Predicting Startup Fundraises Using Job Post Data
May 11, 2022 - 6 min read
The public doesn’t have much insight as to when a startup might raise money; we don’t know when it will happen until it’s dramatically announced on techcrunch. But is there any way we could predict when a company is fundraising? In particular, could we predict if a startup is about to make a fundraising announcement solely by looking at increases in job posts by the company?
Why Try to Predict Startup Fundraising?
Hmm, good question, why would you want to predict if a startup will fundraise soon? A few reasons:
- If you’re applying for a job at startups, applying before a fundraise is announced gives you an advantage: the application process won’t be as competitive as it will be after the announcement, and it’s possible your stock options will be issued at a cheaper strike price than if you joined after. Without writing a treatise on stock options, this is ultimately better for the applicant.
- If you’re a venture capitalist or angel investor, it might be a way to find companies that are fundraising now, giving you a chance to invest yourself before the round is full. This wouldn’t necessarily be helpful for sourcing significant, lead investment opportunities but it would help investors find deals that they didn’t know about and potentially get into them before it's too late.
- Or maybe you’re just a fan of predictions, I don’t know.
coolstartupjobs collects and analyzes job posts from startups over time to help jobseekers identify which startups are growing the most right now. But we can also use this job post data we collect to run all kinds of cool analyses, including predicting fundraising announcements.
Example
As one example, take a look at this chart of historical job post data for Alloy Automation, an automation platform for e-commerce companies:
This chart shows the number of jobs posts Alloy Automation had posted on their website on a given date. You can see a spike in job posts from 1 job post to 8 job posts on January 23. This is a huge increase for an early stage startup that has only had 1 job post for months. Then 1 month later on February 22, this announcement on techcrunch:
The job post spike predates the fundraising announcement by one month!
Depending on the exact timing of the fundraise, that may have been a valuable signal to have: an investor could have contacted the company about investing and maybe squeezed into the round, or a job seeker could have joined when the candidate pipeline wasn’t full and gotten stock options at a lower strike price.
Larger Scale Test
The previous example shows that you can sometimes predict fundraising announcements from job post spikes, but will it work on a larger dataset? Will it blend? Let's see!
In our dataset, from 2021-09-24 to 2022-04-01, we identified 91 “job post spikes” among job post data from 371 companies. Of these 91, 25 occurred shortly after a fundraising announcement; this means the job post spike occurred after fundraising rather than before. You’d expect some companies to do this; they don’t post jobs until the fundraising has been closed and announced.
These spikes can be discounted; it's unlikely a startup would fundraise, then fundraise again a few months later. That leaves us with 91-25 = 66 job post spikes that we think might predict future fundraises. Let’s look at the contingency table for the results:
| Actually Fundraised (in next 3 months) | Did Not Actually Fundraise | |
| Predicted Fundraise | 17 | 49 |
| Predicted No Fundraise | 57 | 248 |
There were 74 actual fundraises that occurred out of the 371 companies. Job post spikes correctly predicted 17 of them (specificity=23%, if you’re into that) within the next 3 months. This means of our 66 predictions, 49 of them were incorrect. Since there were 74 fundraises, this means there were 371-74=297 no-fundraises. We predicted 248 of those correctly and 49 incorrectly (sensitivity=84%).
At first glance, these results might seem discouraging; after all, we missed 57 of the 74 fundraises. This would be terrible if we were testing for a disease (well, it might beat Theranos). But it’s actually kind of amazing you can predict about a quarter of startup fundraise announcements purely from job post data; that’s a lot of signal in a single, publicly available variable.
Using job post spikes as a predictor also correctly rules out a large number of companies that don't fundraise. If were you reaching out to companies to see if they're fundraising, or if you were applying to jobs, this predictor would significantly cut down on the number of companies you would need to contact while, by itself, preserving one quarter of all companies you would want to contact.
Plus, you now have some information about the company that will help you contact them more naturally: "Hey, I'm an investor, I saw your company just posted a bunch of sales roles; are you raising right now?"
Why Not More Than a Quarter?
We consider these great results for just using job post data, but why aren’t they even better? Well, several reasons:
- Companies don’t always announce fundraising rounds. They don’t have to and there are reasons they might not want to, like not alerting competition.
- Companies might be doing a lot of hiring, but aren’t fundraising. This can certainly happen if business is going well and/or they haven’t aggressively burned since their last fundraising round. These would still be great companies to contact; if a company is doing a lot of hiring without fundraising, that means they’re in a great position at the moment.
- Companies did fundraise but didn’t increase headcount. The money might be going to contractors, manufacturing, other non-salary uses, or just straight to the bank account because the CEO thought the fundraising environment was particularly favorable at the moment.
So while it isn’t perfect, identifying a quarter of fundraises purely from job posts is pretty damn good. Will it blend? We say yes!
And if you do happen to be looking for a job, enter your email address to get job posts from startups that growing right now: