Methodology

Source data

The directory is seeded from public reference sources at build time. The primary intent is to use the Wikidata SPARQL endpoint to retrieve agencies typed as advertising or marketing companies, including names, countries, founding years, and known specializations. Where the primary source returns insufficient coverage, the build falls back to a curated set of well-known IMC agencies compiled from publicly available holding-company registers, trade-press references, and industry directory listings.

The actual source used in the most recent build of this site is recorded below. The build is reproducible: rerunning the seed script with the same configuration regenerates the same profiles. Inclusion is therefore an editorial output of the source-selection rules described here rather than the result of any ad-hoc curation.

• Source used in this build: Wikidata SPARQL endpoint (https://query.wikidata.org/sparql)
• Build timestamp: 2026-05-03T03:25:04+00:00
• Total agencies in this build: 506

Profile structure

Each agency profile has a consistent shape: name, country of headquarters, founding year (where recorded), one or more specialization labels, and a written overview composed of six standardised sections — overview, practice and specializations, geography, era and lineage, position in the wider IMC landscape, and how to use the profile. The standardised structure makes profiles directly comparable to one another and makes it obvious when a particular field is missing rather than absent by accident.

The written sections are generated from the underlying data using a fixed set of templates. They are intentionally descriptive rather than evaluative; nothing in a profile constitutes a rating, ranking, or recommendation. The profile does not invent claims about clients, awards, or leadership that are not supported by the source data.

Categorisation choices

The country field reflects the agency's headquarters as recorded by the source, not the full footprint of its offices. Many agencies operate across multiple jurisdictions; the directory does not attempt to catalogue every office, because the headquarters location is the more stable and more comparable field across thousands of entries. Country pages should be read as "agencies with their main address in this market" rather than "agencies that work in this market".

The founding year is the earliest credibly recorded founding date in the source. Agencies frequently have several defensible founding dates depending on whether you count the first office, the first registered legal entity, or the most recent material rebrand. Where the source is ambiguous, the directory takes the earliest date and notes the ambiguity in the era section of the profile.

Specialization labels are heuristic. They reflect the craft an agency is most strongly associated with in public reference material. Most agencies in the directory operate across a wider range than any single label suggests, and the labels should be read as a leading edge rather than as a fixed boundary.

Update cadence

The seed script is intended to be re-run periodically against the same source. When it is re-run, profiles can change — agencies can be added, retired, recategorised, or have their founding year corrected — to reflect updates in the underlying source. The build timestamp above reflects the last time the data was refreshed for this site. The directory does not maintain a per-profile change log; the source itself is the authoritative trail.

Things explicitly out of scope

The directory does not catalogue individual campaigns, individual creative assets, or specific client relationships. Those exist in trade-press archives, awards databases, and case-study libraries that are better suited to the granularity. The directory also does not catalogue freelance practitioners, internal in-house marketing teams, or pure technology vendors — the unit of analysis is the agency, defined as an organisation hired by a client to deliver IMC work.