Methodology.

1. Abstract

The Other League Table is a full-population study of the public social media performance of the 116 clubs of English football’s top five tiers: the Premier League, the Championship, League One, League Two and the National League. Using public data only, the study measures output, video share, platform presence, follower-normalised engagement and post-match publishing speed across Facebook, Instagram, TikTok, X and YouTube during the closing nine weeks of the 2025-26 season, and reports audience size relative to attendance as a separate reference index. The dataset comprises 193,332 public posts (190,444 in the study window and 2,888 on winter deadline day) carrying 977,683,366 public interactions, with 1,241 league fixtures timed for the speed measure. Clubs are ranked within the division they played in during 2025-26 on a weighted composite locked before collection, and across all 116 on the same five measures rescaled. Every EFL club receives its own results directly, free of charge; the rankings are editorial and final.

2. Governance

Two rules sit above everything else in this document. First, the lock: the measures and weights were locked on 30 July 2026, before any data was collected, and the full counting rules on this page were published before any score was computed. No table was, or could be, tuned after the scores were known; section 8 shows what happens to the results when the locked specification is varied anyway. Second, the ranking is editorial and final: no commercial conversation with any club moves a position in either direction.

3. Population and scope

The study population is the complete membership of the top five tiers of English football in 2026-27, 116 clubs in total: Premier League (20), Championship (24), League One (24), League Two (24) and National League (24). There is no sampling; every club is measured.

The English league system includes four Welsh clubs, Cardiff City, Swansea City, Wrexham and Newport County, all in scope. Scottish and Irish clubs compete in separate national league systems and are not part of this population: the study covers English football, which is not the same thing as British football.

In scope are each club’s official first-team accounts on Facebook, Instagram, TikTok, X and YouTube. Women’s, academy, foundation and player accounts are out of scope, stated here explicitly because “you missed our women’s account” is the predictable dispute: the study measures the first-team channel a club presents as its main public voice.

4. Study window and data

The measurement window is retrospective: Saturday 1 March to Sunday 3 May 2026, the closing nine weeks (exactly 63 days) of the 2025-26 season. The window contains a full league programme for every division, and all engagement had settled months before collection. Clubs are scored against the division they played in during 2025-26, and the published tables present those same divisions, with the four clubs promoted from the National League’s feeder divisions flagged (section 6 sets out how they are handled).

The run-in is chosen deliberately. It is the season at full stretch: title races, promotion pushes, play-off chases and relegation scraps are live in every division, media teams are at peak output under peak pressure, and fan attention rises with the stakes, so the engagement being measured means the most. A ranking of club social operations should measure clubs when it matters, not in a quiet spell. The window is also structurally fair: every club plays a comparable league programme inside it, and it closes at the one full stop the whole league shares, final-day weekend. The play-offs are excluded on the same principle, since a dozen clubs would gain weeks of high-drama content the other hundred cannot access. Nine weeks is long enough that no single viral post or quiet week moves a club's numbers, and recent enough that the period described is the football everyone has just watched. Deadline day, Monday 2 February 2026, is the one out-of-window capture (measure 5.7): a different kind of pressure test, reported as a standalone section and never weighted into the composite.

All data is public: posts, timestamps, interaction counts, follower counts and fixture kickoff times. No API permissions, no insider numbers, no paid-reach data. Collection ran in July and August 2026 and gathered 190,444 in-window posts plus 2,888 deadline-day posts, 193,332 in all, carrying 977,683,366 public interactions; follower counts are as of collection.

5. Measures and formulas

Every formula below is the exact rule the scoring pipeline runs; the published counting rules and the computed tables cannot drift apart because the former is generated from the latter’s source.

6. Scoring and weights

The five division tables are the primary rankings: clubs are compared with their own 2025-26 division, where resources are broadly comparable. Within each division, each scored measure is scaled 0 to 100:

scaled = (value - division minimum) ÷ (division maximum - division minimum) × 100

Matchday speed is inverted after scaling (100 - scaled), so that faster is higher, and if every club in a division shares the same value on a measure, all score 50 on it. The scaled measures combine into the composite with the following weights, fixed before collection:

composite = 0.30 × engagement + 0.20 × output + 0.20 × speed + 0.15 × video + 0.15 × presence

• Engagement: 30%
• Output: 20%
• Matchday speed: 20%
• Video share: 15%
• Platform presence: 15%

Unavailable measures reweight; nothing is imputed. Where a measure cannot be computed for a club (no qualifying fixtures for matchday speed; no in-window posts or no recorded followers for engagement), the composite is the sum of the available weighted scores divided by the sum of the available weights. For the four clubs promoted from the National League’s feeder divisions, which have no 2025-26 league fixtures in the study’s sources, the composite therefore reweights over the remaining four measures (divided by 0.80). Those four clubs are scored within the National League cohort, because a two-club cohort cannot be min-max scaled, and are flagged in the published table.

The overall table ranks all 116 clubs on the same five measures under the same weights and the same reweighting rules, min-max rescaled across the full population instead of within a division: a deliberately size-blind view of performance in which a well-run League Two club can rank above a Premier League club. The division tables remain the primary presentation.

Deadline day is reported but never weighted. The global reach multiplier (measure 5.5) is a reference index only: it appears in its own table and carries no weight in the composite, the division tables or the overall table, because it measures brand gravity rather than performance.

7. Worked example

Casterbridge Rovers are a fictional club in a fictional four-club division, invented so the full calculation can be followed by hand without singling out any real club. Their raw measures for a nine-week window: 126 posts, of which 63 are video, posted across four of the five platforms; 25,200 public interactions; 40,000 combined followers; and a mean of 24.0 minutes from estimated full-time to first post. The division’s minimum and maximum on each measure (set by the other three clubs) are shown alongside.

composite = 15.0 + 10.0 + 15.0 + 7.5 + 7.5 = 55.0

Note the speed row: 24.0 minutes scales to 25.0 on the raw scale, and the inversion (100 - 25.0) turns it into 75.0, because faster must score higher. If Casterbridge Rovers had no qualifying fixtures, the speed term would drop out entirely and the composite would reweight over the remaining measures: (15.0 + 10.0 + 7.5 + 7.5) ÷ 0.80 = 50.0. Nothing is imputed.

8. Sensitivity analysis

With five defensible measures there is no single “true” table: a different analyst could weight the same data differently and produce a different order. The study’s defence is the lock in section 2, and this section shows, rather than asserts, how much the results depend on it. All 116 clubs were re-scored under ten alternative specifications: three realistic re-weightings that keep all five measures, and seven structural changes that remove a measure entirely or change the scaling scheme.

Under the realistic re-weightings, the results hold. Equal weights, engagement at 25% and engagement at 35% retain 16, 18 and 19 of the overall top 20 respectively. A reader who disagrees with the exact weights is disagreeing about the order within a stable band, not about who belongs in it.

Under structural changes, the results move, and they should. Removing a measure entirely retains between 10 and 19 of the top 20 (removing output is the most disruptive), and switching min-max scaling for rank percentiles retains 13. A composite that did not respond to removing a fifth of its inputs would mean the measures were redundant. Division winners are decided by finer margins than the top-20 band: dropping video or presence preserves all five winners and engagement at 35% preserves four, but equal weights preserves two and dropping engagement entirely preserves one. This is exactly why the specification was locked before a single post was collected: the numbers below are what stops that lock being a matter of trust.

Re-weighted specifications scale the untouched weights pro rata so the total remains 1.00. The 5th-95th percentile cap winsorises each measure within its cohort before min-max scaling, testing whether any single outlier club compresses the field; it does not.

9. Exclusions and limitations

Deliberately not measured

• Paid advertising: not visible in public data, and not the organic craft this study ranks.
• Stories and other ephemeral formats: not retrospectively measurable.
• Reach and impressions: private platform data; follower-normalised engagement is the public proxy.
• Websites, apps, newsletters and podcasts: outside the study’s definition of social output.
• Tone or quality judgements in the scoring: the table is arithmetic. Qualitative observations appear only in per-club scorecards and are clearly marked as opinion.

Known limitations

• Public interaction counts undercount platforms that hide share totals; the undercount applies to every club equally.
• The full-time estimate introduces up to ten minutes of noise per match; averaged across a run-in of league matchdays it does not move a ranking band.
• Attendance normalisation in the reach multiplier slightly favours clubs with capacity-constrained grounds; the study names this rather than adjusting for it.
• Follower counts are read at collection (July and August 2026) while posts date from spring 2026; per-1,000-follower engagement therefore slightly understates the fastest-growing clubs. The rule is identical for all 116.
• A small number of accounts cannot be read publicly at all (one club’s Facebook page is unavailable to logged-out visitors). These are recorded as unmeasurable and noted, never scored as zero.
• Four clubs promoted from the National League’s regional feeder divisions have no 2025-26 fixture data in the study’s sources; their matchday-speed measure is recorded as unavailable and handled by reweighting rather than imputed.

10. Disclosures

• Funding: the study was funded and conducted by NBK Social Research, the research imprint of NBK Social.
• Competing interests: NBK Social provides social media services to sport organisations. Commercial relationships, past, present or prospective, had no influence on inclusion, measurement, scoring or publication.
• Review status: this study has not been academically peer reviewed.

11. Versioning, data availability and citation

The study published on its permanent address on 1 August 2026. Any post-publication amendment is dated in the changelog below.

• 1 August 2026: the platform-gaps summary on the study page was corrected on publication day. The clubs without TikTok accounts had also been described as lacking YouTube channels; their channels exist, are verified and were measured. The corrected summary instead counts two other clubs whose existing YouTube channels went the window without an upload. No score, rank or table was affected.

Summary data behind the published tables is available to journalists and researchers on request. Contact NBK Social Research.

Frequently asked questions