| HN Mirror

But that could easily just be a selection bias effect. It’s marketed as a web app with a mobile interface, so it self-selects for people who already prefer that, and it would be a potential usage confounder, and would especially skew age demographics (as would questions about data privacy).

Also it’s not clear if this is just general page views or account activity that actually engages the functionality. Pageviews generally wouldn’t matter because it would be swamped out by a prior heavily favoring the model that people prefer to browse the web on mobile devices, unrelated to what they prefer for other products or application interfaces.

For example, I tend to browse job sites on either mobile or desktop indifferently. But I would absolutely only ever apply when browsing from a desktop, because application portals are too variable and require me to verify things and read more closely.

Hypothetically, if someone looked at traffic data naively and saw it was “50/50” between desktop and mobile for some job portal, they might wrongly conclude they need to change the interface to support mobile differently, when really the difference is between cursory traffic that has no intention to apply (uses mobile device) vs. someone intending to apply and thus would not use a mobile device to do it no matter what changes were made to the job portal (uses desktop with strict desktop-only preference for anything involving e.g. resumes, cover letters, etc.).

From a statistics point of view I suspect that baseline traffic numbers contain nearly zero information about customer preferences regarding app vs web form vs desktop spreadsheet when they finally sit down to actually enter the data and maintain it across time.

The other thing is that I’d suspect weeks, or even months, worth of data would be far too little to draw any type of significant conclusion from for this type of usage question.