Worked example: market research

Most market sizing is a snapshot with a growth rate bolted on. With 17 years of replayable register history you can run the identical query at each year end and measure the change instead.

Write the segment once

Sector, legal form, geography and vintage are all register facts, so the definition is stable across time. That is the property that makes the replay meaningful.

$ for y in $(seq 2009 2026); do
    curl -s "https://api.spotit.ai/v1/search?country=DE&nace=52.29&legal_form=GmbH&count_only=true&as_of=${y}-12-31" \
      -H "Authorization: Bearer $SPOTIT_KEY"
  done

Three questions the register answers directly

• Formation and dissolution rates by commune. entity.status_changed and incorporation dates, rolled up through NUTS/LAU. Municipal granularity, not national.
• Foreign-branch openings by parent jurisdiction. Branch entities carry their parent's seat, so you can see which countries are expanding into a market and when they started.
• Sector migration. NACE changes are filings. A company reclassifying itself is a signal, and it is dated.

Cite the gazette

Every count decomposes to entities, and every entity decomposes to filings. GET /v1/entities/{id}/lineage returns the document behind each fact, which is what turns a chart into something a reader can check.

For anything larger than a few thousand rows, take the warehouse share instead and run the whole study in SQL.

Caveats worth printing next to the chart

• Financial coverage is 61% in Tier 1 and 9% in Tier 2. Revenue-weighted analysis is only honest where accounts are filed.
• Registers differ in what they consider a dissolution. Cross-country comparisons should use the status taxonomy we normalise to, and should say so.
• A register that digitised its back catalogue in 2013 will show an artificial step in that year. The coverage table marks these.