Building a Peer Set That Actually Compares

The quality of a comparable company analysis depends almost entirely on the quality of your peer set. Include the wrong companies, and your valuation is garbage—a sophisticated garbage, but garbage nonetheless. Include the right companies, and your valuation is grounded in reality.

Yet peer selection is often done carelessly. Analysts download a list of companies in the same GICS industry classification and call them peers. They do not ask whether those companies are actually similar. They do not adjust for size, growth rate, profitability, or capital intensity. They simply take the median multiple and apply it, blind to the fact that one "peer" is a high-growth, asset-light SaaS company and another is a low-growth, capital-intensive manufacturer—and they both happen to have "software" in their name.

Building a good peer set is hard work, but it is foundational. This article walks through the criteria, the common traps, and the discipline required to build a peer set that actually compares.

Quick definition

A peer set is a group of publicly traded (or recently acquired) companies selected because they share similar characteristics to the target company in terms of industry, business model, scale, growth rate, profitability, and capital structure. Peers should be as similar as possible on dimensions that drive valuation—if they are not similar, the multiples extracted from them will not apply to the target.

Key takeaways

• Peer selection is not mechanical; it requires judgment about which characteristics drive valuation in your industry
• Start with industry classification (GICS or SIC), then apply secondary filters: scale, growth, profitability, business model
• Aim for 5–10 peers; too few and you have no cushion against outliers, too many and you dilute the "comparable" standard
• Real peers rarely exist; you will almost always have a "best available" peer set, not a perfect one
• Document your peer selection rationale so others can debate and improve your work
• Be prepared to explain why each peer is included (or excluded) in plain language
• Median is better than mean; consider trimming outliers on the high and low end
• Segment analysis or separate peer sets may be necessary if your target operates in multiple business segments

The hierarchy of peer selection criteria

When building a peer set, there is a natural hierarchy of what matters most. Think of it as a series of filters, starting broad and narrowing down:

Filter 1: Industry and sector

Start with industry classification. Is the target a healthcare company, a financial institution, a technology company, a retailer? Most databases use GICS (Global Industry Classification Standard) or SIC (Standard Industrial Classification). These are a starting point, not gospel.

For a healthcare company, you might start with all companies in the Healthcare sector (GICS 4010). Then narrow to Healthcare Equipment & Services (GICS 403010) or Pharmaceuticals (GICS 401010). These filters narrow the universe but are still too broad. A regional hospital operator (asset-heavy, steady margins, high debt) is not a peer to a biotech firm (asset-light, volatile, often unprofitable).

Filter 2: Business model

Companies in the same industry can have vastly different business models. A software company selling perpetual licenses has a different economic profile than one selling subscriptions. A manufacturer with in-house production has a different model than one that outsources to contract manufacturers.

For example, in industrials:

• An asset-light company that designs and outsources manufacturing (like many fabless semiconductor firms) should not be grouped with an asset-heavy company that owns factories.
• A company with recurring revenue (e.g., a maintenance contract business) should not be grouped with one that has one-time project revenue.
• A company with 80% gross margins should not be compared to one with 35% gross margins, unless you understand why.

Filter 3: Scale (size)

Company size (by revenue, market cap, or EBITDA) often correlates with valuation multiples. Larger companies tend to trade at higher multiples than smaller ones, all else equal, because they are more stable, more liquid, and more predictable. A $50 billion software company trades at a different multiple than a $500 million software company, even if growth rates are similar.

Include peers that are within roughly 0.5x to 2x the target's scale. If your target has $500 million of revenue, peers with $250 million to $1 billion revenue are appropriate. Peers with $50 million revenue or $5 billion revenue are stretches.

Filter 4: Growth rate

A fast-growing company trades at a higher multiple than a slow-growing company with identical profitability. This is such a strong effect that it often dominates other factors. A 20%-growth-rate software company should not be grouped with a 3%-growth-rate one.

Define a reasonable growth band. If your target is growing at 10% per year, look for peers growing at 8–12%. If your target is in a mature, low-growth industry (e.g., utilities), look for peers with similar 2–4% growth. Do not force a 20% growth story to fit a 3% peer set; they are not comparable.

Filter 5: Profitability and return on capital

Two companies can have the same revenue and growth rate but vastly different profitability. One might have 20% EBITDA margins; the other might have 5%. This difference is fundamental and will drive different multiples.

Similarly, return on invested capital (ROIC) varies widely. A company generating 15% ROIC on reinvested capital is not a peer to one generating 8% ROIC, even if growth rates are identical, because the quality of earnings (and thus the multiple it deserves) is different.

Filter 6: Capital structure and leverage

Companies with very different debt levels might have different multiples. A highly leveraged company might look cheap on an EV/EBITDA basis but actually be expensive to equity holders. A debt-free company looks expensive on that metric but might be a safer equity investment.

Typically, you want peers with similar leverage ratios (e.g., all companies with debt-to-EBITDA below 2x, or all with net debt above 3x). If your target has very high or very low leverage for its industry, seek peers with similar leverage.

A visual framework for peer selection

A practical example: building a peer set for a mid-cap retailer

Let's say your target company is Main Street Retail, a mid-cap specialty retailer with $2 billion in revenue, operating 500 stores across North America, focusing on apparel and accessories. Profitability: 7% EBITDA margins, growing 4% per year. Debt-to-EBITDA: 2.5x.

Step 1: Industry filter

Start with all companies in the Specialty Retail industry (GICS 304030). This gives you maybe 20–30 public companies. That is a starting universe.

Step 2: Business model filter

You realize that some companies are omnichannel (stores + e-commerce), while others are pure-play e-commerce. Some focus on apparel, others on electronics or home goods. You filter for traditional retailers with stores as their primary channel and apparel/accessories as their core category.

This cuts the list to 15 companies: specialty apparel retailers like Ulta Beauty, Gap Inc., Abercrombie & Fitch, Foot Locker, etc.

Step 3: Scale filter

Your target has $2 billion revenue. You want peers with $1 billion to $4 billion. Ulta Beauty ($9 billion revenue) is too large. Small indie brands are too small. You are left with 8–10 companies.

Step 4: Growth filter

Your target grows at 4% per year. This is mature, steady-state apparel retail growth. You exclude hot growth plays (if any in your peer list) and focus on peers with 2–6% revenue growth. This keeps you at 8 peers.

Step 5: Profitability filter

Your target has 7% EBITDA margins. You exclude any peers with margins below 4% or above 12%, since they likely have different business models. You end up with 6–7 solid peers.

Step 6: Leverage filter

Your target has 2.5x debt-to-EBITDA. You include peers with 1.5x to 3.5x, excluding those with very high leverage (distressed) or very low leverage (asset-light, online-only).

Final peer set for Main Street Retail:

1. Gap Inc.
2. Abercrombie & Fitch
3. Foot Locker
4. Lululemon Athletica (borderline—higher growth, included)
5. Nordstrom
6. Dick's Sporting Goods
7. Urban Outfitters

You now have 7 solid peers. You can look up their current trading multiples and extract an EV/EBITDA multiple to apply to Main Street Retail.

Real-world peer sets: the limits of perfection

In practice, no peer set is perfect. Your target has some unique characteristics that no public company matches exactly. This is OK. The goal is "best available," not "perfect." Here are some realistic challenges:

Challenge 1: No direct public peers exist

Some industries are dominated by a single large public company or by private companies. If you are valuing a niche manufacturer with no direct public peers, you must either:

• Expand your peer set to include somewhat less-similar companies (e.g., include larger, faster-growing, or lower-margin peers and note the adjustment)
• Look for transaction comps (recent M&A of similar companies) instead
• Rely more heavily on DCF and less on comps

Challenge 2: Peers are in different geographies

An industrial company with only European operations might have no perfect U.S. peer. You can include U.S. peers but note that currency exposure, tax rates, and regulatory environments are different. You might apply a small multiple discount (5–15%) to account for geographic risk differences.

Challenge 3: Peers have different revenue mixes

A diversified company might have 60% of revenue from Product A and 40% from Product B. True peers might have 80% / 20% or 40% / 60%. You can include these peers but should consider running separate analyses for each business segment and weighting the results.

Challenge 4: One peer is an outlier

You have seven peers, and one is trading at 2x the median multiple on the same EBITDA. Why? Maybe it is a special situation, a distressed seller, or a deeply undervalued stock. Maybe it is overvalued. Either way, it will skew your median. Consider excluding it (if you have enough peers) or, at minimum, using median instead of mean (median is more robust to outliers).

Documentation and transparency

Build a peer set memo that explains:

1. Target company overview: Industry, business model, scale, growth, profitability, leverage
2. Selection criteria: What filters you applied and why
3. Peer list: Name, revenue, EBITDA, growth rate, debt-to-EBITDA, and the specific reason each was included
4. Peers excluded: Companies you considered but excluded, and why (e.g., "Too small—only $200M revenue; order-of-magnitude below target")
5. Outlier notes: If any peer is an outlier on one dimension, note it

This transparency allows others—your boss, an audit committee, another analyst—to review and critique your peer set. They might say, "You excluded company X, but I think it should be included because..." and you can revise. Or they might say, "Peer Y is trading at 3x the multiple of the others; did you investigate why?" and you can dig deeper.

Common mistakes in peer selection

Mistake 1: Over-reliance on industry classification

Just because two companies are in the same GICS sector does not make them peers. A software company and a data center company are both in "Technology," but they have completely different economics. Use industry classification as a starting point, not an ending point.

Mistake 2: Including peers that are too different on one key dimension

If your target is a 15% EBITDA margin business and you include a 4% EBITDA margin peer because it is "in the same industry," you will distort your multiple. Profitability is too fundamental to overlook.

Mistake 3: Using too many peers

Fifteen or twenty peers sounds like a robust sample, but it dilutes the "comparable" standard. You are essentially averaging together companies that are only loosely related. Aim for 5–10 high-quality peers rather than 15 mediocre ones.

Mistake 4: Ignoring size and scale effects

A $500 million company is not a peer to a $50 billion company, even if they are in the same industry. Large companies trade at premium multiples due to liquidity, stability, and scale advantages. Include size-appropriate peers.

Mistake 5: Forcing historical peers when the business has changed

If your target pivoted its business model last year (e.g., from a transactional model to a subscription model), peers from three years ago are stale. Find new peers that reflect the current business model, even if they are new to your peer set.

Mistake 6: Using transaction comps peers that are not representative

If your target is a mid-cap industrial company and the only recent M&A deal in your space was a distressed seller being acquired at a fire-sale price, that one transaction comp will skew your entire valuation. Use it, but weight it carefully and supplement with multiple deal examples.

Segment-by-segment peer sets

Some companies operate multiple business segments with different profiles. A conglomerate might have a fast-growing tech division, a mature manufacturing division, and a slow-growing utility division. These cannot share a peer set.

In such cases, build separate peer sets for each segment:

1. Value each segment using its own peer set (and DCF)
2. Sum the segment values to get total enterprise value
3. This "sum-of-the-parts" valuation often differs from a single-peer-set approach and is more accurate

For example, valuing Diversified Corp with three divisions:

• Division A (Tech): 30% of revenue, 40% of EBITDA, 25% growth → Peer set: high-growth tech companies
• Division B (Manufacturing): 50% of revenue, 45% of EBITDA, 3% growth → Peer set: industrial manufacturers
• Division C (Utilities): 20% of revenue, 15% of EBITDA, 2% growth → Peer set: regulated utilities

Value each segment separately, then sum the equity values.

FAQ

Q: How do I know if a company should be in my peer set?

A: Ask yourself: "If I replaced the target company's name with this peer's name in my analysis, would the analysis still make sense?" If you are unsure, it probably should not be in the set.

Q: Should I include foreign (international) peers?

A: Yes, if the target competes globally or if no good domestic peers exist. However, be aware of differences in accounting standards (IFRS vs GAAP), tax rates, and regulatory environment. A U.S. bank and a European bank operate under different capital requirements and thus may have different profitability profiles; this should be disclosed.

Q: What if my target is much smaller than all potential peers?

A: You have two options: (1) Use the smallest peers available and apply a small-cap discount (typically 10–20%), noting the adjustment, or (2) Build a separate valuation using transaction comps (smaller, more recent M&A deals) instead of trading comps. Both are legitimate approaches.

Q: How many peers do I need for a robust analysis?

A: Five to ten is ideal. With five, you have enough diversity but still manageable detail. With ten, you have good cushion against one or two outliers. Above ten, you are diluting the "comparable" standard. Below five, you are vulnerable to being skewed by an outlier or by a peer that is not truly comparable.

Q: Should I use the mean or the median multiple?

A: Always median. It is more robust to outliers. If your five peers have multiples of 8x, 8.5x, 9x, 10x, and 20x (where that 20x is an outlier), the median is 9x (better) while the mean is 11x (skewed by the outlier).

Q: What if my peer set shows a very wide range of multiples (e.g., 6x to 15x EV/EBITDA)?

A: That wide range signals that your peer set is not homogeneous. You have included companies with genuinely different characteristics. Investigate why: differences in growth, profitability, leverage, or business model. Then either (1) narrow your peer set to exclude the outliers, (2) segment your peers into sub-groups based on the characteristic driving the difference, or (3) document that the wide range reflects true differences and pick the narrow band that best matches your target's profile.

Q: Can I use peers from different countries?

A: Yes, especially for multinational companies. A global software company peer set might include U.S., European, and Asian software firms. Be transparent about geographic differences and note any adjustments for accounting standards, currency, or regulatory differences.

Related concepts

• Valuation multiples: The ratios (EV/EBITDA, P/E, etc.) extracted from peers and applied to the target
• Business model comparability: The importance of ensuring peers have similar economics
• Scale effects on multiples: Why larger companies often trade at premium multiples
• Growth rate adjustment: How to adjust multiples when peer growth differs from target growth
• Segment valuation: Valuing a multi-division company by valuing each division separately

Summary

Building a peer set is both art and science. Use industry classification as a starting point, then apply secondary filters: business model, scale, growth, profitability, and leverage. Aim for 5–10 peers that are genuinely comparable on the dimensions that drive valuation. Document your selection rationale so others can scrutinize and improve your work. Acknowledge that no peer set is perfect; the goal is "best available" not "perfect." Use median multiples, not mean. Be prepared to defend each peer inclusion and exclusion in plain language. A good peer set is the foundation of credible comparable company analysis; a poor one is garbage in, garbage out.

Next

Read the next article, Screening comps from a starting universe, to learn the practical mechanics of filtering a large universe of companies down to a focused, defensible peer set.