I think this is a very interesting business model. I took the test myself and found the results quite accurate to what I already know I like/am good at. What I wonder though is that there are attributes that might not be easy or even possible to measure. While the success of this type of models will somehow democratize access to jobs (reducing the value of a brand name school or a shiny resume), it could also over simplify the recruiting evaluation process, setting up barriers for people who “in theory” don’t match with a specific position but could certainly do a great job at it and be successful due to other attributes.
Really cool tool. I agree that the selection bias exists. I don’t think it is a problem though. If this service is already providing an incentive to opt-in by offering discounts to people who evidence a lower risk driving behavior, Progressive could automatically assume that everyone who did not opt-in will be a riskier driver and hence, should have a higher price. This only becomes a problem if by making the default option more expensive to compensate for discounts, we stop being competitive compared to other players in the market. This would create a positive selection for Progressive and would push riskier drivers to the competition. If the resulting volume can still make the economics of the business work, then it is not a problem to have a selection bias.
I completely agree with the concerns you raise. Comparing behavioral variables to psychometric or demographic variables will always result in a better discrimination power from the first ones because they measure objective observable actions. Nevertheless, those preferable risk models are unavailable for populations of unbanked individuals and micro businesses, specially in developing economies. EFL’s model adds predictive power on top of traditional scores. Which means that banks who implement EFL’s score in their evaluation process will not stop using Equifax all together. While its susceptibility to “gaming” is significantly higher than a traditional score EFL has taken measures to counteract the risk of faulty data. I asked the Director of Innovation about this and he explained how sometimes it’s now “what” the individual answers, but “how” (e.g. how long does he take to reply, how many times does he change his answer).
About the regulatory outcry, I can totally see that happen in the U.S. In Latin America, on the other hand, we can even use demographic variables to find correlations with default rates. This, in the U.S. is discrimination and is forbidden. I understand how both these policies makes sense in the markets where they’re applied; having a less mature market where high quality behavioral data is scarce, many times you need to use all resources available regardless of how precise they are.
I think this is a great idea. For all the reasons you mention above, this initiative should provide a never ending flow of new ideas to explore (for free) to stay relevant to what their customers want and both to maintain a highly engaged customer base. My only concern with this would be that since as a company you’re receiving thousands of new ideas, that will most likely get discarded, there will be a lot of disappointed customers whose ideas where not selected. As in any contest or lottery, “winners” should be broadly publicized to make sure your audience sees this as a feasible track to get their ideas listened.
Love the concept! I am a junkie myself so couldn’t help to take the quiz and give it a try. Nevertheless, I also wonder if their strategy will ever take them to a “mainstream” position in the market. It’s true that they have a privileged position to do faster, more efficient product development. Nw they need to develop more traditional distribution channels to grow their sales volume and leverage their outstanding design capabilities.
Very good post. Really insightful analysis from the data you collected. I wonder if asking users to rank a subset of options is the alternative though. It is known that recent memories are more vivid that older ones, which could certainly introduce a bias towards the rank the place you just visited “higher” than the ones it’s compared against. I wonder if adding some type of disincentive to over-rate a restaurant would work in addition to forcing all customers to rate their experience before using the app again. For example, enforcing to rate individual criteria when giving a 4 or higher might disincentive customers to just give 4’s to most restaurants.
I loved this post! Totally agree with the potential of creating “offline matching” with this new service. I don’t really think that was the main purpose of the service but it definitely provides users with more opportunities to connect with people outside of their phones. In my own experience (and I’ve also heard this from some other dating app users), I don’t feel as excited to go on a first date with someone I haven’t personally met as I would feel when I have a first date with someone who I already know (and like). Generating this first casual, risk-free interaction might solve this problem. Nevertheless, the chances of meeting someone you’ll actually like on a random UberPOOL ride are undoubtedly lower than on dating apps.
I find some things very interesting here. I agree that Spotify is the clear leader in this space (even when it doesn’t have the largest user base compared to Pandora). Both services have direct network effects (since there not a lot of bargaining going on in the supply of music to these platforms) which primarily affect their ability to give better recommendations and to build more relevant playlists to their users. Nevertheless, they both provide very different customer value propositions: Spotify is 100% on-demand and Pandora is a radio service (I would dare to say that they address different needs too; I would assume that the Pandora user doesn’t really want to invest any time or effort in their “listening to music” experience… this is what is called a lean-back experience in the streaming industry). What does not make a lot of sense to me, is that Pandora has managed to convert 5% of its user base into a paid tier that offers only slightly better experience (no ads, unlimited skips). I wonder if Pandora developed a user experience comparable to Spotify’s, would it be more successful than Sporify in leveraging their huge userbase to convert them into a $9.99 tier?
Very interesting post. I myself am trying the Apple Music service and I am still on the fence wether to go with Spotify or Apple. I wonder about a couple of things:
1) It is true that network effects have a great impact in the sustainability of either of the two services. Nevertheless, Spotify was the first mover in this space and it already has a significant userbase with very high migration barriers (moving all your music from Spotify to Apple is too difficult or virtually impossible if you’re a heavy user).
2) Apple Music’s service still has a couple of bugs and somethings are not as seamless as one would want (e.g. moving from the U.S. iStore to anywhere else in the world could potentially cause massive losses of data). Will its brand name hold consumers and give them enough time to polish their functionality to provide a better service than Spotify?
Finally, I don’t have stats on this but I am sure that given the huge initial iTunes userbase (and credit card information) should trigger adoption in customers who were not affiliated to any streaming service, expanding the market. On the other hand, their product portfolio (no free ad-supported tier) should also make it easier to drive users to pay, whereas Spotify is having a really hard time convincing their free users to pay 9.99 for the premium service. I wonder if these two last attributes alone will allow Apple Music to reach their expected growth targets.
Thank you for your comment Jeff!
The challenges that you mention are definitely real and I think songkick should work on them going forward. I’ll give you my thoughts on each of them:
1. I did not find any explicit clarification on what happens if a user learns of a concert through songkick and then goes directly to the venue, but from what I understood from their business model, the small fee they receive per ticket sale is paid when the purchase goes through their app. One of the problems that they usually encountered is that some sales would not go through due to technical problems in the ticketing platforms, and that would also decrease their revenues. I believe that finally owning the whole process end to end, will allow them to capture all the value they are creating and be responsible for the platform stability, avoiding those losses in ticket sales.
2. I totally agree with your suggestions. I imagine they notify as soon as the concert is announced to allow you to save the date and track the event for a future purchase. Notifying when tickets are available is definitely a no-brainer, and I assume that once they integrate the ticketing platform to the app, they will have even more incentives to trigger purchases, so that should be a natural move.
3. Truth is, I think large ticketing platforms could definitely develop features similar to the ones of Songkick. What I do believe, though, is that it will be easier to change purchase behavior than to change discovery behavior. I see it this way, you get directed to whichever platform sells the tickets because it has the best price or because it’s the only one selling the concert you want. You are not necessarily attached or have regular interaction with ticketing platforms due to loyalty, but due to convenience. On the other hand, you do discover shows mostly in the same place, and this platform redirects you to buy the tickets. I believe that Songkick has an advantage since they already have a network of 10 million users and they are the ones who hold control of the “redirection” process.
Everything is uncertain still, though. I’m curious to see what happens next!
Really good post! What I found most interesting is when you talk about brand value. You say “Reviews on websites like TripAdvisor and social media have made it easy to trust unknown brands”. Airbnb, in this case, has gone one step further, allowing us to trust unknown individuals… going into their homes, or letting them in our own homes. When I think about it, a similar model existed before Airbnb with the common practice of “couch surfing”. This practice was obviously less structured and riskier. Attaching Airbnb’s brand to the same players and providing a two sided platform where you give and receive reviews, provides some sense of safety for both the host and the guest. I wonder if what’s happened is that hotels’ brand value has been eroded or that when we book a random person’s home for the weekend, it’s Airbnb’s brand value what’s competing with the hotels in our decision process.
Very interesting. I actually agree that this technology could be integrated with the activities customer support employees if the application with which they interact is well designed and user friendly. For example, call center operators usually have predefined scripts that they have to follow, that indicate what to say under which circumstances. If the application is able to classify data inputs from the customers voice inflections, tone, volume, pitch, speed, etc, into simple moods (e.g. indifferent, angry, curious, etc) this new classification variable just adds another layer to the decision tree that guides the operator speech.
Moreover, this could be integrated with voice recognition technologies (already available) to identify which words or particular scripts frequently trigger negative reactions in the consumers (we all know customer support employees often make us angry), and thus, companies could refine their customer support strategy.
Tilt’s model reminds me of Kickstarter in the sense that you can actually post any event or cause and get funded for it with very little control or supervision on wether you fulfilled what you intended to use the funds for (although Kickstarter actually does ask for results). The Sochi example you gave made me think on the risk of having “fake” causes or “jokes” posted that get funded as they go viral, sometimes just for fun (e.g. the potato salad Kickstarter project: https://www.kickstarter.com/projects/zackdangerbrown/potato-salad). I wonder how can Tilt avoid that kind of behavior from both the organizers and the funders.
This sounds great. I only see one potential challenge to maintain the integrity of the data they’re collecting. I understand they have an initial baseline to launch the app, which will be improved and updated with the images collected by future users. Therefore, whenever a user “scans” a new face, will he be able to create a new profile associated to that face? If so, the application should have processes in place that ensure that new users are created with the correct information and avoid having multiple profiles for the same user.