Quibb lets you share what you're reading for work. Use Quibb to share news about your industry, discuss what matters, and see what colleagues are reading.
Our mission is to connect professionals over business news and informed commentary — targeting every industry, profession, and geography.
Excellent advice from Feastie co-founder Valerie Coffman.
"You can drive serendipitous discovery simply by offering users a selection of most popular content or editor’s picks." -- completely agree with this. If you can segment your userbase and tailor your editor's picks to each segment, even better. This can even scale up with you for a time, as you create and recommend specific content to increasingly specific segments. Of course, if you collect engagement data throughout this whole process, when you've hit product market fit, you already have a dataset on which to train your recommendation system.
So I guess we need some kind of definition of what exactly is a recommendation engine, because segmenting users and displaying content based on that could definitely be considered a recommendation engine, in my view. I suspect the author had something more specific in mind, but really - it's just an implementation detail, no?
Maybe I am off base, but I use Quibb primarily as a recommendation engine. It is where I find the articles I read on a daily basis. It isn't done algorithmically, instead it is done by peer recommendations. I'm not sure that doing it with an algorithm would be per se any better/worse, but I certainly wouldn't be opposed to it. I don't see a problem with a rec engine as a service or MVP. The challenge is obviously creating one that works and returns good results. The same could be said for any MVP it needs to work and deliver on the core value. MVP isn't code for a shitty or hastily built product. An MVP could take 5 years if there is really that level of technical challenge.
The fundamental idea behind recommendation engines is relevancy. This can be accomplished in a variety of different ways ranging from the simple rule to the extremely complex algorithm. While I agree that "one does not simply build an MVP with a recommendation engine" if the definition is an extremely complex algorithm. I disagree that a basis of relevancy in what you present to a given user profile can't be included out of the gate. I have worked with a number of organizations, from startups to enterprises who have leveraged some of our tech to create recommendation engines, or relevancy engines to help present optimized content to a given user.
I totally agree with the author's main point that you shouldn't build a collaborative filter or content analysis recommendation engine as part of your MVP. The effort required to do so would be immense. And even if you had the world's best algorithm, it wouldn't do much good until you had tons of content in your system. Of course, one exception is if your MVP is a collaborative filter or content analysis recommendation engine. :)