How investors triangulate valuation from comps

The question with no honest answer

A founder sits across from an investor and asks the question everyone asks: "What should our valuation be?" It feels like a request for a fact, like asking what the weather will be. The investor gives the answer everyone gives, which is some version of "it depends," and the founder leaves thinking the number is arbitrary, set by vibes and confidence and whoever blinks first.

That read is wrong, and believing it is expensive. The investor was not being evasive. They were telling you the truth in a compressed form. The number does depend, on a specific list of things, and the investor already knows that list because they price companies for a living. You walked in treating valuation as a single number to defend. They are treating it as the output of a calculation they run on every deal, and you are one input.

Here is what actually happened in the silence after your question. The investor pulled up a mental comp set: three to six companies they consider similar to yours by stage, sector, and revenue model. They anchored on what those companies raised at. Then they started adjusting, moving your number up for the risks you have removed and down for the risks you have not. By the time they said "it depends," they had a range. They just were not going to hand it to you, because the range is their leverage.

The founders who get a fair price are not the ones with the best story. They are the ones who reconstructed that calculation before the meeting and showed up arguing inside it.

What founders do instead, and why it underprices them

Most founders set their valuation one of three ways, and all three leave money or credibility on the table.

The first is the desired-dilution method. You decide you want to raise $2M and give up no more than 20%, so you announce a $10M pre. The number is real to you because it protects your ownership, but it is disconnected from anything the investor cares about. They hear a round size and an ownership target dressed up as a valuation, and they discount it immediately because it contains no information about the company.

The second is the comparison-to-the-best method. You read that a company in your space raised at $40M and decide your number should be in that neighborhood because you are at least that good. The problem is you are comparing your headline to their headline without any of the adjustments. That company may have had two prior exits among the founders, a year of revenue, and a bidding war. You are matching their price without matching their risk profile, and the investor sees the mismatch in the first two minutes.

The third is the let-the-market-decide method, where you refuse to name a number and wait for term sheets. This is not wrong, but founders who do it without knowing their own comp set cannot tell a good offer from a bad one. They get a term sheet at a number, feel relief, and accept it with no way to know whether they left 40% on the table or got a gift.

All three fail for the same reason. They skip the triangulation the investor is doing and substitute a number the founder wants to be true. You cannot negotiate inside a calculation you have not reconstructed.

The five adjustments investors actually make

Comps are the anchor, not the answer. An investor starts from what similar companies raised at and then moves the number along five axes. Learn the axes and you stop guessing.

Stage and traction. This is the biggest mover and the one founders most often fudge. Pre-revenue with a prototype is a different price than $20k MRR growing 15% month over month, even in the same sector. The investor is not pricing your idea. They are pricing how much of the risk between idea and outcome you have already burned down. Every milestone you can point to, first revenue, first retention cohort, first enterprise logo, moves you up the comp ladder toward the better-priced rows.

Growth rate. Two companies at the same revenue are not the same company if one is growing 20% a month and the other 5%. Investors price the slope, not the point. A high growth rate is the single most defensible reason to argue for the top of your comp range, because it is the hardest input to fake and the one most correlated with the outcomes they are paid to find.

Market size and quality. The comp set assumes a certain ceiling. If your market is genuinely larger or structurally better than the companies you are being compared to, that is an argument for adjustment up, but only if you can show it with a real wedge and an expansion path, not a top-down "1% of a huge number" slide. If your market is smaller or more crowded than the comps assume, the investor adjusts down whether you raise it or not.

Quality of revenue. Not all revenue is priced equally. Recurring, high-retention, high-margin software revenue prices at a premium to one-time, services-heavy, or low-retention revenue at the same dollar amount. An investor will discount $30k MRR with 6% monthly churn well below $20k MRR with negative net churn. The quality of the revenue often matters more than the quantity, and founders quoting only the top-line number miss the adjustment that is being made against them.

Competitive heat. This is the adjustment founders pretend does not exist and investors weight heavily. Heat is real demand in your specific round: multiple investors moving, a credible lead, a tight timeline. Heat moves the number up, sometimes more than fundamentals justify, which is also why it is dangerous. The corollary matters too. A round with no heat gets adjusted down regardless of how good the company looks on paper, because price in venture is partly a function of scarcity and competition, not just quality.

The trap: outlier AI rounds as your comp

There is a specific way this calculation breaks in 2026, and it is worth naming because it is actively misleading founders right now.

You read that an AI company raised at a number that looks unhinged relative to its revenue. You take it as the new baseline and set your comp set accordingly. The problem is that those rounds are outliers priced on factors that do not transfer: a founding team with a specific research pedigree, a perceived land-grab in a category investors are terrified of missing, or a strategic bet where the fund is buying optionality, not pricing fundamentals. Using an outlier as your comp is like pricing your house off the one on the street that sold to a buyer who needed that exact address for reasons that have nothing to do with the house.

When an investor sees a founder anchor on an outlier round, it does the opposite of what the founder intends. It signals that the founder does not understand how pricing works, which makes every other number in the conversation suspect. The fix is to build your comp set from the median of comparable companies, not the maximum, and to treat the outliers as evidence of category heat rather than as your personal price. You can mention the heat. You cannot borrow the number.

The artifact: comp adjustment checklist

Run this before you name a number to anyone. It turns "what should our valuation be" into a defensible range you can argue inside.

Build the comp set

• List 4 to 6 companies genuinely comparable to you by stage, sector, and revenue model. Not the most famous ones. The most similar ones.
• For each, record what you know or can reasonably estimate: round size, valuation, approximate revenue or traction at the time, and how hot the round was. Mark estimates as estimates.
• Throw out the single highest and single lowest. Take the median of the rest as your anchor, not your answer.
• Flag any comp that is an outlier AI or hype round and exclude it from the anchor. Note it separately as category-heat evidence only.

Apply the five adjustments, with direction and reason

• Stage and traction: am I ahead of or behind my comp set's milestones at raise? Up or down, and by what specific evidence?
• Growth rate: is my slope steeper or flatter than the comps? This is your strongest up-argument if steeper. Name the number.
• Market size and quality: larger and better, or smaller and more crowded, than the comps assume? Up or down, with a real wedge, not a top-down slide.
• Quality of revenue: recurring and high-retention, or one-time and churny, relative to the comps? Adjust the effective revenue, not just the headline.
• Competitive heat: do I actually have multiple investors moving and a credible lead, or am I hoping? Be honest. No heat means no up-adjustment, regardless of fundamentals.

Convert to a position you can defend

• State your range, not a point: anchor plus or minus the adjustments, as a band.
• For each up-adjustment, write the one sentence of evidence you will say out loud. "We're at the top of the range because we're growing 18% month over month against a comp set averaging 7%."
• For each down-adjustment the investor will make, write your honest counter or your acceptance. Knowing the down-arguments before they raise them is half the negotiation.
• Decide the number below which you walk, and the number above which you stop pushing because the heat is real and you would rather close.

The decision rule: never name a valuation you cannot decompose into a comp anchor plus five labelled adjustments. If you cannot show your work, the investor assumes there is no work, and prices you accordingly.

What the same company looks like before and after

Take a seed company at $25k MRR. Here is the difference between walking in with a number and walking in with a triangulation.

The numbers are illustrative. The shape is not. The same company, same metrics, gets a different reception depending on whether the founder shows the calculation or hides it behind a single figure.

Where this breaks in practice

The checklist is only as good as the comp data behind it, and that is where most founders stall. The information you need is scattered. The round that a comparable company raised lives in a TechCrunch headline, a half-remembered conversation, a founder friend's offhand comment, an investor's "we passed on someone like you last month." The adjustments live in your own metrics, which are in a dashboard somewhere, and in the investor's private read of the market, which you never see directly.

So the real work of pricing is assembly. You are reconstructing a comp set from fragments, trying to remember which investor said your sector was hot and which one said it was crowded, attempting to hold five adjustments in your head while a partner asks why your number is what it is. The founders who price well are the ones who did this assembly on purpose, in advance, instead of improvising it under questioning.

This is the part of fundraising RoundOS is built for. As you talk to investors, it captures what each one says about your space, your stage, and your number: the offhand "that feels rich for seed," the "we'd want to see another quarter," the comp they mention in passing. Those comments are the market pricing you in real time, and they usually evaporate into your inbox and your memory. Held against the investor they came from, they become a picture of how the market reads your price, so when you sit down to set a range you are working from what investors actually told you, not from a headline you misremembered. The point is not that software invents your valuation. The point is that the market's price logic is already being spoken to you across a dozen conversations, and most of it is lost. The founders who hear it are the ones who priced the round before the round priced them.