How wrong can you afford to be?

Paul Neto
Image referencing the blog page title

Our spending behaviors have become fragmented. Once upon a time the financial inputs and outputs were simple. You were paid in cash or cheque, and you paid for things in cash or cheque. Then you had a credit card, or two, or three, then a debit card, then a tap and go, then you paid with your phone, and now you pay with your watch. Furthermore, you don’t just buy groceries, you subscribe for recurring items, purchase on Amazon, maybe Costco, and your Netflix comes from your internet provider, and your apps, iTunes purchases, Office, Apple storage, and Apple music are billed directly as subscriptions. 

Unfortunately, most payment and spend research is still done by recall or submission of messy receipts. This fragmentation makes these methods highly unreliable. 

We strove to understand how people perceive some of their purchases, and in particular, in-app and subscription purchases made through their iPhones. In this study, we asked nearly 600 individuals to estimate what they spent on via in-app purchases and subscriptions in the previous month. We then asked individuals to share their in-app purchases with us using our Retro behavioral data collection methodology. Since this data is actual data, there is 100% accuracy as it does not rely on recall, collection of paper or digital receipts.


What we found was not surprising. 

The analysis showed that 41% of respondents were incorrect by at least 75%, and 70% of respondents were wrong by at 25%. We did find that 62% of respondents over-reported how much they spent, and 38% under-reported how much they spent. Thus on average, the estimation error was 237%. That’s a 2x error.

This obviously creates a lot of questions about the accuracy of reported purchases, particularly for this type of purchase data. If you’re asking about purchase data in your survey, you need to check out Retro - now. These complex areas have a simple solution; get the data, unbiased, directly from the source. Ultimately, if you are using a survey to get to this data - you shouldn’t be.

Check out our report to learn more.