I recently spent time with some of my friends at dinner. We went around the table discussing weekend plans and work schedules. Many of my friends work in medicine, and I like to pester them with questions about bumps, scratches, and rashes on the regular. That evening, they returned the favor. Since I work with startups, I regularly get into conversations that begin with “I have an idea for a company.” But after some time (and explaining), the discussion tends to veer to queries and suggestions about KingsCrowd’s process of evaluating companies.

KingsCrowd customers and founders, like my friends, have limited visibility on how startups make their way through our pipeline. I want to lift the veil. I want to answer the questions that my friends asked me — because you are probably thinking the same thing. 

I’ll start by walking through our diligence process, beginning with the startup going live on a platform. Then I’ll follow up with answers to some of the most common questions that I get.

When a startup goes live on one of the many platforms that we cover — such as Wefunder, StartEngine, Republic, or another — the first step is to collect basic information. KingsCrowd takes note of data such as the name of the company, the platform it’s raising on, the company’s headquarters, and its website to track the raise. At this point, there is no numeric rating data available. The company’s page on KingsCrowd would look something like this:

Once the startup is in our system, it goes into a queue for our data team. Our amazing team of data analysts take time each week to gather tons of data that then goes into our rating system. Our primary data analysts will size up the market for each startup, collect data on the product margins, and evaluate the background of the founder, among other metrics. All of that information goes into  our proprietary algorithm, which rates each startup on a scale of one (the lowest rating) to five (the highest rating). 

Gathering more than 50 data points on a single company, however, can be daunting. And while many of our data points are empirical, some are more subjective. We try to make our information as objective as possible, adhering to a data input manual that clearly outlines the limits of each data point. Still, analysts will have differing opinions from time to time. For example, one analyst might think a company’s product is highly differentiated, while another might consider it just average. That is why company data is never up to just one analyst. 

Once an analyst has collected data, a second analyst will come through to check that data for quality and send the original analyst any suggestions or corrections. Only then is the company turned over to the investment research team. At this point, a KingsCrowd user would see a company page that looks like this:

This is where I come in as a member of the research team. Each week, the research team gets a batch of companies that have already been rated by our amazing data analysts. In a typical week, there are around 20 companies. I spend nearly one whole day each week going through each company and deciding on whether or not I think it deserves coverage — and my colleagues do the same. 

Once we’ve reviewed these newl;y rated startups, I meet with my team of four other researchers and our managers to discuss each company. Before the meeting, we submit a form to vote for each company (to cover or not to cover). Beginning with the companies that received a universal coverage vote and working our way through the more controversial raises, we assign one or two companies for each researcher to cover the following week. After that meeting, we get in touch with the founder(s) of our respective company. And from there we are off to the races, working on what will eventually become an Analyst Report.

This is the general framework that our data analysts and researchers follow each week. But like my friends, I am sure this explanation gave you more questions. Here are some of the questions I received between plates of green beans and pie:


How many companies can you cover in a week?

We are a small team. Right now, we can cover a maximum of around eight companies each week. Every company will receive a rating, but only a select few will receive an Analyst Report. As we grow our team, we will expand that coverage. 


How do you decide who covers what company?

This is a great question! And the answer is typically general interest. We try our best to have researchers cover a company that they are particularly bullish on or one that matches the researcher’s experience. For example, I have a particular interest in artificial intelligence and machine learning. 


How long does it take for an Analyst Report to become available from the time the company goes live?

The entire process of data collection and review takes around a week. When I initiate coverage on a company, much of the timeline is actually in the hands of the founder. The quicker I can talk to a founder, the quicker I can write my report. And talking to founders is a crucial step in our research process for Analyst Reports. No one knows a company better than the person who founded. We often learn nuances about industries, business models, or competition from founders that would have been nearly impossible to find otherwise. It’s those hard-to-get insights — paired with our analysis of the raise’s investment terms — that we want to provide to KingsCrowd members.

Assuming a founder gets on the phone with me the week that the company was assigned, I can usually produce a Report by the end of the following week. The report then goes through our content team members, who correct our grammar, polish our writing, and fact check our pieces. A quality report takes time, but we would much prefer quality over speed. Please be patient with us! 


Can a company that is rated highly be passed on for coverage?

We flag companies that are rated above a 4.5 and automatically review them, regardless of the voting outcome. In the same vein, we also try to cover raises that are especially popular and have raised significant funding very quickly, whether or not we think they make a good investment opportunity.


What happens to the companies that don’t receive coverage?

Companies that do not receive coverage  still have a numerical rating from our algorithm. But there will not be an Analyst Report for that particular raise. These tend to be companies that we are not particularly bullish on, are not rated above a 4.5, and/or are not hugely popular investments. 


Beyond the 4.5 rating threshold for coverage, do the numeric ratings impact your decision to cover companies?

This is one of my favorite questions to answer because it speaks to the intricacies of early stage investing. Our algorithm is impressive. It can synthesize hundreds of data points and, in most cases, spit out a numeric rating that we find appropriate for a company. But it isn’t perfect. Even though a company might look flawless on paper, we often still need human intervention to get a true feel for a startup’s potential. 

Uber is one fantastic example of this conundrum. If our algorithm traveled back in time to 2009 when Uber was freshly founded, the company would likely have earned a very poor market rating. Back then, the ride share market hardly existed. Sometimes startups create the market in which they exist, as is the case with Uber.  

Just like startups can create markets in and of themselves, oftentimes macroeconomic factors contribute to market creation or growth. Most recently, the COVID-19 pandemic accelerated the market for services like video conferencing. It would be impossible for our algorithm to consider the infinite number of factors that go into market climates. Of course, we try to minimize the effect that outliers may have by weighing different features more heavily and having hundreds of features that contribute to the end score. 

The Uber example is exactly why a startup with a rating of 2.3 might receive a Deal to Watch Analyst Report and a company with 4.3 stars may be passed up for coverage. Investing is an art, not an exact science.