Price tracking sounds trivial. Store a route, check the fare periodically, send a message when it drops. The mechanical part genuinely is that simple. The part that determines whether it’s useful is different: deciding which changes deserve to interrupt you.
The easy half
A tracker records the route and the dates you care about and re-queries the fare on a schedule. That’s it. Anyone can build this half.
The half that matters
Fares move constantly, often by small amounts and often back again within a day. A tracker that alerts on every movement becomes noise within a week, and noise gets muted. Once muted, it may as well not exist.
So the real work is judgement:
- Is this movement meaningful, or normal jitter? A $6 change on a $700 fare is not news.
- Is this price actually low, or just lower than yesterday? Those are different claims. A fare that rose 20% last week and fell 5% today is still expensive.
- Does the change survive the total cost? A drop in base fare that coincides with a change in fare brand — fewer bags included, stricter change rules — may not be a drop at all.
Answering the middle question requires history. You cannot tell whether today’s number is genuinely low without knowing how that specific route has behaved.
Why alerts get ignored
Almost every muted notification stream started useful. It degrades when the ratio of “worth reading” to “not worth reading” tips. Tying alerts to real movement rather than to a fixed schedule is what keeps the ratio survivable — and attaching the reasoning means you can judge the alert rather than trust it blindly.
You can read each stage of how BookWithAI tracks and alerts, including where it deliberately stops.
The honest limit
Tracking tells you what a price is doing. It does not tell you what a price will do. Where a route has thin history, any judgement about it rests on less evidence — and that’s a property of the data, not something a confident interface can fix.