3 May 2026
A founder asks an investor what their pre-seed company is worth. The investor names a number. The founder asks how. The honest answer is: I ran four methods, and three of them agree.
Pre-seed valuation feels arbitrary because most people pick one method, usually the one that gives the highest number, and pretend it's a market price. The methods are the same ones professional investors use; the difference is that professionals run all four, look at where they disagree, and the disagreement is the finding. The number you'd defend in a term-sheet conversation is the range where the methods converge — not the high end of any one of them.
Four methods. Each has a known failure mode. None is right alone. Run all four, lay them side by side, and the answer is the band where they agree.
A pre-seed company has almost no past and a wide-open future. Revenue is too small for multiples to mean anything. Future cash flows are too speculative for discounted-cash-flow analysis to work. Most of the equity value is option value — the chance that the company survives five to ten years and ends up worth meaningfully more than today.
That structure breaks the methods that work for public companies. A trading multiple of one times revenue is decisive for a $100M-revenue business and meaningless for a $30K-revenue business. A DCF requires a forecast horizon long enough that the assumption error compounds past usefulness. So pre-seed valuation borrows from a different toolkit, designed for companies whose value is mostly conditional on outcomes that haven't happened yet.
The four methods that work are:
The first three are roughly co-equal at pre-seed; the fourth is a sanity check, not an anchor.
Pull recent rounds in the same geography, stage, and category. The comp set is whatever similar companies actually priced at.
For a software company at pre-seed in Iceland in 2024–2025, recent rounds typically priced at 80–300M ISK pre-money ($580K–$2.2M), with the band tightening by profile. A founding team with no working product is at the lower end. A working product with paying customers and demonstrated retention is near the upper end. The Nordic comp set adds another tier (typically 1.5–2× the Icelandic number for the same profile), and US/EU comps are 2–4×.
This method is the most useful when the comp set is real — four or more recent rounds, with similar product profile and revenue stage, in the same geography. It's the least useful in a thin market or a novel category, where any individual deal can sit far outside the band.
The honest version of this method names sources. The published Kría announcements, Frumtak Crowberry's portfolio updates, the Icelandic startup press, and direct knowledge of recent rounds in the network. Where the comp is anonymised because it was network-known, label it that way. A "comp set" with no named transactions is decoration.
When the comp method works cleanly, it usually produces a tight range — the four or five recent rounds form a cluster, and the company being valued is either inside the cluster or just above or below it. When it doesn't work, the symptom is a range so wide it's not actionable. That's a sign the comp set is wrong, the geography is wrong, or the category is wrong.
Dave Berkus, an active angel, published a method in the early 2000s that's still used. It assigns up to $500K each across five qualitative factors, capping the maximum pre-money at $2.5M.
| Factor | Up to | What it measures | |---|---:|---| | Sound idea | $500K | Is this a real problem with a plausible solution and a market that wants it? | | Prototype | $500K | Is there a working product in front of customers? | | Quality team | $500K | Are the founders capable of executing? Do they have relevant prior wins? | | Strategic relationships | $500K | Distribution partnerships, channel access, advisor network? | | Product rollout / sales | $500K | Are real customers paying real money? |
Score each from 0 to $500K. Sum.
The original method's absolute scale is rough — $2.5M as a ceiling came from US deals two decades ago, and the equivalent today varies by geography. But the relative scoring still works. The method's real value is forcing the analyst to articulate where each factor sits on a defensible scale, in writing, with one or two sentences of reasoning per score.
Berkus is most useful as a ceiling check on the comp method, and as a structured way to compare two pre-seed opportunities. Two companies at the same stage in the same category should have similar Berkus scores; if they don't, the difference is informative.
The method's weakness is precision. A 350K score on "quality team" versus 400K isn't meaningful; the granularity is fake. What's meaningful is whether a factor scores zero, low, mid, high, or maximum. The number is just a way to add them up.
Work back from probability-weighted exit outcomes. The steps:
Define the realistic exit-outcome distribution. For a software company at pre-seed in Iceland, the typical distribution is:
| Outcome | Typical probability | Typical exit value | |---|---:|---:| | Wind-down / fail | 50–70% | $0 | | Acqui-hire / soft landing | 15–25% | $1–10M | | Modest strategic exit (acquisition by category leader) | 10–15% | $10–40M | | Real exit (regional category winner, IPO-track) | 3–8% | $50–200M | | Tail outcome (unicorn-shaped) | 1–3% | $500M+ |
These probabilities are calibrated by company. Strong retention data, international traction, and a defensible niche shift the distribution toward the upper tiers. Weak unit economics and concentrated geography shift it down.
Calculate the expected exit value. Sum across outcomes. For typical Icelandic pre-seed software, the expected value lands somewhere in $5–25M, depending on the company's profile.
Apply a target return multiple. Angels target 8–15× on individual positions. Pre-seed VCs target 20–50×. The multiple has to clear the failure rate, the time value of money, and the opportunity cost of the capital. Patient angel capital can run lower (8–10×); fund-cycle pressured VC capital runs higher (20–30×).
Work back to the implied present valuation.
Required ownership = Target multiple × check size / Expected exit value
Implied post-money = Check size / Required ownership
Implied pre-money = Post-money − Check size
Worked example. Investor writes a $50K check, expected exit value is $13M, target multiple is 10×.
The VC method works when the exit-outcome probabilities can be calibrated against actual recent exits. It fails when the probabilities are guesses. The most common failure: assuming the tail outcome is more likely than it is. Most pre-seed companies do not become regional category winners; they acqui-hire at 5–15M, and the probability-weighted answer is dominated by the boring middle of the distribution.
Apply a market multiple to current ARR. For consumer subscription software at pre-seed: 8–25× ARR. For B2B SaaS: 10–30×. For marketplaces: 1–3× of GMV.
The output is a floor, not an anchor. Pre-seed companies are systematically worth more than their current revenue suggests, because most of the value is option value. A company with $30K ARR isn't worth $300K. It's worth what it would be worth at $1M ARR three years from now, discounted for the probability that it gets there.
The method is most useful as a sanity check. If the comp method or Berkus produces numbers more than 5× the upper revenue multiple, the valuation may be running on hope rather than fundamentals — somebody, somewhere, has assumed unusually high growth or unusually high outcomes. Worth re-examining.
The method fails when ARR is in the single-digit thousands. A 25× multiple of $5K ARR is $125K, and that's not a useful anchor for any conversation.
Run all four. Lay the outputs side by side. The defensible range is usually where the middle two converge — typically the comp method and either Berkus or the VC method. The revenue multiple sets the floor. The Berkus ceiling caps the upside.
What the disagreement tells you:
The triangulation is the finding. Four methods that agree at $1.2M pre-money give you a defensible $1.0–1.5M range. Four methods that disagree from $300K to $3M give you the more useful information that the company is at a stage where one or two specific data points (typically retention or international traction) are doing all the load-bearing work, and the valuation will move sharply once those are known.
Three things.
First: come to the meeting with a range, not a number. A founder who says "150M ISK pre-money" without explaining the range it sits in has either anchored on someone's headline or pulled the number from a US comp set. A founder who says "100–200M ISK depending on the comp set you weight, the four methods triangulate around 130M" is talking the same language as the investor across the table.
Second: be honest about the option value. Most of what makes a pre-seed company valuable is what it could become, not what it is. Founders who try to justify the valuation on current ARR are anchored on the wrong method. Investors who try to negotiate it down to current ARR are using the floor as the anchor. Both miss what the round is for: a bet on the option, priced relative to the realistic exit distribution and the failure rate.
Third: the negotiation is rarely about the number. It's about the lead investor's confidence in the company. A 130M pre-money with a credible lead and a board partner who knows the segment is more valuable to the company than a 200M pre-money with an absentee investor. Founders who optimise for headline pre-money over composition of the round usually regret it within a year.
We disagree with the conventional founder wisdom that "valuation is what matters and the rest is fluff." For a Rich-motivated founder optimising on dilution, valuation is one variable in a five-variable problem. For a King-motivated founder, valuation is barely on the list. The other terms — board, liquidation preference, drag-along — are what determine whether the founder still runs the company in three years.
If you're raising:
If you're investing:
If you want a second read on a valuation, write to us at pitch@valaris.is.
The Berkus method: Dave Berkus, "The Berkus Method: Valuing the Early Stage Investment", berkonomics.com, 2009 update of his original 1990s framework.
The Scorecard method (related to and often combined with Berkus): Bill Payne, "Scorecard Valuation Methodology", 2011. Useful when the comp set is large enough to compute averages.
The VC method: variants taught at most early-stage venture programs; the canonical text is Sahlman, "The Structure and Governance of Venture-Capital Organizations", Journal of Financial Economics 27, 1990. Practitioner-friendly version: Brad Feld and Jason Mendelson, Venture Deals, 5th edition, Wiley, 2022, chapter on valuation.
Pre-seed practitioner data: AngelList Quarterly, Carta's State of Private Markets reports (covers US deal economics; Iceland-relevant only at the multiplier level). For Iceland specifically: Kría grant announcements, Frumtak Crowberry portfolio updates, Brunnur Ventures press releases, and direct network knowledge.
Public-market subscription benchmarks (for the revenue-multiple sanity check): public-company comparables in the same category, 10-K filings, and EV/Revenue multiples from BVP's Cloud Index (cloudindex.bvp.com) for software companies.
Probability-weighted exit modelling: Mark Suster, "The Math Behind Why Most Founders Don't Get Rich", bothsidesofthetable.com — useful framing on outcome distributions for VCs and angels.
Aswath Damodaran (NYU Stern), "Valuing Young, Start-up and Growth Companies" — academic but practical treatment of the structural problem of valuing companies with most of their value in option terms.