PocketBill Blog
AI answers pull from sources that feel real, are directly useful, stay evergreen, and get judged neutrally. But the two rules that actually change what I write are information gain and typicality — a page that only repackages what's already public has nothing to cite. And since AI queries run 25 words against Google's 6, the win is the narrow question nobody answered. PocketBill has no support tickets to mine, so I started asking in forums instead.
Last week someone asked an AI assistant to recommend an expense tracker that works on a plane with no wifi and no account. It named three products. PocketBill was not one of them, and the reason was not that we are worse.
I read a breakdown of how overseas AI engines actually pick their sources, and four rules stuck. Real beats official. Useful beats popular. Evergreen beats topical. Neutral beats flattering — the cited reviews include the bad ones.
The first is information gain. If a page is a repackaging of what is already public, the model has no reason to show the source. It can summarise that itself. A page earns a citation by containing something that exists nowhere else.
The second is typicality — the sense that a real, specific operator stands behind the words. For a small app this is oddly good news. The only thing I have that a giant aggregator does not is the actual details of how ours behaves.
The numbers that reframed everything: a traditional Google query averages about six words. An AI query averages about twenty-five. People are no longer typing keywords; they are describing a situation.
That sounds harder, and it is the opposite. Those long queries are uncompetitive because nobody writes for them. "Which expense app keeps working when my phone has no signal and I refuse to make an account" has no commercial page fighting for it. It just needs one honest answer.
The usual advice is to dig long-tail questions out of your support inbox. PocketBill is local-first and account-free — there is no inbox. No tickets, no sales calls, no analytics on what people typed before they left.
So I went where the questions already were and read the forums instead. The unglamorous, oddly specific complaints people post about other apps. That is the page I am writing next, and it is the only kind of page that can speak for us at all.