Traffic report

Notes

What each figure above cannot tell you, stated here rather than left out. The counts in these notes are read from the same data block as the charts; the dates and the row caps are stated by hand.

1. The two totals. The 6,126 figure is GoatCounter’s own total for the window, and the browser and system bars sum to it exactly, while the location bars (67 short) come from a capped list; the daily columns sum to 6,160, 34 apart. Both count the same period, and the difference has run in both directions and is not a fixed offset, so it is not that one of them is short. This page does not know what the difference is, so it prints both rather than deriving one from the other. Counts are GoatCounter “visits”: cookieless, identity-free.
2. The window, and the root path. This is a rolling two months, August 11 – October 11, 2026, 62 days ending on the last full day. It used to run from launch on 10 July and grow without bound; tracking began that day, so nothing before it exists, and everything before this window does and is simply not drawn here. Until August 24 the Illinois app was served from / at chidistricts.com; that day it moved under /il/ and / became the fleet landing page. The series spans both, so / counts as Illinois on every day before the cutover and as the landing page on the cutover day and after, a whole day at a time, which is the granularity the dashboard publishes per path. August 25 is not a slump. Google Safe Browsing flagged the day-old districtry.com domain that morning, and Chrome shows its warning before the page loads, so blocked visits never reached the counter at all. The flag was a false positive and has since been cleared; the day is under-counted by an unknown amount and cannot be reconstructed. The two Illinois SEO pages the narrative tracks are at 28 visits (/il/police-district.html) and 255 (/il/school-board.html, the third most-visited path on the site).
3. The instance split. Twelve bars: eleven apps and the fleet landing page. Read the ranking as age, not quality: until August 24 the non-Illinois tags pointed at GoatCounter sites that did not exist and every one of their visits was dropped, unrecoverable. Illinois has the whole window of history; each of the others starts at the cutover or at its own later go-live, which is why the newest apps have the fewest days to have been counted at all. Iowa joined the fleet manifest at 01:24 UTC on 28 August, so it is listed on the landing page now; until then it was live and in the sitemap but unlisted, which is how it drew a visit from a page that did not offer it. These twelve sum to 2,123 of 2,568 page visits. The split reads the dashboard’s own pages list, which caps at ten rows, so 445 visits on the long tail are in the total and in no bar. Bars reading zero (San Francisco, Minnesota, Kentucky, Indiana, North Carolina and Oklahoma) mean those paths are not among those ten rows, not that the counter recorded none: their visits sit in that long tail, and this page cannot tell how many, zero included. Some of those bars also had less of the window to count: San Francisco (49 days), Minnesota (12 days), Kentucky (12 days), Indiana (5 days), North Carolina (3 days) and Oklahoma (2 days) spent only that long on the front door, which is where a reader finds an app at all. The root path is divided by date rather than guessed at: each day’s root visits go to whichever page was actually there.
4. Layer ids share one namespace across the fleet, so a layer more than one app registers reports into ONE row and its count sums across them. Five of the ten drawn here are registered by more than one app, and a toggle has carried its own app's tag since September 29; an unattributed figure is toggles from before that date, or ones whose per-app row fell below the ten each app's list holds: Congress (Illinois 35, New York 9, San Francisco 1 and 73 unattributed), County (Illinois 33, Iowa 4, Michigan 8, Minnesota 3, Oklahoma 1, Wisconsin 1 and 51 unattributed), County board (Illinois 24, Wisconsin 4 and 76 unattributed), Ward (Illinois 65 and 123 unattributed) and us-house (Indiana 4, Iowa 5, Kentucky 2, Michigan 7 and Wisconsin 3). The other five are registered by one app and are unambiguous: County precinct, IL House, IL Senate, School board and Ward precinct. How much blending there could be is measurable even though the blending itself is not: the other ten instances have 345 page visits between them against Illinois’s 1,342, so a shared id could carry at most that share and the single-app ids carry none of it. That ceiling grows as the other instances do, and this note is where it will be visible when it stops being small. The layer list is the top 10 of those recorded.
5. The referrer list is a top six. GoatCounter’s dashboard shows a “Show more” beyond it that this page cannot query into. The six sum to 2,924 of 6,126 visits, so a named search engine’s true total could run higher than what is shown here. utm_source tags count in the referrer list; the campaigns list counts only visits from links tagged with ?utm_campaign=….
6. Search counts combine 2 Search Console properties, because the site renamed mid-window and the old domain still appears in results and redirects across; those are two different results for one search, not one counted twice. Google withholds rare queries before anyone sees them, so these will not reconcile exactly with Search Console’s own reports. This is also the one thing the visit counter cannot see: Google strips the query from the referrer, so every one of these searches arrives in the referrer list as an anonymous “Google”. On Bing the figures combine 2 verified properties: districtry.com first appeared in these results on September 13, with 6,682 impressions against chidistricts.com’s 756. districtry.com carries most of the figure now; chidistricts.com’s share is the early days of this window, which the rolling window is dropping.