The median seed pre-money valuation reached $16 million in Q3 2025, up 14% year over year (Carta State of Private Markets, Q3 2025). The median post-money rose to a record $24 million in Q4 2025 on roughly $4 million raised. Founders read those numbers and try to back into a target. That is the wrong way to read them. Seed valuation is the output of a process, not the input. Investors work backward from fund math and forward from how much proof has been assembled. The number you defend depends on how well you understand both ends of that calculation.
This guide walks through the actual methods you will encounter at a seed round in 2026. It explains why founders who arrive with structured proof close at the top of their range, and why founders who arrive with a deck and a wishlist almost always end up below the median.
What investors are actually solving for
A seed valuation is two equations meeting in the middle.
On the investor's side, the math starts at the fund level. A typical seed fund needs to own 10% to 15% of any company at entry to make the math work, because most of the portfolio will not return the fund and the winners need to carry the rest. So the investor calculates the check size that buys 10% to 15% at a given post-money. At a $20 million post-money seed, a 10% target means a $2 million check. The valuation is downstream of that ownership target, not upstream of a discounted cash flow model.
On the founder's side, the math starts at runway. A typical seed round funds 18 to 24 months of operations. The Carta median seed cash raised was $4 million in 2025. Round size determines how much equity the founder has to sell to make the runway equation work. The Carta 2025 Founder Ownership Report shows median seed dilution sitting at roughly 20%, down from 23% in 2019. After a seed round, the median founding team owns 56.2% of their company. After Series A, it drops to 36.1% (Carta Founder Ownership Report 2025, carta.com/data/founder-ownership).
Where the two equations meet is the valuation. There is some negotiation flex inside the bracket, but the bracket itself is determined by fund-ownership targets and founder-runway targets. Most "valuation conversations" are actually conversations about whether the founder has enough proof to move the bracket up.
This is the part most first-time founders miss. They prepare to argue valuation. The investor is not arguing valuation in the abstract. The investor is asking: does the proof on the table justify pricing this round at the top of the bracket, the middle, or the bottom?
Why paper does not move the number
Founders walk into seed conversations with three things. A deck. A model. A list of comparable rounds.
None of those move the bracket on their own. Here is why.
A deck is curated by definition. The Fu and Taylor study at NBER analyzed 21,000 venture deals using cell phone behavioral data and found that 95% of deals show zero detected in-person diligence visits, and average diligence drops 33% in normal markets and 84% in hot markets (Fu and Taylor, NBER Working Paper 33987, July 2025). Investors know decks are designed to persuade. They discount accordingly.
A model is assumptions on top of assumptions. Only 30% of VCs use quantitative financial models in their decision process, according to Gompers, Gornall, Kaplan, and Strebulaev's survey of 885 institutional VCs in the Journal of Financial Economics (Gompers et al., JFE 2020). The model is a sanity check, not an input.
A list of comparable rounds is reference data, not proof. Investors have access to the same data through Carta, PitchBook, and AngelList. Pointing to a $20 million post-money round at a competitor does not move your number unless you can also point to comparable proof points that justify the comp.
What moves the bracket up are demonstrated facts. Real revenue. Real retention. Real team specifics. Real evidence the market will pull. Without those, the investor's default is the middle of the bracket, because the middle is the safest place for the math to land.
The 2025 Equidam Startup Valuation Delta report makes the same point from a different angle. After a slight recovery from 2023 lows, pre-seed valuations remain below 2024 highs. The reason is not market sentiment. It is that investors are now expecting demonstrable unit economics and clear go-to-market plans before they price a round, rather than treating the projection slide as a reasonable proxy (Equidam Startup Valuation Delta H1 2025). Demonstrable, not described.
The four valuation methods you will encounter
Most seed-stage VCs blend two or three of the following. Knowing which one is being applied changes how you negotiate.
The Berkus Method
Originally developed by angel investor Dave Berkus, this method assigns up to $500,000 in value to each of five qualitative factors: a sound idea (technology risk), a working prototype (technology risk), a quality management team (execution risk), strategic relationships (market risk), and product rollout or sales (production risk). Maximum theoretical pre-money value: $2.5 million. Modern angels often double or quadruple these caps to fit current market conditions, but the structure is the same: five qualitative factors, dollar values per factor, sum to a pre-money number.
When you see it: angel rounds, pre-revenue pre-seed, very early seed.
The Scorecard Method
The Scorecard Method starts with the average pre-money for similar companies in your region and stage, then weights factors against that average. Typical weights: management team 30%, market opportunity 25%, product 15%, competition 10%, sales channels 10%, capital needs 5%, other 5%. Each factor gets a multiplier above or below 1.0 based on how the founder compares to peers. The result is a pre-money number anchored to comps.
When you see it: angel groups, traditional pre-seed and seed firms.
The Risk-Factor Summation Method
This method also starts with a comparable-anchored baseline, then adjusts up or down across 12 risk dimensions: management, stage of business, legislation, manufacturing, sales and marketing, funding, competition, technology, litigation, international, reputation, exit. Each factor gets graded from very high risk to very low risk and the cumulative grade adjusts the baseline.
When you see it: angel networks, micro-VCs that lean structured.
Comparable Company Analysis
This is the dominant approach at seed and the one most institutional VCs default to. The investor looks at recent rounds in your sector, stage, and geography, then anchors the bracket to the Carta benchmark. AI seed startups currently price around 42% higher than non-AI peers, with median AI seed pre-money near $17.9 million across multiple 2026 trackers (Eqvista AI Startup Fundraising Trends 2026; TechCrunch March 2026 analysis of AI seed valuations). Inside the comp set, your specific proof points push you up or down within the bracket.
When you see it: every seed round above $1 million.
Comparison table
Method | When investors use it | Typical range | Strength | Weakness |
|---|---|---|---|---|
Berkus | Pre-revenue, angel-led pre-seed | $0 to $2.5M (modern: $0 to $5M) | Simple, explicit, no model required | Caps very low for hot sectors |
Scorecard | Angel groups, traditional seed | Anchored to regional avg | Comparable-grounded, weighted | Sensitive to peer set selection |
Risk-Factor Summation | Angel networks, micro-VCs | Anchored to comp baseline | 12-dimension diagnostic | Subjective grading per factor |
Comparable Company Analysis | Institutional seed, $1M+ rounds | $10M to $30M post-money in 2025 | Reflects actual market clearing prices | Comps can be cherry-picked either way |
The practical takeaway is that most seed founders today face Comparable Company Analysis with a Scorecard Method overlay. The investor anchors to the Carta median, then runs you against a peer set in your sector, then adjusts within the bracket based on the proof you can show.
What actually moves you up the bracket
Three things shift a seed round from the middle of the bracket to the top.
1. Demonstrated traction, not described traction
Investors run a baseline expectation. The 2025 Forum Ventures and Waveup studies found that round expectations have jumped one full stage compared to 2021. Pre-seed founders are now expected to show what seed founders used to show, and seed founders are expected to show what Series A founders used to show (Forum Ventures State of the VC Market, 2024 to 2025; Waveup Survey of 56 VCs, 2025). The line on what counts as traction has moved.
In 2026, most seed-stage VCs expect a working product, $10K or more in monthly recurring revenue or signed pilot agreements, a clear ideal customer profile, and a credible 12 to 18 month plan to Series A milestones. Hitting that bar lands you in the middle of the bracket. Beating it, with hard numbers an investor can confirm in your live data, moves you up.
The verification matters. A founder who claims $30K MRR is making an assertion. A founder who shows $30K MRR live in Stripe is presenting evidence. Investors weight evidence dramatically higher than assertion.
2. Founder credibility that compounds
Repeat founder data shows the asymmetry. According to Crunchbase analysis using Equidam survival rate data and First Round's 10-Year Project, a previously successful founder has a 30% success rate, a previously failed founder 20%, and a first-time founder 18%. Repeat founders also negotiate better terms: more retained equity, more board control, less dilution per round. Serial founders with prior exits can close $10 million seed rounds in a week.
If you are not a repeat founder, the closest substitute is structured credibility: real domain experience, named customers, named advisors, a team page that shows specific track records. Vague founder bios cost valuation. Specific ones add it.
3. Convicted demand from multiple investors
The Snowballing Signaling Theory review in the journal Cogent Business and Management synthesized 344 news articles and 970 academic publications and found that valuation signals accumulate over time, reinforcing investor expectations and driving cyclical dynamics (Cogent Business and Management, 2025). One investor's pricing decision shapes the next investor's pricing decision. This is why running a process matters.
A founder with three term sheets at the same valuation has more pricing power than a founder with one term sheet at that valuation, even if the underlying business is identical. This is also why the Aran and Packin "Due Diligence Dilemma" framework calls a lot of late-stage seed diligence "proxy diligence": investors signal off other investors rather than running independent verification (Aran and Packin, University of Illinois Law Review, forthcoming 2025).
What this means in practice: convicted demand at a target number is itself a valuation lever. Get to multiple yeses at the same price, and the price holds.
The repetition trap that kills seed valuations
Here is the structural problem most first-time founders run into.
Each fund builds its own conviction. No fund reads another fund's notes. The investor across the table runs the Comparable Company Analysis from scratch, asks the same first-call questions, scores the team on the same axes. Then the next investor does it again. Then the next.
That repetition has compounding cost on the founder side and erosion cost on the valuation side. Carta data shows the median seed close now takes 142 days, more than double the 69 days it took in 2021. The median time between seed and Series A is 616 days as of Q2 2025. That is over 20 months of fundraising drag on top of building the business (Carta State of Private Markets, Q2 and Q3 2025).
The valuation cost is subtler. Long processes leak signal. By round day 90, the founder has typically lowered the asked valuation once. By day 120, the founder has lost negotiating leverage. Term sheets that should have priced at the top of the bracket end up in the middle because the founder has been on the back foot for too long.
Valuation is a function of speed and parallelism. A short process with multiple convicted investors at the same target prices high. A long process with sequential investors, each running their own evaluation from zero, prices low.
The proof layer
This is where SeedForge changes the valuation conversation. A 30-minute AI session walks the founder through the same proof points that drive a Comparable Company Analysis: team specifics, traction reality, market evidence, unit economics, the actual numbers behind the narrative. Live data connects from Stripe, GitHub, and LinkedIn so the metrics are real, not slide-edited. The output is a Living Profile, one shareable link, that an investor explores before the first call.
Now the negotiation is about the number. Not about what is true. Investor #1 sees the proof, then investor #2 sees the same proof, then investor #3. The founder runs the process in parallel rather than rebuilding conviction with each fund from zero. Visit seedforge.com to run the session.
This is the wedge between a seed round priced at the top of the bracket and one priced in the middle. Founders who arrive with structured proof give the investor less room to discount. The bracket itself does not move, because the bracket is set by fund math. But the position inside the bracket moves significantly.
Practical: a seed valuation defense checklist
Before you walk into your first seed conversation, run through this list.
Know your fund-math comp set. Identify the funds you are pitching. Look up their typical seed check size and ownership target. Calculate the post-money each fund implies if they hit their target ownership at their typical check size. That is your bracket per fund.
Know your dilution math. Cash needed for 18 to 24 months of runway, divided by the post-money you can defend, equals the percentage you sell. If that number is above 25%, you are either raising too much or pricing too low.
Have proof connected, not promised. Stripe, GitHub, LinkedIn, deck, model, references. Connected and ready to share via one link, not "available on request."
Build a peer comp slide. Not "Stripe and Plaid raised at $30M." A real, honest peer set: companies in your stage, sector, and geography that priced recently. Carta Launch and AngelList show real recent rounds.
Run multiple funds in parallel. Sequential processes lose. Parallel processes hold pricing. Aim for 8 to 12 fund conversations starting within the same two-week window.
Know the AI premium math. If you are an AI company, the bracket is roughly 42% above the non-AI peer median. If you are not an AI company, do not try to claim the AI premium. Investors test this immediately.
Track your touchpoints. Date of first meeting, date of partner meeting, date of term sheet ask, by fund. If a fund has not progressed in three weeks, deprioritize them. Process velocity protects valuation.
What a typical 2026 seed cap table looks like
A clean way to picture how all of this lands together: imagine a founding team raising a $4 million seed at a $16 million pre-money valuation, which produces a $20 million post-money. The investor takes 20% of the company. Combined with a typical 10% to 15% option pool reserved for hires, the founders end the round owning roughly 56% of the cap table on Carta's 2025 median benchmark (Carta Founder Ownership Report 2025). Move the round to $5 million on $20 million pre-money and the post-money becomes $25 million, dilution lands near 20%, and the founders give up an additional 4 to 5 percentage points compared to the prior structure. Move the round to $3 million on $12 million pre-money, and the dilution math holds at the same 20% range while the founders raise less runway. The ratio of cash to dilution is roughly fixed by the bracket. Founders who try to optimize one without modeling the other usually end up regretting the trade twelve months later, when Series A investors look at the cap table and quietly walk away.
FAQ
What is the median seed valuation in 2026?
The most recent data point is Q4 2025: median post-money seed valuation of $24 million on Carta, with median pre-money near $16 million in Q3 2025 and median cash raised of $4 million. AI companies clear roughly 42% above non-AI peers, with AI seed median pre-money around $17.9 million (Carta State of Private Markets, Q3 to Q4 2025; TechCrunch AI seed valuations analysis 2026).
What is the difference between pre-money and post-money valuation?
Pre-money is the value of the company before new investment. Post-money equals pre-money plus the new round size. Investor ownership equals new round divided by post-money. At a $16 million pre-money raising $4 million, the post-money is $20 million and the investor owns 20%. Founders should always confirm which valuation a term sheet refers to before signing.
Is the AI premium real?
Yes. Recent 2026 seed data shows AI startups pricing roughly 42% higher than non-AI peers at seed, with revenue multiples between 10x and 50x and median typically 20x to 30x. The premium reflects investor belief in faster scaling, not different fundamentals. Non-AI founders cannot claim the premium without AI-grade traction, and investors test the claim immediately (TechCrunch 2026 AI seed valuations analysis).
How much equity should I expect to give up at seed?
Median seed dilution sits near 20% in 2025, with the most common range 20% to 24% (28% of rounds in that bracket). After a typical seed, the median founding team retains about 56% of the company. After Series A, that drops to 36% (Carta Founder Ownership Report 2025, carta.com/data/founder-ownership). Selling more than 25% at seed is a yellow flag for later-stage investors.
What if I have no revenue yet?
Pre-revenue valuations rely on the Berkus, Scorecard, or Risk-Factor Summation methods more heavily, and the bracket is lower: pre-seed median pre-money was $7.7 million as of Q3 2025 (PitchBook-NVCA Venture Monitor). The lever is not financial projection. It is structured proof of team quality, market evidence, and prototype completeness. Demos and pilot LOIs price higher than slides about future revenue.
Can investors actually calculate a startup's valuation?
Not in the precise sense. Only 30% of VCs use quantitative financial models in their decision process (Gompers et al., JFE 2020). What investors do is run a comparable-anchored bracket from fund math, then position you within the bracket based on proof. Your valuation is less the output of a calculation and more the output of conviction at a particular comp price.
How long should I expect a seed round to take?
The Carta median seed close in 2025 was 142 days from first conversation to wire (Carta Q2 2025 data), more than double the 69 days it took in 2021. Top quartile founders close in 30 to 45 days. The differentiator is process velocity: parallel conversations, structured proof shared upfront, and a clear competitive process.
Bottom line
Seed valuation is fund math meeting founder math at a comparable-anchored bracket. The bracket is set by Carta's market clearing data. Your position in the bracket depends on demonstrated proof, founder credibility, and convicted demand from multiple investors at the same price.
Founders who walk in expecting to negotiate a number lose. Founders who walk in with proof connected, comp math done, and parallel investor conversations running, hold pricing.
The shift from a seed priced at the bottom of the bracket to the top depends on how much risk the founder has converted into evidence before the conversation starts. Pitch quality plays a small role in that. Connected proof, structured before the first call, plays a much larger one.