Selecting a Defensible Comparable Set

Ask two analysts to value the same mid-size software company using comparables and they can arrive at valuations 40% apart — not because they made arithmetic mistakes, but because they picked different peer groups. One included a slower-growing enterprise incumbent trading at 6x revenue; the other included a hot, fast-growing niche player trading at 14x. Same target company, same method, wildly different answer, because the method's entire output depends on a judgment call that happens before any multiplication.

What "comparable" should actually mean

Same-industry membership is a starting filter, not the answer. Two companies in the same industry can deserve very different multiples if they differ on the variables that actually determine a multiple: growth rate (higher growth generally justifies a higher multiple), margins and returns on capital (more profitable, more capital-efficient businesses justify higher multiples), and risk (more leveraged or more cyclical businesses justify lower multiples for the same growth). A slow-growing, capital-intensive telecom operator and a fast-growing, asset-light software company both sit in "technology," but they are not comparable on any dimension that drives a multiple.

Practical screens that narrow a raw industry list into a defensible set: similar revenue growth (within a reasonable band), similar gross or operating margins, similar leverage, similar business model (subscription vs. transactional, for instance), and, where feasible, similar geographic exposure.

Worked example

Valuing a mid-cap specialty retailer with 8% revenue growth and 15% EBITDA margins.

1. Raw industry screen pulls 20 "retail" companies, with multiples ranging from 4x to 16x EV/EBITDA — too wide a spread to be useful without further filtering.
2. Apply growth filter: keep only companies growing 5–12% (excludes hyper-growth e-commerce names and declining legacy chains). This narrows to 11 companies.
3. Apply margin filter: keep only companies with EBITDA margins between 12% and 18%. This narrows further to 6 companies.
4. Median multiple of the final set: 9.2x EV/EBITDA, with a tight range of 8.1x–10.4x — a far more defensible number than the 4x–16x raw industry range.
5. Applying 9.2x to the target's EBITDA of $40 million gives an enterprise value of 40 \times 9.2 = \368$ million, a figure an analyst can actually defend line by line against a challenge on peer selection.

What this means in practice

Every comparable-company analysis should show its work: list the companies excluded and why, not just the final set, since the exclusions are exactly where disagreements happen in practice — in a pitch, a client challenge, or an interview. Small peer sets (under 5–6 companies) are more sensitive to one outlier and should be reported with the full range, not just the median.