The same site looks like three different sites depending on the dates you select. Yesterday is whoever just linked you. A week is a rhythm. A month is an audience. If you change the range after you have already decided what the chart "means," you are shopping for a number.
Pick the range from the decision, then open the report.
Match the range to the question
| Question | Range | Why |
|---|---|---|
| Did the widget just start recording? | Today, or the last hour if you have it | You only need your own test visit |
| Did a post or a newsletter land? | The day of the send, plus the next two days | Most of the click arrives immediately, then a tail |
| Which countries actually read this site? | 28 days, compared with the 28 before | A week is still one news cycle |
| Is a page worth translating? | 28 days or a quarter, and only that page | One country on one afternoon is not a language |
MapMyVisitors lets you set a date range on the stats page. The map widget on the public site is a picture of recent visits, not of the range you picked in the dashboard. Do not describe the widget as "this month." Describe the table, and name the dates.
Compare equal lengths
"Up from last week" only means something if both weeks are seven days and neither one contains a holiday, a launch, or a bot spike you already understand. Put the two ranges next to each other before you write the sentence. If the earlier range includes a day the site was down, the growth is not growth.
A partial current month against a full last month is the most common bad comparison. Either wait until the month closes, or compare the first N days of each month.
When a short range is the right one
Checking an install, a referrer after you publish, or whether a deploy stopped visits: use a day. Then stop. Do not take that day's country list into a planning meeting. The people who arrived because of the post are real, and they are not yet "the audience."
Bots love short windows because a scan is a spike. If a single hour dominates the day, check the user agent before you tell anyone traffic doubled. Telling bots from visitors is a five-minute pass and it belongs in every short-range reading.
Long ranges hide launches. If you only ever look at 90 days, a successful post is a small bump and you will not see which link caused it. Keep one short look and one long look. They are not substitutes.
