BCG Growth-Share Matrix

What the matrix is — and the machine Henderson actually built

Around 1970, Bruce Henderson — founder of the Boston Consulting Group — distilled portfolio strategy into a 2×2 in his perspective piece "The Product Portfolio". Each business unit is plotted by market growth rate (vertical) and relative market share (horizontal — your share divided by the largest competitor's, on a log scale in the original). Four quadrants result: Stars (high/high), Cash Cows (low growth, high share), Question Marks (high growth, low share), and Dogs (low/low).

What most retellings drop is that the matrix was a cash-flow machine, not a labeling exercise. Two BCG-era ideas power it: the experience curve — costs fall predictably with cumulative volume, so relative share is a proxy for cost advantage and margin — and the product life cycle — high-growth markets consume cash (capacity, marketing) while mature ones release it. Put together: cows generate the cash, question marks consume it in pursuit of share while growth lasts, stars are the question marks that won, and when their market matures they become the next cows. Sequence, not snapshot.

That logic was built for the conglomerate era, and its proxies have aged unevenly — which is the honest half of this page. But the underlying question the matrix forces — which businesses fund which, and is the portfolio a system or a heap? — has not aged at all. Context: decision-making models; industry-level structure is Five Forces' job, one level below this one.

When to use it — and when not to

The matrix earns its keep when:

• ✓Allocating across genuinely distinct businesses. Multi-line companies deciding where investment, attention and talent flow — the matrix forces the portfolio conversation many firms never have.
• ✓The portfolio has silent cross-subsidies. Plotting reveals which units are living off which — often a surprise to everyone, including the units.
• ✓As a conversation opener, deliberately crude. One chart that gets a leadership team arguing about the right things beats a hundred-page review nobody contests.

And where it misleads:

• ✗Where the experience curve is weak, share ≠ advantage. In software, platforms, and network businesses, advantage comes from switching costs and ecosystems, not cumulative unit volume — the horizontal axis measures the wrong thing.
• ✗Growth ≠ attractiveness. A high-growth market with brutal five-forces structure can be a value trap; a 'mature' niche can print money for decades.
• ✗Labels become prophecies. Call a unit a Dog and watch investment, talent and morale drain until the label is true. The quadrant is a measurement, not a sentence.
• ✗Market definitions get gamed. Every unit is a Star in a market defined narrowly enough. The definition IS the analysis — and it's exactly what never gets argued.

Step by step, with a worked example

Illustrative scenario: an invented industrial group with four units — sensors, legacy gauges, a software platform, and a services arm. The procedure:

1. 1Define the units and their markets — out loud. Is 'software platform' competing in industrial-IoT platforms (huge, growing, giants everywhere → low relative share) or in gauge-fleet management (niche, moderate growth, we lead)? The plot changes completely. Write the definition down; it will be attacked, and should be.
2. 2Measure the axes. Relative share = our share ÷ largest competitor's (a value above 1 means leadership). Growth against a threshold that means something for this portfolio — the original 10% convention is a convention, not a law.
3. 3Plot, sizing bubbles by revenue. Gauges: low growth, share 2.5 → Cash Cow, and the biggest bubble. Sensors: high growth, share 1.3 → Star. Platform (under the honest, wide definition): high growth, share 0.2 → Question Mark. Services: low growth, share 0.6 → Dog, per the axes.
4. 4Read the system, not the labels. The cow funds one big bet. The question mark needs sustained cash to move share before the market's growth window closes — can gauges fund it AND the star's needs? That arithmetic, not the labels, is the analysis.
5. 5Interrogate the boundary cases. Services plots as a Dog — but it drives sensor retention (a linkage the matrix cannot see). Divesting the 'dog' might wound the star. Every real decision lives in exactly this kind of exception.
6. 6Decide per unit, with reasons. Hold star funding; harvest the cow at maintenance investment; fund the platform for eighteen months against named share milestones; keep services but manage it as a component of the sensor offer, not a standalone P&L.

The matrix as an argument tree

In decision-quality terms, the BCG matrix feeds the frame (the portfolio seen as one funding system) and forces values and trade-offs into the open (growth bets versus harvest, this unit versus that one). What it cannot hold is the argument every real portfolio call turns on. On an argument tree:

BCG matrix vs the alternatives