The Imminent-Raise Signal Reference for Watchlist Companies

This is the lookup an early-stage investor keeps open to decide, for a company already on the watchlist, which public behaviors mean a round is coming and which are noise. It is written for partners, platform and talent staff at funds, and angels who already run a watchlist and now need to score its signals. Each signal below carries three things: what it proves about an approaching raise, how much head start it buys, and the specific way it lies to you.

Most public guides list the readiness signals a founder uses to judge their own timing, or sketch deal-sourcing strategy from the air. This one goes the other way. It scores external, observable company signals by lead time and failure mode so you can set an alert on the ones worth watching and mute the rest.

Which signals lead, and by how much

The reliable signals sit at opposite ends of the timeline, not the middle. Engineering acceleration leads by weeks to months because it precedes product milestones; the Form D lags to within days of the round but is legally certain. The mushy middle, a single job post, is where the ghost-job noise lives.

Here is the vendor-reported ladder. Treat the week-counts as estimates to validate against your own closed deals, not as laws.

A separate sourcing view argues the useful window opens 6 to 12 months earlier than the round, treating traffic plus hiring as the core leading pair. The two camps are not rivals. One reads engineering and traffic as the true first movers, early and noisy. The other treats the Form D as the actionable trigger, late and near-certain. You want both wired.

The Form D: the legal tripwire

A Form D is the hardest signal to fake and the latest to arrive. It is legally required within 15 days of the first sale of securities, filed on the SEC's EDGAR system, and it typically appears 7 to 14 days before any press release. That is a legally disclosed funding signal, not an estimate or a rumor.

The documented case: one company filed a $150M Series C Form D 12 days before the announcement. That 12-day gap is the head start on offer. The SEC takes the deadline seriously enough to have imposed penalties of $60,000 to $195,000 in December 2024 for late filing, which is part of why the record is dependable.

What it proves: a round has closed, or a sale of securities has begun. What it does not prove: that a round is coming. A filing is confirmation, not prediction. Non-US entities and some deal structures never file at all, so absence tells you nothing.

Engineering signals: the longest legal head start

Engineering acceleration is the earliest reliable read, at 6 to 12 weeks, and it is hard to fake because commit velocity and contributor growth accumulate publicly over time. You cannot retroactively manufacture a quarter of shipping. Engineering acceleration precedes product milestones, and product milestones precede the fundraise.

What to monitor: commit velocity, contributor growth, and new repos on a company's GitHub org. What it proves: the team is building toward a milestone that will anchor a pitch. What it looks like when it lies: a one-off spike from a migration or a dependency bump reads like acceleration but decays. Require it to compound across weeks.

Because these signals are un-alerted by most crowd-followed tools, they hold their edge longer than hiring or traffic. The durable advantage is not access to the feed. With more than 80% of data-driven VCs now using LLMs in sourcing and 35% saying tools source about half their deals, the feed is shared. The advantage is a sharp thesis that filters it.

The edge is not the feed everyone now shares; it is the un-alerted signal and the thesis that reads it.

Hiring signals: the most watched, least trustworthy

Hiring is the noisiest signal on this page, and it is the one most people over-weight. A raw job-posting spike is discounted heavily before it means anything, because a large share of postings are fake, stale, or theatrical.

Ghost jobs run 18 to 22% of postings, up from 12 to 15% in 2022, and one in three postings never results in a hire. In June 2025, employers reported 7.4 million openings but made only 5.2 million hires. Hires per 10 postings have halved from about eight to about four. Motive matters to you specifically: some firms post jobs precisely to project growth to investors, competitors, or their own staff.

Note the spread by source. On the noisiest public reads, a hiring spike is discounted 18 to 33% before it counts. Ashby's cleaner ATS base rate shows 82% of 2024 jobs filled, so only 18% unfilled. The noise floor itself depends on where you read the data.

There is still signal in the volume, if you read the cluster and not the count. Three postings in one department is a growth signal; five is a reorg or an expansion. And a posting still open after 60 days is a signal of frustration, not growth.

The first senior-finance hire

The exception inside the noisy hiring category is the first senior-finance hire. An upcoming fundraise is the clearest reason to make it, because diligence exposes weak models when you have the least time to fix them. Below roughly $25M ARR, or with a near-term need such as fundraising, a fractional CFO is usually the right call; a VP of Finance at Series A can manage investor relations, forecast, and help execute early rounds.

This hire is rare enough to carry real weight. In Refolk's index there are 4,311 current Head of Finance and VP Finance people across the United States, concentrated in the San Francisco Bay Area, New York, and Boston. Narrow that same query to the software industry with VP or Director seniority filters and matches drop to zero. The senior-finance-at-software-startup pool is thin enough that a single new such hire is a genuinely rare, high-signal event, not background noise.

A rare hire beats a loud one. A five-role posting cluster is already discounted by ghost-job base rates. A first senior-finance hire, clustered with data-room-type activity, is not.

When you need to resolve that thin set of people from plain-English criteria across LinkedIn, GitHub, and the open web, Refolk does the narrowing that manual sourcing cannot: ask for the exact hire and get the people, not a feed to sift.

The signal reference: what each proves and how it lies

This is the row-by-row lookup. Jump to the signal you are looking at and read across.

• Form D on EDGAR. Proves a round closed or securities sold. Buys 7 to 14 days. Lies by: absence for non-US entities; confirms a closed round, not an approaching one. Wire it as a daily poll.
• GitHub commit velocity and contributor growth. Proves the team is building toward a milestone. Buys 6 to 12 weeks. Lies by: one-off spikes from migrations or dependency bumps. Require weeks of compounding.
• Departmental hiring cluster. Proves growth intent when three or more roles open in one department; five signals a reorg or expansion. Buys 4 to 8 weeks. Lies by: ghost jobs at 18 to 22% and roles posted to project growth to investors. Discount the raw count.
• First senior-finance hire. Proves diligence pressure, the documented trigger for the hire. Buys weeks, ahead of the Form D. Lies by: routine scaling can also justify it. The tell is a first such hire clustered with data-room activity.
• Web traffic climb. Proves demand building. Buys 4 to 6 weeks. Lies by: a launch, a viral post, or paid spend inflates it once. Require multi-week compounding.
• Headcount inflection. Proves scale-up when a company crosses 20 to 50 employees or posts a first-ever role in a new function. Buys weeks. Lies by: single hires read as inflection. Watch for the function that did not exist before.

How this reference misleads: the failure modes

Every signal on this page has a documented way of lying. This section is the one to read twice, because a false positive costs you a wasted outreach and a burned first impression.

Hiring spike, false positive. Up to one in five postings is a ghost job. Check whether the role is fresh, filled within normal timelines, or a stale repost. A posting still open after 60 days signals frustration, not growth.

Form D lag or absence. Non-US entities and some structures never file, and a filing confirms a round closed rather than one coming. Verify the entity-to-company match on EDGAR before acting.

Traffic bump. A launch, a viral post, or paid spend inflates traffic without a raise. Require it to compound over weeks, not spike once.

Finance hire misread. A VP Finance can be routine scaling. The tell that it means a raise is a first senior-finance or fractional-CFO hire clustered with data-room-type activity, since diligence pressure is the documented trigger.

Single-signal action. Acting on one signal invites false positives. Enforce the compound-trigger rule: a funding-adjacent event plus a hire in a target role.

Announcement-day herding. Reaching out on the day a round is announced puts you in the same batch as every other vendor who set up the same alert. By then the signal is lagging, not leading.

Conversion-rate over-weighting. A seed company has only a rough 38% base-rate path to Series A, so an early signal is worth less than it feels. Discount lead time by the stage graduation odds.

Discounting by the funnel: why early signals are worth less than they feel

Early signals point at companies that mostly will not clear the next round, so lead time must be discounted by graduation odds. Seed-to-Series-A conversion now sits near 38%, down from 50% or more in 2020 to 2021, and the trend line is worse than the blended figure suggests.

The Q1 2018 seed cohort reached Series A at 30.6% within two years; the Q1 2022 cohort managed just 15.4%. The conversion rate roughly halved in four years. In Europe, median Seed-to-A time is 18 months and only 19% convert within 36 months. Pre-seed-to-seed runs about 45% within 24 months, and Series A-to-B about 55%.

The practical read: when you see an early signal on a seed company, multiply your excitement by the odds it graduates. A crisp GitHub acceleration read on a company that has a 38% shot at its next round is worth roughly a third of the same read on a company already through Series A. This is not a reason to ignore early signals. It is a reason to weight them.

The procedure: from watchlist to logged outreach

Run these seven steps in order. The first and last belong to a partner; the monitoring steps belong to an analyst. The whole loop only pays off if step one is real, because a pre-fundraise feed only rewards investors who already know what they are hunting.

Career-page monitoring deserves a note: it is the freshest hiring source, capturing postings before they syndicate to job boards, sometimes by hours or days. Public Lever and Greenhouse endpoints expose this directly.

The compound trigger is the load-bearing step. A compound signal is two or more conditions firing together, and the most predictive stacks are funding plus a new hire in a target role, funding plus website intent, or a specific tech install. These stacks exist because each signal alone sits inside a documented false-positive band. Pairing them is how you climb out of it.

Verify before you call it a raise

Run this before you promote a watchlist company to active outreach. If you cannot tick every box, you are acting on noise.

Keeping this reference current

Signals decay, so the numbers on this page are a starting calibration, not a permanent one. The week-counts are vendor-reported rather than peer-reviewed, and one practitioner concedes their own signal-half-life claim, roughly 30 days at peak intent decaying to a firmographic baseline by day 90, is pattern-match, not a controlled study. Treat every lead-time figure as a hypothesis you test against your own closed deals.

Two things to re-check on a schedule. First, the ghost-job base rate: it moved from 12 to 15% in 2022 to 18 to 22% more recently, so the discount you apply to hiring signals should track it. Second, the conversion funnel: seed-to-A has halved cohort over cohort, so the graduation odds you use to weight early signals will keep moving. B2B software spending at venture-backed firms rises 40 to 60% in the two quarters after a round, which is a reminder that the post-raise behavior is the loud, lagging half most tools already chase. Your edge is the quiet, leading half, and holding that edge means re-tuning these thresholds as the shared feed compresses everyone's advantage.

Where a signal's false-positive rate is not established publicly, as with domain-registration and trademark filings, do not assign it a lead time you cannot defend. Log it, watch it, and let your own data decide whether it earns a monitor.