate, which is cost-of-capital territory in the Finance galaxy. A "Cash Cow" earning below WACC is being milked, not managed.
{
"skill": "Building a first-pass BCG or Ansoff matrix fast, without ceding the classification itself",
"use": "Have an AI assistant draft the raw comparator table — candidate competitors, their approximate market shares, and the category's approximate growth rate — so you spend your time validating and classifying, not hunting for a first data pull.",
"tools": ["Chat assistants — Claude, ChatGPT", "Research/agentic tools — a browsing-enabled assistant for comparator share data"],
"prompt": "For <CATEGORY> in <COUNTRY/REGION>, list the leading competitors, their approximate market share, and the category's approximate revenue growth rate over the last three years. Name the source of every figure so I can verify it myself before I use it.",
"verify": "Treat every share and growth figure as a draft until traced to a named primary or credible secondary source (a filing, a regulator dataset, a paid industry report) — an AI-generated comparator table can silently mix stale figures, mismatched fiscal years, or a wrong market definition, exactly the market-definition blind spot this section teaches; recompute relative share yourself before classifying anything.",
"diy": "You must already be able to compute relative market share and a market CAGR by hand from two raw figures (this section's core teaching) — AI accelerates the data pull, never the arithmetic or the classification judgment.",
"market": "US"
}