Skip to main content

Comps Analysis Skill

What is Comparable Company Analysis?

Comparable company analysis — universally called “comps” in finance — is a valuation method based on a simple idea: similar companies should be worth similar amounts relative to their earnings, revenue, or other financial metrics. If five enterprise software companies with similar growth profiles trade at a median of 15x EBITDA, then a sixth similar company should trade at roughly the same multiple. This is the financial equivalent of pricing a house by looking at what similar houses in the neighborhood sold for. You would not price a 3-bedroom house in San Francisco by looking at 1-bedroom apartments in Detroit — you would find comparable properties with similar features, location, and condition. The same logic applies to companies. Worked example: Suppose you want to value CloudCo, an enterprise SaaS company with 200Mofrevenueand200M of revenue and 50M of EBITDA. You identify five comparable public SaaS companies:
CompanyEVEBITDAEV/EBITDA
PeerA$3,000M$200M15.0x
PeerB$1,800M$150M12.0x
PeerC$2,500M$160M15.6x
PeerD$900M$75M12.0x
PeerE$4,200M$250M16.8x
Median15.0x
Applying the peer median of 15.0x to CloudCo’s 50MEBITDAimpliesanenterprisevalueof50M EBITDA implies an enterprise value of **750M**. If CloudCo has 30Mofnetcash,theimpliedequityvalueis30M of net cash, the implied equity value is 780M. Comps are arguably the most widely used valuation method in finance because they are intuitive, market-based, and fast to produce. Every investment banking analyst learns to build comps in their first weeks on the job. But the simplicity is deceptive — the quality of a comps analysis depends entirely on two things: selecting truly comparable peers and choosing the right metrics. Why compare companies at all? Because no company exists in a vacuum. Investors allocate capital across alternatives. If Company A trades at 20x EBITDA and Company B (with similar fundamentals) trades at 12x, there is either a reason for the premium (faster growth, stronger moat, better management) or Company A is overvalued and Company B is undervalued. Comps help you identify these discrepancies.

Detailed Worked Example

Let us build a complete comps analysis for a fictional mid-cap SaaS company, DataFlow Inc. (ticker: DFLW), with LTM revenue of 320M,LTMEBITDAof320M, LTM EBITDA of 64M, and a market cap of $1,500M.
1

Select Peers

We identify 5 comparable SaaS companies based on similar business model (B2B SaaS), scale (200M200M-800M revenue), and geography (US-based):
CompanyTickerRevenueDescription
CloudSyncCSYN$280MEnterprise data integration
PlatformXPLTX$450MB2B workflow automation
MetricHubMHUB$310MAnalytics platform
FlowStackFSTK$190MAPI management
DataBridgeDBRG$520MEnterprise middleware
2

Gather Operating Data

CompanyRevenueRev GrowthGross MarginEBITDAEBITDA Margin
CSYN$280M22%74%$56M20.0%
PLTX$450M18%71%$108M24.0%
MHUB$310M25%78%$68M21.9%
FSTK$190M30%80%$34M17.9%
DBRG$520M14%68%$135M26.0%
Max30%80%26.0%
75th %ile25%78%24.0%
Median22%74%21.9%
25th %ile18%71%20.0%
Min14%68%17.9%
DFLW$320M20%72%$64M20.0%
DataFlow’s growth (20%) is near the peer median (22%), and its EBITDA margin (20.0%) is at the 25th percentile. This suggests DataFlow should trade at or slightly below the peer median multiple.
3

Gather Valuation Data

CompanyMkt CapEVEV/RevenueEV/EBITDAP/E
CSYN$2,100M$2,000M7.1x35.7x42.0x
PLTX$3,800M$3,600M8.0x33.3x38.5x
MHUB$2,500M$2,400M7.7x35.3x45.2x
FSTK$1,900M$1,850M9.7x54.4xN/M
DBRG$3,200M$3,400M6.5x25.2x28.1x
Max9.7x54.4x45.2x
75th %ile8.0x35.7x43.6x
Median7.7x35.3x40.3x
25th %ile7.1x33.3x33.3x
Min6.5x25.2x28.1x
4

Apply Multiples to DataFlow

Using the peer median multiples:
EV/Revenue: $320M x 7.7x = $2,464M implied EV
EV/EBITDA:  $64M x 35.3x = $2,259M implied EV

Average implied EV = ($2,464M + $2,259M) / 2 = $2,362M

DataFlow net debt = $50M
Implied equity value = $2,362M - $50M = $2,312M

Diluted shares = 40M
Implied price per share = $2,312M / 40M = $57.80

Current stock price = $37.50 ($1,500M market cap)
Implied upside = ($57.80 - $37.50) / $37.50 = 54.1%
Interpretation: The comps suggest DataFlow is significantly undervalued relative to peers. However, DataFlow’s below-median margins may justify some discount. At the 25th percentile EV/EBITDA (33.3x), the implied EV would be 2,131Mandtheimpliedprice2,131M and the implied price 52.03 — still meaningful upside.

Why It Matters

Comps analysis is used in nearly every financial services workflow:
  • Investment bankers use comps to advise clients on fair offer prices for M&A transactions. “Based on where your peers trade, a fair acquisition price would be 12-14x EBITDA.”
  • Equity researchers use comps to determine whether a stock is over- or under-valued relative to its peer group.
  • IPO teams use comps to set the offering price for new public listings — the “IPO valuation” is largely derived from where comparable public companies trade.
  • PE firms use comps to evaluate entry and exit multiples for potential acquisitions.
  • IC presentations almost always include a comps table to provide market context for the valuation discussion.
If you do comps poorly — selecting inappropriate peers, using the wrong time periods, or mixing up metrics — you will anchor your analysis on misleading numbers and reach flawed conclusions.

Key Concepts

TermDefinitionWhy It Matters
EV/EBITDAEnterprise Value divided by EBITDA. The most commonly used valuation multiple.Allows comparison across companies with different capital structures because EV accounts for debt and EBITDA excludes interest.
EV/RevenueEnterprise Value divided by Revenue. Used for high-growth or unprofitable companies.When a company has negative EBITDA, you cannot use EV/EBITDA. Revenue multiples are the fallback.
P/E RatioPrice per share divided by Earnings per share (or Market Cap / Net Income).The classic equity valuation metric. Easy to understand but distorted by capital structure, tax rates, and non-recurring items.
Enterprise Value (EV)Market Cap + Net Debt. The total value of the business available to all capital providers.Using EV in multiples (vs. just market cap) makes comparisons capital-structure-neutral.
LTMLast Twelve Months. The most recent four quarters of financial data.Ensures you are comparing companies on the most current data, not outdated annual figures.
NTMNext Twelve Months. Forward estimates based on analyst consensus.Forward multiples incorporate expected growth, making them more relevant for high-growth companies.
MedianThe middle value when all data points are sorted.More robust than average (mean) because it is not skewed by outliers. Always use median for comps statistics.
Rule of 40Revenue Growth % + EBITDA Margin %. Used for SaaS companies.A SaaS company scoring above 40 is considered healthy. Below 40 signals either insufficient growth or poor profitability.

How It Works

Triggers automatically when: users need to build a comparable company analysis, benchmark a company against peers, or establish market-based valuation ranges.

Step 1: Clarify Purpose

Always establish context first:
  • “What is the key question?” — Valuation, growth comparison, efficiency benchmarking?
  • “Who is the audience?” — IC presentation, quick reference, detailed memo?
  • “What is the context?” — M&A advisory, investment decision, sector overview?
Adapt the analysis depth and metrics based on answers. A 20-slide IC deck and a quick Slack response are both “comps” but take completely different shapes.

Step 2: Select Peers

Select 4-6 comparable companies based on:
  • Similar business model — Same industry, similar revenue mix
  • Similar scale — Market cap within a reasonable range (typically 0.5x to 3x the target)
  • Same industry/sector — GICS classification as a starting point
  • Geographic comparability — US-to-US, not US-to-emerging-market
When in doubt, exclude the company. Better to have 3 perfect comps than 6 questionable ones. A comp that is not truly comparable will distort your statistics and mislead your audience.

Step 3: Gather Data

Using available MCP sources (S&P Global, FactSet, Daloopa preferred), pull:
  • Operating metrics: Revenue, Growth, Gross Margin, EBITDA, EBITDA Margin
  • Valuation: Market Cap, Enterprise Value, EV/Revenue, EV/EBITDA, P/E
  • Industry-specific metrics (Rule of 40 for SaaS, ROE for financials, etc.)
Data Source Hierarchy:
  1. MCP data sources (S&P Global, FactSet, Daloopa) — use exclusively if available
  2. Bloomberg Terminal, SEC EDGAR filings — only if MCPs are unavailable
  3. Never use web search as a primary data source for institutional-grade analysis
MCP sources provide verified, institutional-grade data with proper citations. Web search results can be outdated, inaccurate, or unreliable.

Step 4: Build the Analysis

Operating Statistics Section:
  • Company data rows with revenue, growth, margins
  • Statistics block: Max, 75th Percentile, Median, 25th Percentile, Min
Valuation Multiples Section:
  • Market Cap, EV, EV/Revenue, EV/EBITDA, P/E
  • Same statistical summary
Industry-Specific Metrics:
IndustryMust-Have Metrics
Software/SaaSRevenue Growth, Gross Margin, Rule of 40, ARR, Net Dollar Retention
ManufacturingEBITDA Margin, Asset Turnover, CapEx/Revenue
Financial ServicesROE, ROA, Efficiency Ratio, P/E
RetailRevenue Growth, Gross Margin, Inventory Turnover, Same-Store Sales
HealthcareR&D/Revenue, Pipeline Value
The “5-10 Rule”: 5 operating metrics + 5 valuation metrics = 10 total columns. Enough to tell the story without noise. If you have more than 15 metrics, edit ruthlessly.

Step 5: Formatting and Delivery

Formula rules:
  • Every derived value (margin, multiple, statistic) MUST be an Excel formula referencing input cells
  • The only hardcoded values are raw input data (revenue, EBITDA, share price)
  • Every hardcoded input gets a cell comment with its source
  • Blue font = inputs, Black font = formulas
Statistics that need statistical summary: Revenue Growth %, Gross Margin %, EBITDA Margin %, EV/Revenue, EV/EBITDA, P/E Statistics that do NOT need summary: Revenue ,EBITDA, EBITDA , Market Cap ,EV, EV (not comparable across different-sized companies) Verify step-by-step with the user:
  • After setting up the structure, show the header layout before filling data
  • After entering raw inputs, show the input block and confirm sources/periods
  • After building operating metrics formulas, show calculated margins and sanity-check
  • After building valuation multiples, show multiples and confirm reasonableness
  • Do NOT build the entire sheet end-to-end and then present it

Formatting Standards

ElementFillFont
Section headersDark blue #1F4E79White bold
Column headersLight blue #D9E1F2Black bold
Data rowsWhiteBlack (formulas), Blue (inputs)
Statistics rowsLight grey #F2F2F2Black
That is the whole palette: dark blue + light blue + light grey + white. Nothing else.

Sanity Checks

Before delivering, verify:
  • Margin test: Gross margin > EBITDA margin > Net margin (always true by definition)
  • Multiple reasonableness: EV/Revenue 0.5-20x, EV/EBITDA 8-25x, P/E 10-50x
  • Growth-multiple correlation: Higher growth usually means higher multiples
  • Negative EBITDA: If a company has negative EBITDA, its EV/EBITDA is meaningless — use EV/Revenue instead

Cross-Reference Rule

Valuation multiples MUST reference the operating metrics section. Never input the same raw data twice. If revenue is in C7, then EV/Revenue formula should reference C7.

Common Mistakes

The mistake: Using Market Cap / EBITDA instead of EV / EBITDA, or mixing EV-based and equity-based metrics in the same comparison.Why it happens: Market cap and EV are both “how much the company is worth” and get confused.The fix: EV = Market Cap + Net Debt. EV-based multiples (EV/EBITDA, EV/Revenue) pair with pre-interest metrics. Equity-based multiples (P/E) pair with post-interest metrics. Never cross them.
The mistake: Comparing Company A’s LTM revenue to Company B’s FY2024 revenue to Company C’s Q4 annualized revenue.Why it happens: Different data sources report different periods, and it is tempting to use whatever is available.The fix: All companies must use the same period. LTM (last twelve months) is preferred because it is always current. If using fiscal year data, ensure all companies have the same fiscal year end.
The mistake: Including a hardware manufacturer in a SaaS comp set because both are “technology companies.”Why it happens: Overly broad industry definitions. GICS “Information Technology” includes companies with fundamentally different business models.The fix: Focus on business model similarity, not sector labels. A SaaS company’s comps are other SaaS companies with similar revenue scale, growth profiles, and end markets. A hardware company has different margins, capital intensity, and growth dynamics.
The mistake: Reporting the average EV/EBITDA as 18.5x when the median is 15.0x, because one outlier trades at 45x.Why it happens: Averages are the default in many analyses and feel more inclusive.The fix: Always use median for comps statistics. One outlier can dramatically skew an average. Include the average as supplementary information if needed, but anchor on the median.
The mistake: Computing EV/EBITDA for a company with negative EBITDA and including that negative multiple in the statistics.Why it happens: The formula mechanically produces a number (a negative one), and the analyst does not notice.The fix: If EBITDA is negative, EV/EBITDA is meaningless. Show “N/M” (not meaningful) instead. Use EV/Revenue for companies that are not yet profitable.
The mistake: Typing =3000/200 in the EV/EBITDA cell instead of =C7/G7.Why it happens: Faster to type the numbers directly, especially when building quickly.The fix: Every derived value must reference source cells. Hardcoded multiples cannot be audited, updated, or traced to their inputs. If EV changes, the hardcoded multiple silently becomes wrong.
The mistake: Entering revenue of $280M in a cell with no documentation of where that number came from.Why it happens: Adding comments feels tedious when building quickly under deadline pressure.The fix: Every hardcoded input gets a cell comment citing the exact source: “Source: FactSet, Q4 2025 10-K, Page 42” or “Bloomberg Terminal - CSYN Equity DES, accessed 2025-12-15.” Include hyperlinks when possible.
The mistake: Computing the median revenue across the peer set and presenting it as meaningful.Why it happens: The statistics block is applied to every column without thinking about which columns are comparable.The fix: Only compute statistics (median, quartiles) for comparable metrics: growth rates, margins, and multiples. Revenue in absolute dollars varies by company size and is not comparable. Showing “median revenue = $310M” is meaningless.
The mistake: Using a stock price from three weeks ago to calculate market cap and EV.Why it happens: The analyst built the comps weeks ago and did not refresh the market data before presenting.The fix: Always update stock prices, market caps, and EV calculations on the day of presentation. Add an “As of [Date]” label to the header. If the analysis is for a client meeting on Thursday, refresh the data Thursday morning.
The mistake: Comparing a company with a June fiscal year end to companies with December fiscal year ends without adjusting.Why it happens: The analyst pulls “FY2024” data for all companies without realizing that Company A’s FY2024 ended in June while the others ended in December.The fix: Use LTM (last twelve months) data instead of fiscal year data when fiscal year ends differ. This ensures all companies are measured over the same approximate time period. If using fiscal year data, flag the discrepancy.

Daily Workflow

You cover the enterprise software sector and need to update your comps table after Q4 earnings season.Workflow:
  1. Pull updated LTM financials for all peer companies from FactSet or S&P Global MCP
  2. Update market caps and enterprise values with current stock prices
  3. Recalculate all multiples (the formulas should auto-update if the model is properly built)
  4. Review the statistics block — did any company move significantly relative to peers?
  5. Check for any peer set changes — did any company get acquired, go private, or change its business model?
  6. Update the “As of” date on the header
  7. Send the updated comps to your team with a 2-sentence summary of key changes
Your team is advising a SaaS company on its IPO. You need to build a comps analysis to support the offering price range.Workflow:
  1. Select 6-8 publicly traded SaaS companies with similar growth profiles, scale, and end markets
  2. Build operating metrics (Revenue, Growth, Gross Margin, EBITDA Margin, Rule of 40, NRR)
  3. Build valuation multiples (EV/Revenue NTM, EV/EBITDA NTM — use forward multiples for growth companies)
  4. Calculate the implied valuation range: apply the 25th-75th percentile EV/Revenue to the IPO candidate’s NTM revenue
  5. Present the range to the underwriting team: “Based on peer multiples of 6.5-9.0x NTM revenue, the implied enterprise value is 1.3B1.3B-1.8B”
  6. Discuss whether a premium or discount to the median is appropriate based on growth, margins, and competitive position
A client’s board needs to evaluate whether a $2.5B acquisition offer is fair. You need comps to support the fairness opinion.Workflow:
  1. Build a comprehensive peer set (8-10 companies) for maximum statistical robustness
  2. Include both LTM and NTM multiples to capture different perspectives
  3. Calculate the implied valuation range at the 25th and 75th percentile multiples
  4. Compare the offer price ($2.5B) to the comps-implied range
  5. If the offer falls within the range, the price is “within the range of fairness”
  6. Document everything meticulously — fairness opinions face legal scrutiny

Practice Exercise

Scenario: Build a comps analysis for MediTech Corp, a healthcare IT company with 150MLTMrevenue,150M LTM revenue, 22.5M LTM EBITDA, and a current market cap of 600M(netdebtof600M (net debt of 25M). Given peer data (LTM):
CompanyRevenueGrowthGross MarginEBITDAEBITDA MarginMkt CapNet Debt
HealthSoft$200M18%65%$40M20.0%$1,100M$50M
CareData$120M22%70%$18M15.0%$750M-$30M
MedCloud$180M15%62%$36M20.0%$850M$80M
ClinicAI$90M28%75%$9M10.0%$500M-$20M
PharmaTech$250M12%60%$62.5M25.0%$1,400M$100M
Your tasks:
  1. Calculate EV and valuation multiples (EV/Revenue, EV/EBITDA, P/E) for each peer
  2. Compute statistics (Max, 75th %ile, Median, 25th %ile, Min) for growth, margins, and multiples
  3. Apply the median EV/Revenue and EV/EBITDA multiples to MediTech to get an implied valuation range
  4. Determine where MediTech’s current market valuation sits relative to the peer-implied range
  5. Explain whether MediTech appears over- or under-valued and identify which metrics drive the gap
Remember: CareData and ClinicAI have negative net debt (net cash), which means their EV is lower than their market cap. EV = Market Cap + Net Debt, so net cash is subtracted. Watch for ClinicAI’s low EBITDA margin (10%) — it may warrant a discount to the peer median or use of EV/Revenue instead of EV/EBITDA.

How to Add to Your Local Context

1

Install the Plugin

claude plugin install financial-analysis@financial-services-plugins
2

Customize Peer Selection

Edit skills/comps-analysis/SKILL.md:
## Firm-Specific Peer Selection Rules
1. Always include our core coverage universe first
2. Exclude companies with market cap < $500M
3. All comps must include NTM estimates, not just LTM
4. For healthcare, check with sector team before finalizing peers
3

Set Preferred Metrics

Add your firm’s standard metrics:
## Our Standard Metrics
- Always include FCF Yield alongside P/E
- Report both LTM and NTM multiples
- Include Net Dollar Retention for all SaaS comps
4

Configure Data Sources

Edit .mcp.json to point to your firm’s data provider. If your firm uses Bloomberg instead of FactSet:
{
  "mcpServers": {
    "bloomberg": {
      "url": "https://mcp.bloomberg.com/mcp",
      "headers": { "Authorization": "Bearer YOUR_KEY" }
    }
  }
}

Common Pitfalls

  • Mixing market cap and enterprise value — EV/EBITDA and P/E are fundamentally different metrics because EV includes debt and market cap does not. Do not mix them in the same comparison.
  • Inconsistent time periods — All companies must use the same period (all LTM or all FY2024). Mixing quarterly and annual data produces meaningless comparisons.
  • Including non-comparable companies — A SaaS company and a hardware manufacturer are not comps, even if they are in the same “technology” sector.
  • Using averages instead of medians — One outlier can dramatically skew an average. Always use median for comps statistics.
  • Ignoring negative EBITDA — If a company has negative EBITDA, its EV/EBITDA multiple is meaningless. Use EV/Revenue instead.
  • Hardcoding numbers into formulas — Every derived value must reference source cells. Hardcoded multiples cannot be audited or updated.