crawlora-net/world-cup-sponsors-trends

World Cup sponsors vs. Google Trends search correlation

Annual Google Trends search-interest data (2004-2026, worldwide) for 12 official and regional FIFA World Cup sponsors, correlated against search interest in "FIFA World Cup" itself, tournament by tournament back to 2006. Snapshot 2026-07-21.

Companion to the study [Do World Cup Sponsors Get a Search Boost? We Correlated 12 Brands Against 22 Years of Trends Data](https://crawlora.net/blog/world-cup-sponsors-search-correlation?utm_source=huggingface&utm_medium=referral&utm_campaign=world-cup-sponsors-search-correlation).

Headline findings

• —The full-period correlation is mostly an artifact of the current tournament. Correlating each sponsor against "FIFA World Cup" search interest across the full 2004-2026 series produces r as high as 0.65 (Visa) — but excluding the still-in-progress 2026 tournament, every sponsor falls to r between -0.25 and 0.30. 2026 registers the single highest value in the whole series and acts as a statistical leverage point.
• —Mega-brands show almost no World Cup-year search lift. Coca-Cola, Visa, Adidas, McDonald's, and Hyundai-Kia — five of the longest-running FIFA partners — search only 1.08x-1.21x higher in a World Cup year than any other year. Coca-Cola's excl-2026 correlation rounds to 0.01.
• —The strongest signal belongs to narrower, more football-specific brands. Qatar Airways and Vivo both post r = 0.30 excluding 2026 — the highest in the set — with less non-sports search demand diluting the baseline.
• —Ended sponsorships show it in the data. Sony and Emirates, both of which quit FIFA sponsorship after the 2014 Brazil World Cup, show flat (Emirates, r = 0.05) or negative (Sony, r = -0.25) correlation for the decade since.
• —Wanda Group has zero measurable Google Trends interest across the entire 22-year window despite being a FIFA partner since 2016 covering the 2018 and 2022 World Cups — FIFA suspended Wanda's sponsorship rights in February 2026 over missed contract payments.

Files

Methodology

Pulled via Crawlora's Google Trends interest_over_time endpoint (google_trends_explore_interest_over_time), worldwide, time_range=all (2004 to the present — the widest window Trends exposes, which returns one point per year rather than weekly or monthly) on 2026-07-21. Keywords were batched five at a time: "FIFA World Cup" plus up to four sponsors per call, three calls total to cover all 12 sponsors.

Important normalization detail: Google Trends normalizes every keyword in a batch to a shared 0-100 scale (the single highest raw value across all keywords/points in that call becomes 100). Batch 2 (Budweiser, Hyundai, Emirates, Sony) happens to contain Sony's 2004 value as the batch-wide max, so "FIFA World Cup" reads slightly lower in that batch (78 in 2026) than in batches 1 and 3 (100 in 2026, both anchored by FIFA World Cup's own peak). This does not affect any correlation in sponsor_correlation_summary.* — each sponsor's r is computed only against the "FIFA World Cup" column from its own batch, and Pearson correlation is invariant to a shared linear rescaling of both series — but it does mean the raw "FIFA World Cup" values in annual_search_interest_raw.* differ slightly by batch. Always join a sponsor to the FIFA World Cup row with the matching batch value if reproducing the correlation yourself.

Sponsorship windows (which years each brand was a confirmed official FIFA World Cup sponsor) were compiled from public reporting — see the companion blog post's Sources section for citations (SportBusiness, Fortune, Gulf News). Correlation is the Pearson coefficient (r), computed twice per sponsor: over the full 2004-2026 series, and again with 2026 excluded since that tournament was still in progress at pull time.

Caveats

• —Yearly granularity is a real limitation. Google Trends' all time range returns one point per calendar year for a window this wide, not weekly or monthly — a tournament that runs June-July shows up folded into its entire calendar year's average, and 12 months is coarse relative to a ~4-week event.
• —This is Google's relative search index, not raw volume, sales, or revenue data. A weak correlation with search interest is not proof a sponsorship failed commercially — see the companion post's discussion of Coca-Cola and Visa, whose deals are plausibly valued for broadcast association rather than search demand generation.
• —Wanda Group's flat-zero result is a Google Trends visibility gap, not necessarily zero real-world interest — it likely reflects the brand's search footprint concentrating on Baidu and other China-first platforms outside Trends' worldwide English-language index.
• —Sponsorship windows are summarized from public reporting, not a FIFA primary-source contract database — treat start/end years as best-available, not authoritative to the month.

License & citation

Released under [CC BY 4.0](LICENSE) — use it for anything, just attribute.

Crawlora (2026). World Cup sponsors vs. Google Trends search correlation. https://github.com/Crawlora-org/world-cup-sponsors-trends-data (CC BY 4.0)

See `CITATION.cff` for machine-readable citation metadata.

Extension: full 2026 sponsor tier list + the Lenovo-Vivo swap (added 2026-07-21)

Companion to the follow-up study [We Widened the World Cup Sponsor Study to All 23 2026 Sponsors — Still No Search Boost](https://crawlora.net/blog/world-cup-2026-sponsor-tiers-search-correlation?utm_source=huggingface&utm_medium=referral&utm_campaign=world-cup-sponsors-search-correlation).

FIFA's 2026 World Cup has 23 confirmed sponsors across three tiers (plus a separate regional-supporter tier); the original study above tested 11 of them. This extension adds Google Trends data for the remaining 11 — Aramco, Lenovo, Bank of America, Frito-Lay, Mengniu Dairy, Unilever, Verizon, Home Depot, Diageo, Rock-it Cargo, and Valvoline — plus a full weekly (not yearly) Google Trends series for Lenovo and Vivo, since Lenovo replaced Vivo as FIFA's official smartphone/technology partner ahead of the 2026 tournament.

Headline findings

• —None of the 11 additional sponsors show a credible World Cup-linked search signal, extending the original study's core finding to the full 23-sponsor 2026 roster.
• —Seven of the eleven are essentially invisible on Google's shared 0-100 Trends scale (peak value of 2 or less): Frito-Lay, Mengniu Dairy, Diageo, and Rock-it Cargo return a flat 0 in every year; Aramco, Unilever, and Valvoline barely clear it.
• —The four sponsors with real search volume — Lenovo, Bank of America, Verizon, Home Depot — fit the original study's "mega-brand" pattern: correlation near zero or negative (r between -0.31 and 0.24 excluding 2026), spike ratio 0.93x-1.09x.
• —Weekly data on the Lenovo-Vivo swap tells a more specific story. Vivo's search interest spiked cleanly to its batch-max the exact week the 2022 World Cup opened (Nov 20-26, 2022). Lenovo shows no equivalent spike in 2026 — its biggest rise came in March-May 2026, before the tournament, and it cooled during the event itself, while Vivo (no longer the sponsor) rose during the 2026 tournament weeks to its highest level in over three years.
• —Combined with the original 12 sponsors, 21 of 23 tested 2026-cycle sponsors show no meaningful World Cup search signal.

New files

Methodology (extension-specific)

Same approach as the original pull, described below, with one addition: the Lenovo-Vivo comparison used time_range=today 5-y (not all) with only those two keywords (no "FIFA World Cup" anchor), which returns weekly points rather than yearly — fine enough granularity to see a spike tied to a specific tournament week rather than folded into a whole calendar year average. The 11-sponsor correlation pull otherwise follows the original methodology exactly (time_range=all, worldwide, batched 4-5 keywords per call with a shared "FIFA World Cup" anchor per batch).

*On the "n/a" correlations in `2026_tier_extension_correlation_summary.:** four sponsors (Frito-Lay, Mengniu Dairy, Diageo, Rock-it Cargo) return a flat 0 in every single year — Pearson correlation is mathematically undefined for a series with zero variance, so rfull20042026` / `rexcl2026` / `spikeratio are null. Three more (Aramco, Unilever, Valvoline) technically have a nonzero Pearson r, but with peakvalue` of 1-2 (out of a shared 0-100 scale where "FIFA World Cup" peaks at 100), that number is a statistical artifact of one or two years crossing an integer-rounding threshold, not a meaningful correlation — treat any correlation on a sponsor with `peakvalue <= 2` as not statistically informative.

Additional caveat

• —Sponsor tiers were compiled from FIFA's sponsor list on Wikipedia and cross-checked against individual FIFA sponsorship-announcement press releases — treat tier assignment as best-available public reporting, not a FIFA primary-source contract database.