Portfolio Monitoring Starts With Reports That Arrive Late

Portfolio monitoring is how a venture fund collects, checks and acts on its companies’ numbers, and how its limited partners (LPs) check the fund. Among listed companies bad news tends to arrive late, so a fund should treat a late package as a reason to ask and count missing reports before averaging metrics.

Whether the same holds in private portfolios, where the founder closest to running out of cash might also be the one who has stopped returning calls, has not been measured. What can be measured is the delay built into the reporting chain. Under the industry’s model contract a company has 45 days to send quarterly statements to its investors, and CalPERS, California’s public employees’ pension fund, gives its managers 120 days to report to it.

Exhibit 1. Portfolio monitoring is a chain of reporting clocks

Link	What is owed	When	Rule or standard
Company to its venture investors	Monthly income statement and balance sheet (optional clause)	30 days after month end	NVCA model Investors’ Rights Agreement (2025)
	Quarterly unaudited statements and share count	45 days after quarter end	NVCA model (2025)
	Annual statements, audit optional	180 days after year end	NVCA model (2025)
Fund to its own books	Fair value of each holding	Each time NAV is reported to LPs	IPEV Valuation Guidelines (2022)
GP to LPs	Valuation information for each portfolio company; annual report	Quarterly; annual within 90 days	ILPA Principles 3.0 (2019); ILPA templates from Q1 2026
	Quarterly statement of fees, expenses and performance	45 days, or 75 for a fund of funds	SEC rule (2023), vacated on 5 June 2024
LP’s published reporting	Fund-by-fund performance	GPs have 120 days; about two quarters’ delay	CalPERS (2026)

Source: NVCA (2025); IPEV (2022); ILPA (2019, 2025); SEC (2023); US Court of Appeals for the Fifth Circuit (2024); CalPERS (2026). CX Cash synthesis.

Portfolio monitoring runs on a chain of clocks, and each link adds weeks

The first clock is set in the financing documents. The National Venture Capital Association revised its model Investors’ Rights Agreement in October 2025. The model asks the company to send major investors quarterly statements “as soon as practicable, but in any event within 45 days after the end of each quarter”, and annual statements within 180 days of year end. A bracketed, optional clause adds monthly statements, “within 30 days after the end of each month, an unaudited income statement” among them. The 2025 revision lets a major investor ask for the numbers to be uploaded “to a portfolio management platform” (NVCA, 2025).

The second clock runs from the fund to its LPs, and no statute sets it. ILPA’s Principles 3.0 say that “Valuation information related to the portfolio companies should be disclosed on a quarterly basis”, with annual reports within 90 days of year end (ILPA, 2019). ILPA’s updated Reporting Template, its first revision since 2016, and a new Performance Template are meant to be in use from Q1 2026 (ILPA, 2025).

The SEC tried to fix this clock by rule. Its 2023 private fund adviser rule required a quarterly statement within 45 days of each of the first three quarter ends, and 75 days for a fund of funds (SEC, 2023). The Fifth Circuit vacated the whole rule on 5 June 2024 (Fifth Circuit, 2024), so the deadline is whatever each limited partnership agreement says.

The third clock belongs to the LP. CalPERS publishes its private equity table with a warning: “The General Partners have 120 days to provide Limited Partners with financial data, so there is generally a 2-quarter delay in performance reporting” (CalPERS, 2026). Read on 2 October 2026, the table showed data as of 31 March 2026, 185 days old. Its row for Insight Venture Partners X, a 2018 fund, shows $261.8M paid in, $308.0M returned and $718.7M returned plus remaining value, a 2.7x multiple and a 20.3% net IRR. About 57% of that value was still unrealised (our arithmetic), and CalPERS adds that “Interim IRRs by themselves are not the best indicators of current or future fund performance.”

The delay is paid for in runway. A 120-day window is about 3.9 months, so a company that ends a quarter with 12 months of cash has about 8 left on the day its fund’s LPs are due to see the number (our arithmetic). For an LP, portfolio monitoring reads history. The decisions that change outcomes sit with the GP, and the GP’s own information has gaps.

In theory and in listed companies, late reports lean towards bad news

The gaps may have a shape, and economic theory predicts one. Paul Milgrom’s review of persuasion games covers sellers who hold verifiable information and buyers who decide what to believe. It starts from skepticism: “When evidence is missing, they view that lack of evidence suspiciously, considering that any missing evidence is likely to be unfavorable to the seller” (Milgrom, 2008). If the buyer knows the seller has the facts, disclosure unravels. The best sellers report first to set themselves apart, then the next best, until silence identifies the worst.

Milgrom’s second result is the one closest to a startup’s position: “Uncertainty about the seller’s information mutes skepticism and damages welfare”. A buyer who cannot tell whether a verifiable report exists “allows both that the seller may not know the value and that the seller may be concealing the value because it is low”. In equilibrium, “the seller always withholds very bad news and reports only relatively good news”. Read onto a portfolio, the founder is the seller and the investor the buyer, and the uncertainty that softens skepticism has a familiar form: the books are not closed yet.

A reporting date removes that particular excuse. Once the contract says the package is due by day 30, silence on day 31 can no longer be put down to an unfinished close, although it can still have an innocent cause, such as a new bookkeeper. Milgrom’s own summary of the dynamics is that “good information is readily reported while bad information dribbles out”. The model describes how a skeptical buyer should read silence; it does not say how often a late report hides bad news in practice.

The evidence from listed companies points the same way. Dan Givoly and Dan Palmon found in 1982 that most companies report well ahead of the filing deadline, that “bad news tends to be delayed”, and that prices react more to early announcements than to late ones (Givoly & Palmon, 1982). S. P. Kothari, Susan Shu and Peter Wysocki concluded in 2009 that managers, on average, delay the release of bad news to investors, an inference drawn from the market’s sharper reaction to bad news than to good (Kothari, Shu & Wysocki, 2009). Neither study covers private companies, and we know of no dataset that does. Listed companies face filing deadlines, analysts and a share price that reacts on the day, and a founder reporting to a handful of investors faces none of those.

So the effect in private portfolios could be larger or smaller. It is a hypothesis worth acting on with a phone call, short of a verdict.

Wirecard, a German payments company then in the DAX index, gives a dated record of the pattern at its most extreme. On 28 April 2020 it announced that its 2019 annual report would miss the 30 April deadline. In the chronology of the European Securities and Markets Authority, “The CEO states that this is due to the Coronavirus pandemic”.

On 18 June it postponed the statements again, citing “indications of presentation of spurious balances”, and four days later it said that €1.9bn of balances on trust accounts probably did not exist (ESMA, 2020). It filed for insolvency on 25 June (Wikipedia, Wirecard). Germany’s accounts enforcement panel already listed “Delayed preparation/publication of the annual financial report” among the risk factors it used to pick companies for examination. Wirecard was a listed company, a fraud and an extreme case, and most late reports have duller explanations. The panel’s checklist treated a late annual report as a reason to look closer, which is the most the evidence supports.

Therefore an average of the reporters can flatter the portfolio

Statisticians have a name for data whose absence depends on the missing value itself: missing not at random, or nonignorable nonresponse, in which “the value of the variable that’s missing is related to the reason it’s missing” (Wikipedia, Missing data). If silence is more likely when the numbers are bad, every portfolio summary computed from the companies that reported is biased upward, and better software cannot repair a bias in rows that are missing.

The repair starts with a column: the number of days since each company’s package fell due and did not arrive. It is zero for every company that is on time. It belongs first in the portfolio review, ahead of cash and burn, because it says which of the other columns are current.

Exhibit 2. A hypothetical seed portfolio, read on 30 June two ways

A hypothetical seed fund holds 10 companies, each owing a monthly package 30 days after month end. Seven sent May’s numbers on time; three have gone quiet.

Company	Last package	Days overdue on 30 June	Runway at last package (months)	Runway on 30 June if burn is unchanged (months)
A	May	0	16.0	15.0
B	May	0	20.0	19.0
C	May	0	18.0	17.0
D	May	0	20.0	19.0
E	May	0	15.0	14.0
F	May	0	20.0	19.0
G	May	0	10.0	9.0
H	March	31	6.0	3.0
I	February	61	6.0	2.0
J	January	92	8.0	3.0
Dashboard of the 7 reporters			Median 18.0; none under 10
All 10, aged to 30 June				Median 14.5; 3 under 6

Hypothetical portfolio. Runway is cash divided by monthly net burn in the last package received; the last column subtracts the months since that package and assumes burn did not change.

The dashboard built from the seven reports shows a median runway of 18.0 months and no company below 10. Aged to 30 June, the full portfolio has a median of 14.5 months and three companies with two or three months of cash, all of them overdue. The example builds in the pattern the theory predicts: the silent companies’ last reports already averaged 6.7 months of runway, against 17.0 for the reporters. In a real portfolio the gap may be smaller, larger or absent, and nobody knows until the packages arrive. That is why the overdue count comes first: it says where to ask, and it proves nothing on its own.

The stakes are the reserves. If their numbers are as weak as their last reports suggest, each of the three silent companies will need a decision on follow-on money, a sale or a wind-down. That decision would fall due before the 120 days its fund has to report the quarter to LPs.

But a ratio from a few dozen customers misleads even when the package is on time

The rows that do arrive carry a second bias, set by their size. SaaS Capital’s 2026 survey puts the median contract value of a private software company at $24,266 (SaaS Capital, 2026), which implies about 41 customers at $1M of annual recurring revenue (ARR). Two departures move such a company’s churn rate by almost 5 points. A ratio’s noise shrinks only with the square root of its base, a result Abraham de Moivre published in 1730 (Wainer, 2007). Halving the noise in a churn rate therefore takes four times as many customers.

Lighter Capital, a lender, puts the 2025 median annual customer churn of the 155 private B2B software startups in its data at 16.25% (Lighter Capital, 2025). On a binomial model at that rate, with customers leaving independently, a company with 10 customers reports zero churn in about one year in six. It reports 25% or worse in about one year in five. At 100 customers the reported rate stays between 9% and 24% in 95% of years. A reading of 25% or worse turns up 1.6% of the time, so a bad number at that size is a signal.

Rank 30 companies of mixed size on churn or net revenue retention (NRR), and the smallest will tend to fill both ends of the table. Wainer found the same among 1,662 Pennsylvania schools, where the smallest were over-represented at both ends of the reading scores. No study we know of has measured the effect in venture portfolios. If reserves follow such a ranking, follow-on money drifts toward a lucky year and away from an unlucky one. The fund then gives up ownership in the company that deserved the reserve.

A written rule for when a ratio may enter a comparison prevents that. A ratio is ranked across the portfolio only when its denominator holds at least 100 customers and no single customer accounts for more than 5% of it. Its window must also have run its full length, 12 months for a retention cohort. The 5% cap limits any one customer’s effect on a retention ratio to 5 points. Even the SaaS Metrics Standards Board, which allows two methods for NRR, says its preferred cohort method “may not be helpful for a very early stage company” (SMSB). Below the gate the fund still computes the ratio and prints it beside its customer count and its range, without a rank.

Therefore a portfolio monitoring package can hold the same inputs at every stage

Every company sends the same raw inputs, and the fund computes each ratio from them with one formula of its own, so 30 companies cannot return 30 definitions of NRR. As a company grows, that rule decides which ratios join the comparison (Exhibit 3). Principles 3.0 points GPs to ILPA’s Portfolio Company Metrics Template, but that was built in 2019 “for buyout and growth equity funds” (ILPA, 2019). It asks for EBITDA, net debt, interest coverage and covenant status, and has no field for recurring revenue, burn or customers. Its update adds company-specific KPIs and is due in January 2027 (ILPA, 2026).

Exhibit 3. Portfolio monitoring by stage: the same inputs everywhere, ratios added as denominators grow

	Seed	Series A	Series B	Growth
ARR, as a loose guide	Under $1M	$1M to $5M	$5M to $20M	Over $20M
Customers at the median contract value	Under 41	About 41 to 200	About 200 to 430	Over 430
Inputs collected every month	Cash, net burn, ARR, customers at the start of the month, won and lost, headcount	Seed inputs plus the ARR bridge (new, expansion, contraction, churned), gross profit, sales and marketing spend	Series A inputs plus ARR by customer cohort	Series B inputs plus free cash flow
Ratios the fund computes	Runway; growth as a trend	Plus gross margin, burn multiple, logo churn shown with its range	Plus gross and net revenue retention, CAC payback	Plus Rule of 40, revenue per employee
A true churn rate of 16.25% reads, in 95% of years	4% to 32% at 25 customers	9% to 24% at 100	11.5% to 21.5% at 200	13% to 19.6% at 500

Source: customer counts are our arithmetic from SaaS Capital’s 2026 median contract values ($24,266 across all companies, $25,278 at $3M to $5M of ARR, $46,788 at $10M to $20M); churn ranges are binomial, computed by us on Lighter Capital’s 2025 median. Ratio definitions follow the SaaS Metrics Standards Board and Sacks (2020). CX Cash synthesis; the stage bands are ours.

But a count or a gate only says where to look; attention changes the outcome

Academic research on portfolio monitoring in venture capital measures the attention investors pay, and finds that attention matters. The cleanest test is a natural experiment: new airline routes that cut a venture capitalist’s travel time to companies already in the portfolio. Shai Bernstein, Xavier Giroud and Richard Townsend found that a new route raised the company’s patents and the likelihood of a successful exit, with the effect driven by lead investors (Bernstein, Giroud & Townsend, 2016). Distance is the normal condition: the median portfolio company sat about 200 miles from its lead VC, and around 40% more than 500 miles away.

Exhibit 4. What research on venture monitoring finds

Study	Sample	Finding
Lerner (1995)	VC-backed private firms	VC board representation rises around CEO turnover while other outsiders stay constant; distance shapes which VCs sit on boards
Gompers (1995)	Random sample of 794 VC-backed firms	Investors stage capital, gather information periodically and keep the option to stop funding projects unlikely to go public
Bernstein, Giroud & Townsend (2016)	22,986 companies; survey of 306 VCs	A new direct route raises patents by 3.1%, citations per patent by 5.8%, the likelihood of going public by 1.0% and of a successful exit by 1.4%; almost 90% of surveyed VCs would visit more often
Gompers, Gornall, Kaplan & Strebulaev (2020)	Survey of 885 VCs at 681 firms	60% interact with portfolio companies at least weekly in the first six months; 18 hours a week on portfolio companies; a typical respondent holds 5 board seats

Sources: Lerner, 1995; Gompers, 1995; Bernstein, Giroud & Townsend (2016); Gompers et al., 2020. Survey figures are self-reported.

Lerner’s result shows attention rising when a chief executive is replaced. Gompers’s shows what the information is for: staging lets investors keep “the option to discontinue funding projects with little probability of going public”, an option that loses value with every month of cash spent before the investor knows.

Attention is scarce. Bernstein’s respondents visited a given company 9.93 times a year on average, and the weekly contact in the Gompers survey covers only the first six months. The overdue count is one cheap way to choose which company gets the next hour.

Therefore the silence reaches the marks LPs are paid on

Here portfolio monitoring meets valuation, because a fund that does not hear from a company still has to value it. The International Private Equity and Venture Capital Valuation (IPEV) Guidelines say that “Fair Value should be estimated at each Measurement Date (each time Fair Value based Net Asset Value (NAV) is reported to investors (LPs))”, which under ILPA’s guidance means every quarter. For early-stage holdings they ask the valuer to weigh qualitative factors, including whether cash burn is above, at or below expectations.

The guidelines add that “The Price of a Recent Investment is not a default that precludes re-estimating Fair Value at each Measurement Date” (IPEV, 2022). A company that has stopped reporting leaves the burn question unanswered, and the easiest number left to the valuer is the last round’s price, the default the guidelines rule out.

Marks can carry most of a venture fund’s reported value for years. The median 2017-vintage US venture fund on Carta reported a 1.64x net TVPI in Q1 2026, of which 0.31x had been paid out (Carta, 2026). About 81% of its value was still a mark after roughly nine years (our arithmetic). Secondary buyers price that value at a discount: venture LP interests traded at 79% of NAV in the first half of 2026 (Jefferies, 2026), a 21% cut. The discount has several causes, illiquidity first among them. Marks built on stale information are the one cause an LP can ask about directly, and the fund’s carry and its next fundraise rest on the same marks.

If selection makes a venture fund, why count the silence?

The strongest challenge to this view of portfolio monitoring comes from research on where venture returns originate. Morten Sørensen’s 2007 study found that companies backed by experienced VCs go public more often, partly because the investors help and partly because they pick better companies, “with sorting almost twice as important as influence for the difference in IPO rates” (Sørensen, 2007). Venture capitalists agree. In the Gompers survey, 49% ranked deal selection as the most important source of returns and 27% post-investment value-added (Gompers et al., 2020).

On this view, hours spent on the failing quarter of a portfolio are taken from the companies that will return the fund. Sørensen and the surveyed VCs have the better of the argument about value: a fund is made by what it picks and by its few large winners, and partner hours should follow them. Their case does less damage to the narrow practice argued here.

Counting overdue packages costs a date in a contract and a column in a review, and no partner hours. A company that fails unwatched can cost more than its own write-off, which may be unavoidable. There is the follow-on money committed without current numbers, the mark reported to LPs, and the months in which a sale or an orderly wind-down was still possible. Attention may also matter most at the earliest stages: Bernstein and his co-authors put the effect of shorter travel times on patents at about 4.5 times larger for early-stage companies, though the difference for exits is not statistically significant. Gompers and colleagues offer the simplest reason to keep watching: “even late-stage VC companies have a relatively high rate of failure.”

Count the overdue packages before the averages

The checks fit on the reporting calendar, and the last three belong to the LP.

When	Check	Whose
At the financing	The monthly clause in the investors’ rights agreement, with its 30 days enforced; without a date, a missing package cannot even raise a question	Fund
Day 31 after month end	A call to any company whose package has not arrived, rather than a note for the quarterly review	Fund
Each portfolio review	The overdue count in the first column, every runway aged by the months since its package, and the number of companies each average leaves out	Fund
Before any ranking	Every ratio computed by the fund from the same inputs, ranked only once its base passes the written rule and otherwise printed with its customer count and range	Fund
Before a bridge or follow-on	A current package as a condition of any money from reserves	Fund
Each quarter-end mark	Silent companies marked under the IPEV rules instead of at the last round’s price	Fund
Each GP quarterly report	The share of NAV in companies more than 90 days overdue, how those holdings were marked, and how many companies each portfolio average covers	LP
From Q1 2026	The ILPA templates, with interim IRRs read in light of CalPERS’s caveat	LP
When the GP’s own report is late	The same reading one level up: a reason to ask the GP why	LP

The calendar can prompt the call before the wire request does

A portfolio monitoring dashboard shows the companies that answered, often with ratios their customer counts cannot carry. Theory says a founder with very bad numbers can stay silent while the investor cannot tell whether the numbers exist, and listed-company evidence says bad news tends to arrive late. For private companies the link is untested, so a late package proves nothing; it is a prompt. A fund that fixes its dates and counts overdue packages gets that prompt from the calendar, early enough to ask, instead of from the wire request.

Sources

• Bernstein, S., Giroud, X. and Townsend, R. R. (2016). The impact of venture capital monitoring. The Journal of Finance 71(4), 1591–1622.
• CalPERS (2026). Private Equity Program Fund Performance Review, data as of 31 March 2026.
• Carta (2026). VC Fund Performance: Q1 2026.
• European Securities and Markets Authority (2020). Fast Track Peer Review on the Application of the Guidelines on the Enforcement of Financial Information (ESMA/2014/1293) by BaFin and FREP in the Context of Wirecard (ESMA42-111-5349), 3 November.
• Givoly, D. and Palmon, D. (1982). Timeliness of annual earnings announcements: some empirical evidence. The Accounting Review 57(3), 486–508.
• Gompers, P. A. (1995). Optimal investment, monitoring, and the staging of venture capital. The Journal of Finance 50(5), 1461–1489.
• Gompers, P. A., Gornall, W., Kaplan, S. N. and Strebulaev, I. A. (2020). How do venture capitalists make decisions? Journal of Financial Economics; NBER Working Paper 22587 (2016).
• ILPA (2019). ILPA Principles 3.0: Fostering Transparency, Governance and Alignment of Interests for General and Limited Partners.
• ILPA (2019). Portfolio Company Metrics Template, 19 March.
• ILPA (2025). ILPA Releases Updated Reporting Template and New Performance Template for Industry Adoption, 22 January.
• ILPA (2026). ILPA Portfolio Company Template (draft update and comment period).
• IPEV Board (2022). International Private Equity and Venture Capital Valuation Guidelines, December edition.
• Jefferies (2026). Global Secondary Market Review, July.
• Kothari, S. P., Shu, S. and Wysocki, P. D. (2009). Do managers withhold bad news? Journal of Accounting Research 47(1), 241–276.
• Lerner, J. (1995). Venture capitalists and the oversight of private firms. The Journal of Finance 50(1), 301–318.
• Lighter Capital (2025). 2025 B2B SaaS Startup Benchmarks.
• Milgrom, P. (2008). What the seller won’t tell you: persuasion and disclosure in markets. Journal of Economic Perspectives 22(2), 115–131.
• National Venture Capital Association (2025). Model Investors’ Rights Agreement, revised October 2025.
• National Association of Private Fund Managers v. SEC, No. 23-60471 (5th Cir. 5 June 2024).
• SaaS Capital (2026). What is the average deal size for private SaaS companies?
• SaaS Metrics Standards Board. Net Revenue Retention (NRR).
• Sacks, D. (2020). The burn multiple. Bottom Up, April.
• Sørensen, M. (2007). How smart is smart money? A two-sided matching model of venture capital. The Journal of Finance 62(6), 2725–2762.
• US Securities and Exchange Commission (2023). Private fund advisers; documentation of registered investment adviser compliance reviews. Federal Register, 14 September.
• Wainer, H. (2007). The most dangerous equation. American Scientist 95, May–June.
• Wikipedia. Missing data; Wirecard.

Deadlines come from the model contracts, guidelines and rules listed; the studies by Givoly and Palmon, Kothari and co-authors, Lerner, Gompers (1995) and Sørensen are cited for the findings their abstracts state. Ages, ratios and shares with no citation are calculated from the reported data, as are the comparisons marked “our comparison”; the portfolio in Exhibit 2 is hypothetical. Customer counts divide ARR by SaaS Capital’s median contract values, the churn ranges assume customers leave independently, and the stage bands and the ratio gate’s thresholds are our judgement. Carrying the listed-company evidence over to private companies is our inference, as the text says.