Signals

One of the biggest problems I have with a lot of data-related businesses is that they only provide half of the value that I think they should. While it is indeed valuable for a company to help you look at your information and dissect it into smaller and smaller chunks, to me it’s even more valuable for that company to provide timely recommendations based on that data.

I said it an awful lot in VC pitches, to myriad levels of success: what’s already happened will never be as interesting or valuable as what’s going to happen.

Take something like Google Analytics as an example. numberFire has been set up on Google Analytics since the day it was first put online. It can tell me what piece of content was the most successful, how long people on Android phones stay on the site, and any number of vaguely useful things. What it doesn’t tell me however is that the ideal headline length is seven words, or when an article is showing the initial signals of virality.

This is an equivalency to the frustration that I had when I founded numberFire in the first place. Far too many sports analysts were just regurgitating the information, as if all situations were apples to apples (is playing the Seahawks the same as playing the Browns? Hmm..) and broad comparisons were predictive of future events. By using machine learning and various other data mining techniques, we took wide swaths of sports data and turned it into something predictive, because having predictive knowledge on future events is a very valuable piece of information to have.

I would love a dashboard that would tell me how many page views I can expect, just on the basis of the content matter and the headline. I would love to know when an article is about to go viral, so that I can highlight it on social media and re-order the items on my homepage to reflect that. I would love know on which channels and what times I should double my marketing efforts, suggesting keywords that are undervalued in the market, or that a user segment is being underserved based on current supply. I don’t really see why that’s not possible, given what Google Analytics, Heap, KissMetrics, and everything else in our reporting stack knows.

And what’s even more strange about it is that everyone seems to be creating companies that harness the information, creating largely fungible businesses to what is already out there – I should know, because I get about fifteen cold emails from them daily.┬áNo one is creating a business that simply ingests that information, and serves you the recommendation at the precise moment. It’s a missed opportunity.


No responses yet

Leave a Reply