This Machine is Trying to Take my Job


Wow. I am blown away by how many of you stepped up to help me prank my husband last week. We had him going there. He was convinced Craigslist accidentally gave out his number with the wrong ad. It was hilarious.

Unfortunately, I could not contain my giggling and he figured it out fairly quickly. However, this was the prank that kept giving. Even though he knew it was a prank, every time a text message rolled in over the next several days, it served as a bitter reminder of how good his wife got him.

This was easily my best work yet, and I have all of you to thank for that.

Topic of the Week: This Machine is Trying to Take My Job

Dan Primack reported yesterday that GV (Google Ventures) utilizes an algorithm to permit or prohibit new investments. Per Dan, deal info is inputted into “the Machine,” an algorithm processes the data, and out pops investment recommendations in the form of Red, Green, Yellow traffic light signals.

There’s nothing novel about using an algorithm to make investment decisions. This concept is old news and boring. Algorithms are just a fancy version of investment criteria scorecards.

While investment criteria scorecards vary greatly firm to firm, many private equity and venture capital firms use them. They help investors institutionalize deal evaluation and take subjectivity out of the decision-making process (unless of course, the scorecard comes back with an answer the investor doesn’t like, then they ignore the scorecard all together…kidding…sort of…not really…).

ROND has a version of the “Machine” in an excel spreadsheet. We even use a similar methodology with Red, Green, Yellow scores. Our scorecard contains a series of yes/no questions that help us determine whether a potential investment opportunity aligns with our investment thesis.

Setting aside the debate regarding how much “gut” plays into being a successful investor, there are two primary challenges with investment scorecards:

First, it’s difficult to develop your question list or “criteria” in a way that eliminates subjectivity. Most investors want to invest in businesses with strong growth profiles, but what makes growth strong vs. meh? “Does the Company have strong growth?” is open for interpretation, while “Is the Company growing at least 25% per year for the last five years?” is much less subjective.

As you get into more nuanced categories, such as the quality of management, competitive differentiation, etc., it becomes more difficult to create subjective parameters.

Second, once you have a scorecard that eliminates as much subjectivity as possible, it can be challenging to get access to the necessary information to fill it out. Filling out the scorecard requires an investor to: i) Talk with management; ii) Review company information; iii) Access data in subscription databases; iv) Talk with experts in the industry; and v) Research any publically available information on the business via Google (aka the subsidiary of Alphabet, aka the owner of GV) searches.

Hm…Let’s revisit my earlier statement on GVs algorithm being boring.

When I said the concept of using an algorithm to make investment decisions wasn’t novel, I meant it. What’s not boring and is, in fact, mindblowing, is the amount of data and information GV could in theory access to power the algorithm.

Google, Gmail, Youtube, Android, Google Maps, Waze, Nest, Google Home, the list goes on. Google knows everything about everything.

I’m dying to know just how GV uses Google data to power the algorithm, although it’s probably much more boring than what I’ve been cooking up in my head.

Regardless, every time I hear about GV passing or investing in a deal, I will think to myself: “What does the Machine know?”

Have a great weekend everyone!

About the author

Danielle O'Rourke

Recovering Investor. Mom. Wife.

Add comment

Danielle O'Rourke

Recovering Investor. Mom. Wife.

Recent Posts