Standard valuation training assumes a comp set with a dozen or more relevant public companies and a healthy flow of precedent transaction data. Niche B2B categories rarely offer that. The honest response is not to force a thin set into a false sense of precision. It is to widen the method and be transparent about the resulting range.
A comp set with two or three genuinely relevant names produces a valuation point, not a valuation range, and treating it as more precise than that is how models end up defending a number they cannot actually support under questioning.
Adjacent-vertical comps are underused
When the exact vertical has almost no public comps, software companies serving structurally similar verticals, similar customer concentration profiles, similar switching costs, similar growth ceilings, often trade on comparable logic even if the end market is different. A regulatory compliance software company for credit unions and one for regional insurance agencies may not compete for the same customer, but they share enough structural DNA in growth rate, retention profile, and market ceiling that their multiples inform each other more usefully than forcing a comparison to horizontal SaaS names with none of those characteristics.
Private transaction data has real limitations worth naming explicitly
Precedent private transaction multiples, sourced from databases or advisor networks, are frequently reported without the deal structure details that actually drove the multiple: earnout size, working capital adjustments, whether it was a platform or bolt-on acquisition for the buyer. Citing a precedent multiple without those details attached, or at minimum flagging that they are unknown, overstates the precision of the comparison. A defensible model states plainly what is known and what is not, rather than presenting an unadjusted multiple as if it were fully comparable.
Triangulate rather than pick a single method
In a thin-comp environment, running a DCF, a public comp range (even an imperfect adjacent one), and precedent transactions side by side, then looking at where they converge or diverge, produces a more defensible view than leaning on any single method. Convergence across methods built on different assumptions is genuinely informative. A wide spread across methods is also informative, because it tells the team the valuation carries more uncertainty than a single-point answer would suggest, which is itself useful information for structuring the deal.
What this means for the final number
Present a range with the reasoning behind its width, not a single number dressed up to look more certain than the underlying data supports. An investment committee that understands why the range is wide can price that uncertainty into deal structure, through earnouts or a more conservative entry multiple, in a way that a false-precision single number never lets them do.