Growth-share matrix
The growth-share matrix, or BCG matrix, places a company's products and business units on two measured quantities: the growth rate of their market and their market share set against that of their largest competitor. The grid's four cells are named star, cash cow, question mark and dog, and each carries a cash forecast: growth consumes cash, relative share produces it. A portfolio in which the two offset each other funds its own growth without outside capital. Alan Zakon sketched the grid at the Boston Consulting Group, and Bruce Henderson published it from 1970 in the BCG Perspectives series.
Goal
The growth-share matrix answers an allocation question: among a company's products, which ones receive resources, which ones supply them, which ones no longer deserve any. It places each product on two quantities that can be observed in its market.
Henderson's reasoning rests on two propositions. Market growth drives the use of cash: a line growing fast absorbs capital to hold its place. Relative market share drives the generation of cash, through the experience curve: the competitor that has produced the most cumulative units carries the lowest unit cost and therefore earns the widest margin. The names of the four cells follow from this.
The deliverable is a chart carrying one business unit per circle, with the allocation decision taken for each. Market analysis supplies the input data, growth rates and competitors' shares; financial analysis puts a number on the investment the reading calls for. SWOT analysis and the balanced scorecard sit on the same strategic ground, one as an inventory of position, the other as a standing measure of performance.
Usage
When to use
- An investment budget to split across several lines: comparing products that no common scale otherwise separates.
- Growth to be funded without outside capital: identifying the lines that produce cash and those that consume it.
- A market where the cost advantage comes from cumulative volume: relative share predicts a lasting cost gap there.
When not to use
- A cost advantage unrelated to cumulative volume: the horizontal axis no longer measures anything, so run a market analysis that names the sources of advantage.
- A decision covering a single line: nothing to balance, so weigh the investment through financial analysis.
The two axes and their thresholds
The vertical axis carries the annual growth rate of the market the unit operates in, never the growth of the unit itself. It indicates cash consumption. Its threshold is the part of the technique most often distorted: Henderson ties it to the rate of return the company requires of its investments, holding that growth below that threshold makes cash today worth more than its future equivalent. The threshold therefore belongs to the company drawing the grid.
The horizontal axis carries relative market share: the unit's share divided by that of its largest competitor. A unit holding 8% of a segment dominated by a competitor at 32% sits at 0.25. The threshold follows from the definition: parity with the leader is 1.0, and Henderson notes that only one competitor per segment can sit to the left of that line, all the others being smaller than it by construction. The axis reads from high share to low share, left to right, in the original layout.
A share expressed in percentage points would say nothing: 8% makes a company a leader in a fragmented sector and a bit player in a duopoly. The ratio to the largest competitor places the unit in the contest it is fighting.
The experience curve under the horizontal axis
The experience curve is the observation BCG built before the matrix: unit cost falls by a constant percentage with every doubling of cumulative output, of the order of 20 to 30% according to the ratio table reproduced in the 1973 text. Relative market share is an indirect measure of accumulated experience. Henderson quantifies the stake: a relative share gap of two to one goes with a pre-tax cost differential of about 20% or more on value added.
This assumption carries the whole horizontal axis. Where cost does not depend on cumulative volume, in a service business built on specialised labour or in a market where technology renews costs faster than experience lowers them, the axis keeps its graduations and loses its meaning. Wensley disputes the use of the matrix as a general portfolio rule: it holds where the experience effect operates.
The four cells
| Cell | Growth | Relative share | Cash use | Cash generation | Net balance |
|---|---|---|---|---|---|
| Star | high | > 1.0 | high | high | close to break-even |
| Cash cow | low | > 1.0 | low | high | strongly positive |
| Question mark | high | < 1.0 | high | low | strongly negative |
| Dog | low | < 1.0 | low | low | close to zero |
The star consumes a great deal because it is growing and produces a great deal because it leads.
The cash cow is the portfolio's source of funding. Its position gives it the margin, its low growth spares it capacity investment. Henderson assigns it the payment of the dividend, of the interest on debt and of overheads, ahead of funding the stars and the question marks.
The question mark is the most expensive position: the growth of its market calls for capital, its low share produces none. Henderson sees the portfolio's real wager here: winning market share compounds cash needs that growth has already compounded. Most question marks never cross the parity line; the few that do get there because the company backed them to the end.
The dog consumes little and produces little. Henderson judges it in 1973 to be an essentially worthless asset, a cash trap that deserves little investment. Seeger, a decade later, sets three counter-examples against these names: a dog can carry a margin, a cash cow always demands reinvestment, a star can run out of steam.
The path of a unit
The grid freezes a moment; the technique reasons about movements. The life-cycle assumption requires it: growth rates fall back over time, so no unit stays a star. Its relative share at the moment the market slows decides between two outcomes. A unit that holds that share turns into a cash cow, where it returns the capital invested during its growth; a unit that loses it turns into a dog, in a mature market where taking the share back would cost more than it would return.
The question mark follows the same logic upstream: as long as its market is growing, parity can still be bought. Funding a question mark enough to keep it alive but not enough to make it win is the spending the matrix is there to make visible.
Running the exercise
1. Delimit the product-market segments. Both axes are measured inside a segment. Henderson makes this his main warning: a company can be the largest in its industry and lead no single segment taken separately. The breakdown is documented with its boundaries, failing which nobody will be able to reproduce the chart the following year.
2. Measure the growth rate of each segment, on an annual basis and over the same period for all of them, from industry statistics or from the data the market publishes.
3. Compute relative share: the unit's share divided by the share of its largest competitor in the segment. The result is a dimensionless ratio.
4. Set the two thresholds. The vertical line is at 1.0 by definition. The horizontal line is set at the rate of return required of investments, and that value is recorded beside the chart.
5. Draw the chart: one circle per unit, its area proportional to its revenue, as in the original chart.
6. Read the balance, then decide: does the portfolio produce enough to fund what it consumes and by when? The allocation decision is taken unit by unit, with the amount and the term of the commitment, and is read again at the next review by comparing the new chart with the previous year's.
Where the reading breaks down
The growth threshold becomes a universal rule
The 10% found in the secondary literature comes from the illustrative chart in the 1973 article, where the number has no other status. Applied unchanged by a company whose required rate of return is 5%, it pushes lines into the dog and cash cow cells even though their growth exceeds what the company requires.
The labels are read literally
Seeger warns against reading the labels literally, and the most exposed cell is the dog. A line filed there can stay profitable, occupy a segment the company holds for other reasons or carry a share of fixed costs that its disappearance would redistribute. The cell indicates a cash position; the decision to divest calls for other evidence.
Market share is taken for a cause of profitability
Hambrick, MacMillan and Day tested the matrix's prescriptions against PIMS data on industrial units: performance does not follow the cell as mechanically as the simple model leads one to expect. Wensley adds that neither PIMS nor the matrix accounts for competitive risk or for competitors' expectations. The prudent reading treats the cell as a hypothesis, to be checked against the unit's actual costs and margins.
The segmentation is taken over from internal reporting
The management accounting lines are already there, so the chart is drawn in a morning on segments that belong to no market. The relative shares are then arithmetically exact and meaningless. The check fits in one question: for each segment retained, can anyone in the company name the largest competitor in it?
AI considerations
A model consolidates annual reports, industry statistics and press releases into a table of shares by segment, computes the growth rates and the share ratios, then converts them into coordinates. It also produces the variants: recomputing the chart under two competing segment breakdowns or under two growth thresholds, then showing which units change cell. This is the sensitivity analysis the manual exercise skips for lack of time.
Both axes are numeric, so a model asked for a market share that no source publishes will deliver a plausible number. That number will move a unit to one side or the other of a line, with the investment decision that follows. Every share and every rate carries its source and its date, and a value no source carries is treated as an estimate declared as such on the chart.
Two decisions stay human. The growth threshold is set by whoever decides the company's return requirement. Backing a question mark to parity is a wager across several financial years, which the committee that carries it formulates itself. The table submitted to the model carries the volumes and the margins by line, commercially sensitive material: it stays in a tool whose hosting location is known.
Examples
The artefact is the direct-sales portfolio of a Swiss non-life insurer, taken at the financial year end. Its growth threshold is set at 6%, the return it requires of its investments.
| Line | Premium volume | Segment growth | Insurer's share | Largest competitor's share | Relative share | Cell |
|---|---|---|---|---|---|---|
| Private motor | CHF 180'000'000 | 1.5% | 22% | 14% | 1.6 | Cash cow |
| Micromobility (e-bike, e-scooter) | CHF 12'000'000 | 14% | 26% | 13% | 2.0 | Star |
| SME cyber insurance | CHF 4'000'000 | 21% | 5% | 20% | 0.25 | Question mark |
| Individual legal expenses | CHF 9'000'000 | -2% | 6% | 18% | 0.33 | Dog |
The micromobility line sits at 2.0, the two-to-one gap Henderson ties to a pre-tax cost differential of about 20% or more on value added. He separately describes the star as close to break-even in net cash. The threshold decides that reading: set at 1%, it would move private motor over to the stars, and the portfolio would lose on paper the line that funds the other three. Cyber insurance and legal expenses both hold a low relative share; only their position against the threshold separates the wager from the retreat.
Visualisations
The grid places each unit and weighs what is at stake for it, through position and through circle area, at a given date. The flow diagram carries time and money: which cells fund the others and what tips a unit from one cell to the next.
Cost
| Phase | Level | Rationale |
|---|---|---|
| Preparation | High | Delimiting the segments, then obtaining competitors' shares and growth rates. In a market with no published industry statistics, estimating the shares is a market-analysis job in itself. |
| Execution | Low | Computing the ratios and drawing the chart fit into half a day once the data is gathered, the threshold variants included. |
| Documentation | Medium | One chart and one table of sources. The segment breakdown and the date of each share are kept, without which the next review has nothing to compare against. |
Tooling
A spreadsheet covers the whole technique: one row per unit, the share ratios in a computed column and a bubble chart for the plot. It also carries the threshold variants, one classification column per value tested.
Business intelligence tools (Power BI, Tableau, Qlik) take over when the volumes come from a data warehouse and the exercise is replayed by segment and by financial year: the chart recomputes on every data refresh, and the history of positions stays available.
The input data comes from outside: industry statistics from trade associations, publications from supervisory authorities, competitors' annual reports and press releases, market research panels. A panel subscription is justified when the amounts at stake in the decision exceed the subscription cost by several orders of magnitude.
Sources
- Bruce D. Henderson, The Experience Curve Reviewed, IV: The Growth Share Matrix or The Product Portfolio, BCG Perspectives no. 135, Boston Consulting Group, 1973.
- Boston Consulting Group, What Is the Growth Share Matrix?, company history page.
- Donald C. Hambrick, Ian C. MacMillan, Diana L. Day, Strategic Attributes and Performance in the BCG Matrix - A PIMS-Based Analysis of Industrial Product Businesses, Academy of Management Journal, 25(3), 1982, pp. 510-531.
- Robin Wensley, PIMS and BCG: New Horizons or False Dawn?, Strategic Management Journal, 3(2), 1982, pp. 147-158.
- John A. Seeger, Reversing the Images of BCG's Growth/Share Matrix, Strategic Management Journal, 5(1), 1984, pp. 93-97.