Refolk
PlaybookMarket and talent intelligence

The Private Rival Runway Range: Last Raise to a Graded Cash-Out Window

Turn one named private rival into a dated cash-out range (earliest, central, latest month) with every input graded and a quarterly list of events that would move it.

21 min readLast reviewed November 16, 2026Read as Markdown

Key takeaways

  • Only primary cash counts toward a rival's runway, so every "raised $X" press release needs to be checked against the Form D or SH01 amount, and any secondary sales go in at zero.
  • A first Form D can understate a round, because "total amount sold" is cumulative and grows through amendments: a filing showing $12M sold of $20M offered can read $20M six weeks later.
  • In Kruze Consulting's analysis of more than $900M of startup spending, payroll-related costs were 76% of operating expenses, or 68% with cost of goods sold included. That ratio scales a headcount-based payroll up to total burn.
  • Profile counts can be off by 30-50 percent for small or low-adoption firms, so every function-by-country cell in the headcount series needs its own correction band.
  • In Refolk's index, one sales-intelligence software firm holds 30% of US Account Executive profiles at Computer Software companies. Burn has to be built from the rival's own function mix, not a market average.
  • Using the stale 774-day seed-to-Series A median instead of 616 days shifts the central estimate by about 5.2 months, so the timing prior needs refreshing every quarter.

This playbook takes one named, venture-backed private rival and turns its last raise and its headcount into a dated cash-out window, given as an earliest, central and latest month. It is written for strategy, research and talent-intelligence teams who have to tell leadership whether a rival's price cuts, hiring spree or sudden silence comes from strength or from desperation. Every input carries a source and a grade, and the guide ends with the events that would move the estimate at the next quarterly checkpoint.

What does an outside-in runway range tell leadership?

An outside-in runway range tells leadership the window in which a private rival must raise, cut or sell. It is built only from public records and stated as three months with a confidence grade. It turns ambiguous behaviour into something you can act on: the same price cut means one thing from a rival with years of cash and another from a rival with months.

Public writing on runway assumes you can see the bank balance. From outside you cannot, so the method puts bounds on each input and carries those bounds through to the end. Five terms come up throughout:

  • Runway: the number of months until cash reaches zero at the current net burn.
  • Gross burn: total monthly cash spending, payroll and everything else.
  • Net burn: gross burn minus the cash collected from customers each month.
  • Primary cash: money paid into the company itself for new shares, debt or grants.
  • Cash-out month: the calendar month in which the projected balance crosses zero.

The range is not a forecast of shutdown. Most rivals raise, cut or sell before the balance reaches zero. The window tells you when they have to do one of those things, and that is what makes their current behaviour readable.

Reading a rival's behaviour against its runway

Expanding spendCutting spend
Spending into a raise
Read aggressive pricing and hiring as a bid for the next round and watch for Form D amendments
Funded offence
Treat pricing and hiring as deliberate strategy and respond on product, not by waiting them out
Distress
Expect cuts, silence or a sale, and prepare to recruit their people and contest their accounts
Discipline
Read silence as efficiency, not weakness, and do not plan around their exit
Short runway leftLong runway left
The same spending signal means strength or desperation depending on how much runway sits behind it.

Read the matrix after you have the window, not before. Spending signals without a runway estimate are exactly the guess this playbook is meant to replace.

Which cash counts toward a rival's runway?

Only primary cash counts: money the company itself received for new shares, debt or grants. Secondary sales, where existing holders sell their shares to new investors, change who owns the company but put nothing into its account, however large the headline.

US: Form D

Form D is the notice a company files with the SEC when it sells securities in a private placement. Three things about it matter here:

  • It is due within 15 days of the first sale, so the first filing often captures only a first close.
  • "Total amount sold" is cumulative and is updated by amendment. A filing that shows $12M sold of $20M offered can show $20M sold six weeks later.
  • Under the SEC's instructions, Item 13 amounts "should include all cash and other consideration to be received for the securities, including cash to be paid in the future under mandatory capital commitments." A tranche that has been committed but not paid shows up in the number before it reaches the bank.

UK: SH01

SH01 is the Companies House return of allotment, required under Companies Act 2006 s.555. It is due within one month of the shares being allotted, and late filing carries penalties of £150 to £1,500. It records option exercises as well as funding rounds, so a small allotment between rounds is more likely an employee exercise than new capital. Date each ledger row by the allotment date, not the filing date.

Venture debt, grants and secondaries

I have not checked public sources for venture debt or grants for this guide. Look in secured-lending filings, the UK register of charges and grant award databases, but enter anything you find there as grade C until you have a document that shows the amount. Secondaries go in at zero.

What sits in the cash ledger

  1. Primary cash on a filing
    Form D amount sold or SH01 allotment, grade A
  2. Announced but not filed
    Press release amounts, grade B, may hide secondaries
  3. Venture debt and grants
    Not validated here, grade C until a document shows the amount
  4. Secondary sales
    Shares changing hands between holders, entered at zero
Each layer down is less certain, and the bottom layer adds nothing to runway at all.

How long does the last raise usually have to last?

The last raise has to last until the next one, and the published median interval between rounds gives you a starting assumption, or prior, for when that will be. Carta's guidance, drawn from 9,843 US rounds, is to expect a new round every 18-24 months. Its median seed-to-Series A interval was 1.9 years in Q4 2025.

This prior moves faster than most analysts update it:

PeriodSeed-to-A medianDays
Q4 2021~420 days420
Q4 2024774 days (conflict: 744)774
Q2 2025616 days616
Q4 20251.9 years~694 (derived)

Sources: Carta, and secondary summaries of Carta data. The Q4 2025 day count is derived (1.9 × 365). One secondary source gives the Q4 2024 figure as 774 days and another gives 744, so check Carta's own chart before you rely on it.

To set the prior:

  1. Take the rival's last close month, which is the earliest grade A date in your ledger for that round.
  2. Add the newest median interval for its stage. The result is the expected next-raise month.
  3. Adjust for sector. Consumer companies run much longer: the consumer seed-to-A median was 3.0 years in Q1 2025, against 1.7 years in Q1 2024.

Some limits apply. My sources do not establish stage-specific Series A to B or B to C medians, or the spread of quartiles around any median. Pull those from Carta's current chart before you use them on a later-stage rival. Sources also disagree on what to anchor to: Carta's 18-24 month norm or its full-year review figure of 2.1 years. My practice is to use the newest quarter for the central case and the wider norm for the band.

Treat the prior as a cross-check, not an input to the window. If your central cash-out month falls well before the expected next raise, either your burn is too high or the rival is raising early or arranging a bridge. If it falls well after, you may have missed some cash, or the rival is hiring below its plan.

How do you turn a headcount series into a burn rate?

Build burn from a monthly headcount series split by function and country, multiply each cell by a loaded cost, then scale payroll up to total spending. Multiplying a single headcount total by an average salary hides the two things that drive cost: where people sit and what they do.

Count, then correct

Profile counts are the raw material, and they can be wrong in either direction. One practitioner guide says counts for large, high-adoption firms are "usually within 10-20 percent of actual headcount", but for small or low-penetration firms "the count can be off by 30-50 percent." Former employees who never updated their profiles inflate the count, while under-covered countries and functions deflate it. Coverage ratios by function and by country for companies of 20 to 500 people are not established publicly. Give every cell its own correction band, and widen it for small teams and low-adoption markets. Refolk's guide to sizing private-company headcount covers the correction step in detail.

Use the rival's mix, not the market's

Refolk's index shows why a market average misleads:

Slice (Computer Software)ProfilesTop employer shareRatio to US SWE
US Software Engineer5810.3% (6/58)1.00
India Software Engineer557.3% (4/55)0.95
US Account Executive2330.4% (7/23)0.40

The profile counts and top employer shares come from Refolk's index, and the ratio column is derived. These are small slices from one narrow segment, so read them for mix and concentration, not market size. Across the slice, US engineers outnumber Account Executives by about 2.5 to 1, and there are almost as many Indian engineering profiles as US ones. But one sales-intelligence software firm holds 7 of the 23 AE profiles and shows only 2 engineers. Apply the slice-wide ratio to that firm and you would cost it as an engineering shop when it is really a sales floor.

30%
Share of US Account Executive profiles in one segment of Refolk's index held by a single firm
That firm shows 7 AEs and 2 engineers, the reverse of the 2.5 to 1 engineer-to-AE ratio across the slice.

Building the series by hand means pulling joiners and leavers by month, function and country from public profiles. Refolk returns that as a list from a plain-English request, for example the account executives and SDRs who joined the rival in the last six months, by country. That slice moves burn fastest when a rival is spending into a raise.

Load the cost

  • US staff: multiply base salary by 1.25-1.4x to cover taxes and benefits.
  • Non-US staff: no named firm publishes a non-US multiplier. Use your own local estimate and grade every non-US cell C.
  • Base salaries: use your own compensation benchmark for each function-and-location cell, and record where it came from.

Scale payroll to total spend

Kruze Consulting looked at more than $900M of startup spending and found payroll-related costs were 76% of operating expenses (76.4% exactly), or 68% with cost of goods sold included. Consultants alone were 12.5% of opex. Contractors rarely show up in profile counts, so a payroll built from headcount misses about 16% of payroll-related spend (12.5 ÷ 76.4, derived).

76%
Payroll-related share of operating expenses at venture-backed startups
From Kruze Consulting's analysis of more than $900M of spending; the share falls to 68% once cost of goods sold is included.

Because you will subtract full revenue later, measure burn on the 68% basis so that cost of goods sold is included:

  • Low gross burn: assume your count already includes contractors, and divide loaded employee payroll by 0.68.
  • High gross burn: assume it includes none. Divide by 0.836 to add contractors back, then by 0.68.
  • Central: the midpoint of the two.

How large is the revenue offset?

The revenue offset is a band, not a single number: the monthly equivalent of the ARR range for the rival's stage, chosen by the year it raised rather than today. ARR, or annual recurring revenue, is the yearly value of contracted subscription revenue. Divide it by 12 to get the monthly offset.

Stage and yearMedian ARRBand
Series A 2021$1.3Mn/a
Series A 2024$2.3Mn/a
Series A 2025$2.8M (Carta/SVB); $3.3M (secondary)$1M-$6.5M
Series B 2025$7.1M (secondary)unverified

Sources: Carta, citing SVB data. The $3.3M median, the $1M-$6.5M middle 50% and the $7.1M Series B figure come from a secondary summary of SVB data. Check them against the SVB report itself.

How to set the band:

  • Match the vintage. Median Series A ARR rose from $1.3M in 2021 to $2.8M in 2025, about 2.15x (derived). A rival that raised its Series A in 2021 gets a lower offset than one that raised in 2025.
  • Pair the ends correctly. Use the low end of the band in the earliest cash-out case and the high end in the latest.
  • Grade it. Grade the offset B if you used the Carta/SVB median, and C if you used the unverified band or the Series B figure.

There is a limit here. These figures are revenue at the moment of raising, and my sources do not establish how fast a rival grows afterwards. If you have direct evidence of revenue since the raise, use it and say so. Otherwise, widen the band rather than invent a growth rate.

The runway range procedure, step by step

Eight steps take about 11.5 to 16.5 hours the first time and about 2 hours per quarter after that. These times are my own estimates and have not been validated against other practitioners.

From last raise to cash-out window

  1. Ledger
    Primary cash by date, every row graded
  2. Prior
    Expected next-raise month from the stage median
  3. Headcount
    Monthly function-by-country series with correction bands
  4. Burn
    Loaded cost scaled to total spend, minus the revenue band
  5. Window
    Earliest, central and latest cash-out month
  6. Watchlist
    Events that would move the window next quarter
Cash and burn are each built as bands, and the window is where the bands meet zero.

Runway range procedure

  1. Fix the legal entity
    Identify the operating company, any UK or other foreign subsidiaries, and the officers listed on Form D. Done means every entity that issues shares or employs staff is on the list (about 1 hour).
  2. Reconstruct cash in
    Pull every Form D and amendment, every SH01 and every press release, then take out secondaries. Done means a dated primary-cash ledger with each row graded A, B or C (2 to 4 hours).
  3. Set the timing prior
    Add the newest stage median interval to the last close month and adjust for sector. Done means an expected next-raise month (30 minutes).
  4. Build the headcount series
    Count staff each month by function and country for 12 to 24 months, corrected for coverage. Done means a table with a correction band in every cell (3 to 6 hours).
  5. Apply loaded cost
    Multiply US base salaries by 1.25-1.4x, grade non-US cells C, gross up for contractors, and scale to total spend. Done means a low, central and high gross burn (2 hours).
  6. Subtract the revenue offset
    Take the ARR band for the rival's stage and the year it raised, divide by 12, and subtract it from gross burn. Done means a net burn band (1 hour).
  7. Compute the window
    Run the balance forward month by month in the low, central and high cases until it crosses zero, then check the result against the prior (1 hour).
  8. Grade the inputs and set the watchlist
    Set the overall grade to the weakest of the cash, burn and revenue grades, list the tells to watch, and repeat each quarter (1 hour, then about 2 hours per quarter).

Computing the window

Remaining cash in any month is ledger primary cash, plus carryover from the previous round, minus cumulative net burn since the close. You cannot observe carryover. Set it to zero in the earliest case. In the latest case, use whatever your burn model leaves from the previous round over the previous interval, with zero as the floor. Then run the balance forward month by month using the headcount series. Beyond your last month of data, hold headcount flat for the central and latest cases, and extend the trailing growth for the earliest case.

  • Earliest: low cash, high gross burn, low revenue offset.
  • Central: central values throughout.
  • Latest: high cash, low gross burn, high revenue offset.
Runway range worksheet
RIVAL: [legal entity] | Subsidiaries: [list] | Checkpoint: [month]

CASH LEDGER (primary only)
Date | Instrument | Amount | Source | Grade (A filing / B release / C inferred)
[allotment or sale date] | [equity / debt / grant] | [amount] | [filing reference] | [A/B/C]
Secondaries excluded: [amount and source, or none found]
Committed but unpaid tranches: [amount, or none]
Carryover from prior round: earliest 0 | latest [model output, floor 0]

TIMING PRIOR
Last close: [month] | Stage median: [days, which quarter] | Sector adjustment: [yes/no]
Expected next raise: [month]

GROSS BURN (per month)
Loaded employee payroll = sum over cells of (count x base x multiplier)
US multiplier: 1.25 / 1.325 / 1.4 (low / central / high)
Non-US multiplier: [local estimate], grade C
Low = payroll / 0.68
High = payroll / 0.836 / 0.68

REVENUE OFFSET (per month)
Stage and raise year: [stage, year]
Low = [band low ARR] / 12 | Central = [median ARR] / 12 | High = [band high ARR] / 12

WINDOW
Earliest (low cash, high burn, low offset): [month]
Central: [month]
Latest (high cash, low burn, high offset): [month]
Overall grade: [weakest of cash, burn, revenue grades]

Use one worksheet per rival per checkpoint. Replace the bracketed fields and keep the grade column even when the answer is C.

How outside-in runway estimates go wrong

Most bad runway estimates fail on cash rather than burn: they count money the rival does not have, or miss money it does. Check the seven failure modes below before the estimate leaves your desk.

  1. Reading Form D as the round size. A first-close filing, such as $12M sold of $20M offered, understates the cash raised. How it lies: the runway ends implausibly soon after a well-publicised raise. Check: search for amendments, because amount sold is cumulative.
  2. Counting secondaries as runway. A "raised $X" press release can include secondary sales that add no company cash. How it lies: the announced figure is higher than the filed figure. Check: reconcile every release against the Form D or SH01 amount and count only the filed primary.
  3. Using a stale timing prior. Using the 774-day Q4 2024 figure instead of 616 days shifts the central estimate by about 5.2 months (derived). Check: use the newest Carta quarter.
  4. Ignoring contractors. Consultants are 12.5% of opex in Kruze's data and are often missing from profile counts, so a burn built from headcount misses about 16% of payroll-related spend. How it lies: a rival that relies heavily on contractors looks cheap and long-lived. Check: the high case must include the contractor gross-up.
  5. Mixing up which way profile counts are wrong. Former employees inflate counts, while low-adoption regions deflate them. How it lies: a rival that cut staff quietly still shows its old headcount, so burn looks high and the window looks early. Check: look for profiles that still list the employer after the person has visibly moved on, and cross-check against the leaver series.
  6. Treating no WARN filing as safety. Firms under 100 employees, which is most of the 20-500 band, never trigger federal WARN. Check: apply state rules, such as New York at 50 or more employees and California at 75 or more.
  7. Applying the US multiplier everywhere. The 1.25-1.4x figure is a US number, and a non-US multiplier is unverified. How it lies: a rival with a large offshore engineering team gets mispriced in whichever direction your guess leans. Check: grade every non-US cell C and let that grade cap the overall grade when those cells dominate payroll.
A runway estimate without a confidence grade is a guess with a decimal point.

Which events move the estimate at the next checkpoint?

Two kinds of event move the estimate: new filings that change cash, and observable spending changes that change burn. Only one has a statutory lead time, the federal WARN notice, which must reach the required parties at least 60 days before a layoff or closing at employers with 100 or more staff. Treat every other tell as a sign of direction, not timing.

TellWhat it provesHow it liesKnown lead time
WARN noticeA layoff or closing is plannedMissing for most sub-100 rivals60 days federal; 90 in NY (50+) and Nebraska
Form D amendmentCumulative primary cash roseIncludes unpaid committed tranchesWithin 15 days of first sale
SH01 allotmentNew UK shares were issuedSmall ones may be option exercisesWithin one month of allotment
Hiring freeze or net leaversBurn is being cutSeasonal pauses, evergreen postingsNot established
Office move or lease changeFixed cost is changingA move can mean growthNot established
Event sponsorship dropDiscretionary spend was cutBudget shifted to other channelsNot established

The public record does not establish lead times from lease changes, posting freezes, sponsorship withdrawals or price changes to a down round or shutdown. If you want to use those tells for timing, build your own record of how long past rivals took to go from the tell to the event, and state its sample size.

For context, Carta data reported by its Head of Insights, Peter Walker, showed 966 shutdowns among Carta customers in 2024, against 769 in 2023, a 25.6% increase. Enterprise SaaS made up 32% of them. Shutdowns are also concentrated early: 74% of shutdowns since 2023 were pre-seed or seed companies, which are the rivals least likely to trigger WARN. Layoffs.fyi, run by Roger Lee, is another place to check for announced cuts. Remember that all of this covers Carta customers and announced events, not every startup.

At each quarterly checkpoint, pull the filings again, update the headcount series, and recompute the window. Re-running the same joiner and leaver requests in Refolk each quarter gives you a series that is comparable from one checkpoint to the next, which matters more than any single month's count. Then send leadership a brief they can read in a minute.

Leadership brief: rival runway range
Subject: [Rival] runway range as of [checkpoint month]

Bottom line: If it neither raises nor cuts, [Rival] runs out of cash between [earliest month] and [latest month], with [central month] most likely. Confidence: [A/B/C].

What this means for its behaviour: [matrix quadrant] - [one sentence on what its pricing, hiring or silence now signals].

Inputs
- Cash: [amount] primary, from [filings], grade [ ]
- Gross burn: [low]-[high] per month, from [n] people across [functions and countries], grade [ ]
- Revenue offset: [low]-[high] per month, from the [stage, raise year] benchmark, grade [ ]
- Expected next raise (timing prior): [month]

What would move this at the next checkpoint
- Form D amendment or SH01 allotment: window moves later
- WARN notice or net leavers in sales: rival is cutting, window moves later, read as distress
- Hiring acceleration in [function, country]: window moves earlier

Next checkpoint: [month]

Keep it to one page. The confidence line is not optional, and every input names its source.

Before the estimate goes to leadership

  • Every entity that issues shares or employs staff is listed.
  • Every cash ledger row has a date, a source and a grade of A, B or C.
  • Every press-release amount has been checked against the Form D or SH01 amount, and secondaries are recorded at zero.
  • Form D amendments have been searched for after the first filing.
  • The timing prior uses the newest available quarter and the median for the rival's own stage and sector.
  • The headcount series covers 12 to 24 months with a correction band in every cell.
  • Every non-US cell is graded C.
  • The high gross-burn case includes the contractor gross-up.
  • The revenue band matches the stage and year of the rival's last raise.
  • The brief states the overall grade and at least three events that would move the window.

Questions practitioners ask

How accurate is an outside-in runway estimate for a private competitor?

It is accurate enough to place a rival inside a window of months, not to name the exact month. The biggest errors come from profile counts, which can be off by 30-50 percent for small or low-adoption firms, and from unverified inputs such as non-US loaded-cost multipliers and revenue bands. That is why the output is an earliest, central and latest month, with an overall grade equal to the weakest of the cash, burn and revenue inputs.

Does a competitor's funding announcement tell me how much cash it raised?

Not reliably. A "raised $X" headline can include secondary sales between existing holders, which put no cash into the company. Check every announcement against the SEC Form D amount sold, which is cumulative and updated by amendment, or against the UK SH01 return of allotment. Count only the filed primary amount, grade figures that appear only in a press release B, and remember that a first Form D may capture just the first close.

How do I convert competitor headcount to a burn rate?

Take a monthly headcount series split by function and country, multiply each cell by base salary and a loaded-cost multiplier, then scale payroll up to total spend. For US staff the multiplier is 1.25-1.4x base. In Kruze Consulting's data, payroll-related costs are about 76% of opex, or 68% with cost of goods sold included. Add a contractor gross-up too: consultants are 12.5% of opex and rarely show up in profile counts.

When will a startup competitor need to raise again?

Start from the published median interval for its stage. Carta's guidance is to expect a new round every 18-24 months, and its median seed-to-Series A gap was 1.9 years in Q4 2025. Add that to the rival's last close to get an expected next-raise month, adjust for sector, and compare the result with your cash-out window. Refresh the median every quarter, because it moved by about 158 days in two quarters.

If a rival has not filed a WARN notice, is it safe from layoffs?

No. Federal WARN requires 60 days' notice only from employers with 100 or more employees, so rivals below that line, which is most of the 20-500 band, give no statutory signal. New York's rule applies at 50 or more employees with 90 days' notice, and California's at 75 or more. No notice can mean the rule did not apply or that no layoff was planned, and it does not tell you which.

Try it on the search you came here for

Stop building boolean strings. Just describe the person.

Type one sentence. I plan the search, read GitHub, public LinkedIn and Crunchbase records, and the open web as it is right now, and hand back a ranked list with the reason next to every name.

  1. 01Describe them

    One plain sentence. Role, city, stack, stage, whatever matters to you.

  2. 02I read the web live

    GitHub, public LinkedIn and Crunchbase records, the open web. Not a database that went stale last quarter.

  3. 03You read the shortlist

    Ranked, with the reasoning under every name. Open a profile, ask a follow-up, narrow it down.

  • No boolean, no filters, no seat to buy. One box.
  • Read at search time, so a profile updated yesterday counts today.
  • Every step visible as it runs, every name with its reason.

500 free credits on sign-up. No card, no demo call. See real searches.

Read next