The drubbing of Grubhub. Could it have been predicted?

Yesterday, Grubhub’s stock took a beating. Goldman Sachs, downgraded its GRUB stock recommendation, post-facto, from “Buy” to “Neutral”. Goldman said they got it wrong and “significantly underestimated the impact of competitor investments on customer behavior across the space while overestimating both the potential for industry-wide growth and Grubhub’s ability to maintain share”. Ouch.

While we aren’t in the exciting business of poring through 10Ks and 10Qs, we decided to take a quick look at our real-world US mobile app measurement data to see if there was anything in there that could have helped them predict this earlier (than largely rely on management representations and financial performance indicators). Spoiler Alert: The answer is Yes! We compared Grubhub app usage against their competitors Ubereats and Doordash. These are derived from a large longitudinal mobile consumer panel spread across the country.

Here’s what we observe (in less than 20 minutes of analysis):

  • Fewer people use Grubhub than Ubereats or DoorDash during any given week. While the number of panelists that used each of these Food delivery apps was about the same in Jan 2019, the spread has widened since. People appear to prefer DoorDash over UberEats and Grubhub.

Notes: The underlying measurement panel grew substantially from August 2019. So the uptick from week 27 across all three apps is to be expected. However, notice the widening gap between DoorDash and its competition starting week 31.  Notice the unexpected uptick in Doordash in week 40. It coincides with the Doordash’s announcement of a large scale security breach that affected 4.9m customers. Likely Doordash sent their users an email or an in-app notification asking them to change their passwords, which in turn led to increased usage.

num panelists opening food delivery app by week

  • Incidence rate of actively using food delivery app user (when normalized by the size of the panel) again indicates that Doordash remains steadfastly more popular

app incidence rate by week

  • Doordash and UberEats users appear to use the app a lot more frequently than do Grubhub users. Notably, the average number of Grubhub app sessions per week within the panel has dropped by a good 12.5% (from around 8 in Feb 2019 to less than 7 at end of Sept 2019). Competition’s app_usage rates are significantly higher and gaining over Grubhub. Since Q2, frequency of UberEats app usage is trending upward, while Grubhub’s is trending downward. Doordash has largely held flat.

average num food delivery app sessions per panelist per week

  • Average length of app_session for each of the food delivery apps is largely the same now. It was considerably higher for UberEats earlier in the year and has dropped off since. If I were an Analyst covering this sector, chances are that I’d be asking some difficult questions during the Quaterly Earnings call

avg food delivery foreground time in millis

In sum, there were some early signs in the behavioral data that could’ve helped the folks covering this sector detect GrubHub’s recent under-performance.

Would I be willing to bet my personal 401k off mobile behavioral data alone, probably not. Would I prefer engaging a financial advisor whose analysis factors real world behavioral data over ones that don’t? Absolutely.