dayparting-and-geolisted

# Dayparting and geo On a small budget these two settings move the number more than most creative changes, because they are the difference between spending on demand that converts and demand that never does. They are also where thin data most easily produces a confident wrong answer, so the refusal matters as much as the analysis. ## Aggregate before you look A small account has almost no data in any single hour of any single day. Looking at a 24 x 7 grid guarantees you find something and guarantees it is noise. Aggregate until each bucket holds a usable number of conversions: 1. **Weekday against weekend** first — two buckets, the most likely real effect 2. **Broad parts of the day** next — roughly morning, working hours, evening, overnight 3. Only then finer detail, and usually you will not get there The same applies to geography: country or state before city, city before postcode. ## What a real pattern looks like A genuine effect is **large, stable and explicable**. - **Large** — cost per conversion several times higher, not a few percent - **Stable** — the same direction across separate months, not one bad weekend - **Explicable** — there is a reason. Nobody answers the phone on Sunday. Overnight traffic is a different intent. A region is outside the service area A pattern that is large and stable but has no explanation is worth investigating before acting on. Often it turns out to be one campaign, one device, or a tracking gap rather than time or place.