How to Select Precedent Transactions for a Defensible Valuation
Selection is the hardest step in precedent transaction analysis. Here's how M&A practitioners filter for comparability, handle a thin set, and present a defensible range.
The biggest source of error in a precedent transaction analysis is not the multiples themselves — it is the selection of deals those multiples are drawn from. A banker can pull fifteen transactions from a screen, cherry-pick the five with the highest multiples, and produce a range that bears little relationship to what the market would actually pay for the company in front of them. Practitioners on the other side of the table have seen this enough times to push back hard on any set that looks curated.
Building a defensible set means applying selection criteria transparently, documenting the logic, and being consistent enough that a counterparty or board can reconstruct your reasoning. This is not the same as applying the most generous criteria possible; it is applying criteria that a reasonable person would agree belong in the analysis.
When an LLM is helping with the screen, those criteria are only one control. Why finance AI valuation needs sourced precedent transaction data explains how status, value type, currency, financial period and source provenance need to travel with every selected deal.
Start with four primary screening criteria
Most practitioners begin with four variables that define whether a transaction belongs in the initial pool at all.
Sector and subsector. Multiples are not portable across industries. A software company trading at 14× EBITDA and a distribution business trading at 7× are not interchangeable references even if both are labeled “technology-adjacent.” Start at the industry level, then narrow to subsector if you have enough transactions. A medical device business is not the same as a healthcare services business; a B2B SaaS company is not the same as a managed-service provider. The more precise your subsector filter, the fewer transactions you will typically find, so you need to decide early how much precision you can afford given the size of your database.
Deal size. Smaller companies trade at lower multiples than larger ones in the same sector — typically because they are less liquid, more dependent on key people, and more exposed to customer concentration risk. The practical rule is to filter for targets within roughly one order of magnitude of your subject company by revenue or EBITDA. A $10 million EBITDA business should not be compared against $1 billion EBITDA transactions if you can avoid it, and vice versa. When your subject is in a thin size band, you may have to widen the range and apply a size-discount adjustment explicitly, rather than leaving it implicit in an average that does not distinguish.
Time period. Market multiples shift with credit conditions, interest rates, and sector appetite. Deals from 2020–2021 reflect a financing and valuation environment that may be meaningfully different from today. A common starting point is five years of completed transactions, with closer scrutiny of any deal more than three years old. When market conditions have changed sharply, a shorter window is more defensible even if it produces fewer comparables. Include the date of each transaction in your analysis so the reader can see the temporal distribution of the set.
Buyer type. Strategic acquirers with synergies routinely pay more than financial sponsors paying with leverage. Including both in the same undifferentiated set will widen your range artificially and reduce its explanatory power. Where possible, identify whether each transaction was a strategic acquisition, a private-equity buyout, or a secondary PE transaction. Presenting a sub-table that separates these can be more useful to a client than a blended average that obscures the driver of the spread. Strategic buyer vs. financial sponsor multiples in precedent transactions covers the structural reasons for the gap, when to segment versus blend, and how to identify buyer type from SEC filings.
Expand or narrow from the initial screen
Your first screen will return either too many transactions or too few, and the response to each is different.
When the initial set is large — more than fifteen to twenty deals — you can tighten your criteria. Move from industry to subsector. Narrow the size band. Shorten the time window. Exclude acquisitions by strategic buyers if your subject is being sold to a sponsor, or vice versa. The goal is not to minimize the set; it is to maximize its relevance.
When the initial set is thin — fewer than five to eight deals — you have to expand it, and the expansion decisions are where practitioners most often diverge. The least defensible approach is to loosen criteria quietly and present the expanded set as if it were the original one. The more defensible approach is to document your expansion step by step: “We began with subsector X in size band Y over the last three years. That produced two transactions. We then expanded to the broader industry and to five years, which brought in six additional transactions, and applied adjustments for the following differences.” A reader may disagree with your expansion decisions; they cannot dispute that you disclosed them.
What to do with outliers
Every set of multiples has transactions at the high and low end. The question is whether they belong, not whether they are inconvenient.
An outlier may be legitimate. A hostile bid, a distressed sale, or a transaction completed during an unusual market period can produce a multiple that no longer reflects what an arm’s-length buyer would pay. In those cases, excluding it — with an explanation — is reasonable. The explanation is not optional. A set with outliers removed and no disclosure is grounds for credibility questions later.
An outlier may also be revealing. If one transaction in your set trades at twice the multiple of the others, understanding why is more valuable than removing it. The answer might be a synergy-driven premium, a control contest, a materially different business model, or a different financial period used to calculate the multiple. Any of these could be relevant to how you interpret your subject company’s positioning.
The standard output is to show the full set, note any exclusions with reasons, and present a trimmed mean or median alongside the full range. Showing the minimum, 25th percentile, median, 75th percentile, and maximum gives the reader a distribution, not just a point, and it makes your analysis harder to dismiss as selectively presented. Identifying and handling outlier transactions in a precedent set covers the five categories of analytical outlier, the quantitative flag criteria, and the documentation a fairness opinion requires for each exclusion.
Document the financial period for each transaction
This is the step most often skipped and most often questioned. The same deal can produce materially different multiples depending on whether you used trailing twelve-month EBITDA from the last fiscal year, trailing twelve months from the quarter before announcement, or a forward year estimate. If the target was growing quickly, these figures can diverge by 20 to 30 percent.
Before including any multiple in your set, check what period was used to produce it. If your source gives you an EV but not the underlying EBITDA, you will need to reconstruct it from the target’s filings — and note that you did. A multiple labeled only as “EV/EBITDA” without a period is not comparable to a multiple where you know the measurement period precisely.
For the same reason, distinguish between a multiple reported by the parties or a financial source and a multiple you calculated yourself. Reported and calculated multiples can differ when an acquirer adjusts the target’s EBITDA in their announcement, or when transaction costs or assumed liabilities affect the enterprise value differently than your source assumes. EV/EBITDA multiples in precedent transactions require this level of specificity to be reliable.
Presenting the range
A defensible range has three features: the selection criteria are stated, the underlying transactions are listed, and the reader can see the spread rather than just a single number.
The range will almost always be wide enough to be uncomfortable. That discomfort is real information. If the range for your sector spans 6× to 14× EBITDA, the relevant question is where in that range your subject company belongs and why — and the answer should come from the company’s fundamentals relative to the set, not from choosing which transactions to include.
For a board presentation or a fairness opinion, the selection methodology is as important as the multiples it produces. A set of transactions chosen carefully and consistently is more useful than a wider set chosen loosely — and it is far easier to defend if a counterparty challenges it.
PrivyLogic is a searchable database of precedent M&A deal terms and valuation multiples. Use its transaction context to identify candidates, then verify material figures against public disclosures; undisclosed fields are left blank rather than inferred. If you are building a transaction set and need comparable deals with available valuation and financial context, see the available plans. For enquiries about enterprise access, contact the team.
For background on what information a useful precedent record contains, see What Is a Precedent Transaction in M&A?