How does a failure in a retailer’s mobile app impact shoppers’ purchases in its online (website and app) and offline (brick-and-mortar store) channels? Our main hypothesis is that an app failure has a negative effect on offline purchases and no effect on online purchases – after an app failure, shoppers who use the app primarily to search for information but make purchases mostly in the offline channels (search users) engage in search abandonment and forego or reduce their offline purchases; shoppers who use the app mainly to make purchases (checkout users) and online shoppers who use the app for search but purchase on the retailer’s website as well engage in channel switching, i.e., switch their intended action to the website as it is just a click away. An alternative hypothesis is that an app failure has no effect on purchases because shoppers engage in temporal substitution, i.e., simply delay their purchase and in channel switching for both search and purchase. It is challenging to empirically examine these hypotheses and estimate the causal effects of an app failure because field experiments inducing such failures are infeasible. We use a natural experiment, an exogenous two-hour failure in a large offline-dominant omnichannel retailer’s mobile app, to examine the effects of app failure on purchases in each channel. The results are consistent with our main hypothesis. An app failure has a significant overall negative effect on shoppers’ frequency, quantity, and monetary value of purchases across channels. This decrease in purchases after an app failure is driven by lower purchases in offline stores and not in the online channel, consistent with search abandonment and channel switching. Finally, an app failure has a smaller (larger) negative effect on the purchases of shoppers who spent more in the past (purchased more recently,searched for products on the app in the past) with the retailer.