Methods · 08/10
BCG Portfolio Matrix — Steering Business Units by Growth and Share
A portfolio grid that orders business units by market growth and relative market share in order to allocate resources between them.
Purpose
The BCG matrix — the growth-share matrix — positions a company's business units on two dimensions: the growth of their market and their market share relative to the strongest competitor. Four categories result: stars (high growth, high share), cash cows (low growth, high share), question marks (high growth, low share), and dogs (low on both). The underlying logic ties market share, via experience-curve effects, to earning power, and market growth to investment need.
The instrument addresses the central question of diversified companies: which businesses should generate funds, which should receive them, and which should be divested? It emerged in the late 1960s at the Boston Consulting Group around Bruce Henderson and shaped the era of conglomerate management. Its normative pattern — cash-cow surpluses fund selected question marks into stars, while dogs are reviewed and usually pruned — remains the reference point of every portfolio discussion, including those that argue against it.
Procedure in five steps
- Delineate business units
First, the company is decomposed into strategic business units that act independently in their markets and can be accounted for separately. This cut determines the outcome: too coarse, and troubled businesses vanish into averages; too fine, and the picture dissolves into noise.
- Determine axis values
For each unit, market growth and relative market share are established — the latter as the ratio of own share to the largest competitor's. Both presuppose a defensible market definition; this is the methodically most delicate step and should be documented so the positioning remains contestable.
- Visualize the portfolio
The units are plotted as circles in the matrix, with circle area proportional to revenue or capital employed. Only this picture reveals the shape of the portfolio: concentration risks, a missing pipeline of future stars, an overweight of aging cash cows.
- Derive norm strategies critically
For each position, the norm strategy — invest, hold, harvest, divest — is stated as an opening hypothesis and then tested against unit-specific knowledge: interdependencies, strategic importance, turnaround prospects. The norm strategy is a basis for debate, not a verdict; mechanical application is the matrix's best-known abuse.
- Reallocate and follow up
The debate concludes in capital allocation: which question marks merit a genuine bet, how much is harvested from cash cows, which dogs receive a divestment mandate. Each position gets target coordinates and a date, checked against actual movement in the next cycle.
Strengths and limits
- The matrix forces a view of the portfolio as a whole rather than an isolated defense of individual businesses.
- It anchors the uncomfortable insight that businesses play different roles — not every unit is supposed to grow.
- The visualization instantly exposes structural flaws such as a missing pipeline of future businesses or dependence on a single cash cow.
- Two obtainable metrics suffice for a first working picture — the barrier to entry is low.
- Two dimensions cannot carry the evaluative load: profitability, synergies, and competitive dynamics stay outside the frame.
- The assumed link between market share and earning power often fails in platform, niche, and project businesses.
- The market definition is manipulable — almost any business can be computed into a star by drawing its market narrowly enough.
- Norm strategies applied mechanically can become self-fulfilling, as when a business declared a dog is starved into actually becoming one.
With AI and agents
The most obvious effect of LLMs and agents concerns cadence. The BCG matrix lived on the annual rhythm of the strategy off-site because its axis values were laborious to obtain. Agents can estimate market growth and competitive shares continuously from industry data, annual reports, and tender databases, keeping the portfolio as a permanently current picture. The yearly snapshot becomes a time series — showing not just position but drift velocity: a cash cow whose market shrinks faster than planned no longer waits for the next strategy meeting to be noticed.
The method's weakest point, market definition, can also be hardened. LLMs can compute alternative market delineations systematically and show how stable a positioning is against the choice of definition — a business that qualifies as a star only under a flattering cut becomes recognizable as such. For question marks, agents can work the decisive follow-up question the matrix itself never answers: what would have to be true for this business to reach the share threshold, and how much of that is plausible on available evidence?
The core of the portfolio decision stays human: divestment. Whether a dog is sold depends on interdependencies, responsibility for people, political costs, and strategic options that appear in no metric. A further caution: a continuously updated portfolio invites continuous steering. Businesses need time to respond to investment — the judgment of when a signal justifies a course correction and when it is noise becomes more important with denser data, not obsolete.
Relation to scenarios
The BCG matrix stands in a productive tension with scenario thinking, because its most critical input — market growth — is an assumption about the future dressed up as a number. Scenarios make that assumption honest: positioning the portfolio under several images of the future reveals which stars shine in only one scenario and which cash cows dry up abruptly in certain futures. A scenario-differentiated matrix replaces the single portfolio chart with a range — and moves the discussion to where it belongs: the businesses whose role flips depending on which future arrives.
How ready is your organization for AI and agents? — Take the AI Readiness Check