DIZKO
Dizko · Articles · Methodology
Methodology

How Dizko measures event coverage

Dizko editorial · Published

Dizko measures event coverage from a rolling index of public listings. This page documents how records enter that index, how overlapping listings are consolidated, which fields count as observed, and where the data remains incomplete.

Reporting window and collection

City reports use a rolling 45-day window beginning on the report date. Dizko collects public listings from ticketing platforms, venue calendars, scene publications, cultural institutions, and community event tools. The collection target is one refresh every six hours. A report records its own generation time and the latest inventory refresh time.

How overlapping records become one event

A source record is an observed listing. A canonical event is the retained event shown in the index. Dizko groups likely matches by UTC date and applies a shared collision rule using geographic distance within 300 metres, title overlap, and source priority. When records collide, the index retains one canonical event and can preserve evidence from more than one source.

The process reduces duplicate listings, but it is not a claim of perfect identity resolution. Venue aliases, schedule changes, missing coordinates, and differently written titles can affect the result.

Observed fields and derived measures

Field or measureHow it is treated
Title, date, venue, source linkObserved from a public listing
Coordinates, genre, artist, promoter, priceCounted only when present on the retained record
Source shareRetained events attached to a source divided by all retained events
Duplicate rateConsolidated source records divided by all observed source records
Coverage percentageRetained events with the named field divided by all retained events

Freshness and comparison rules

A city receives live status when the latest refresh is within 12 hours, at least three sources are active, and source-link coverage is at least 80 percent. A historical comparison is reported only when the earlier snapshot is at least seven days old and at least 80 percent of active sources overlap. These thresholds prevent a partial scrape from appearing as a market change.

Known limitations

Reproduce the Berlin example

The August 17, 2026 Berlin report contains 2,630 retained events from 2,745 source records across 28 active sources. Read the interpretation of that snapshot, query the public report endpoint, or download the source rows as CSV.

Method and sources

This methodology describes Dizko's production city-signal report as of August 17, 2026. Dizko operates the collection and consolidation system. For a citation, include the city, snapshot date, reporting window, and link to either the public endpoint or the dated CSV.

Frequently asked questions

How often does Dizko refresh event data?

The collection target is every six hours. Each city report exposes the actual latest refresh time, so a reader can distinguish the target from the observed freshness.

Does an active source mean every listing from that source is present?

No. Active means the source contributed retained records to the snapshot. It does not certify complete recall against the source's own inventory.

Why publish a dated CSV as well as a live endpoint?

The endpoint shows the current report and can change as the window moves. The dated CSV preserves the source distribution used by a specific article so its claims remain reproducible.