GX Index August 2026: What UK Gigs Paid (£330–£1,489)Issue 05 of the GigXchange Index: 4,431 data points, a city league flattened to £22 by a model change we made on purpose, 407 published cells, and 160 seed rows withdrawn in public
TL;DR: GX Index Issue 05, August 2026
The GIGXCHANGE Index Issue 05 is the one where the city league stopped being a league. Thirteen UK cities now sit inside a £22 spread on wedding fees, from Edinburgh at £1,067 to Manchester at £1,045, where July showed £523 between top and bottom. The market did not do that. We did, by letting national evidence count towards every city instead of none of them. The pool grew by 141 to 4,431 observations, published cells jumped to 407 because every city now answers the same grid, and the UK wedding median for a 3–4 piece eased £52 to £990. Pubs slipped to £350. And on 8 August we withdrew 160 seed rows that had been sitting in our highest-trust source since launch.
This write-up explains the model change in plain English, prints the tables, scores July's promises, and says where the sample is still too thin to quote. Live percentiles at /rates; the full issue is at the August report page.
- 4,431Approved observations in the pool, up 141 on July’s 4,290
- £990UK median wedding fee, 3–4 piece band, £52 under July
- £22Spread across all 13 cities in the wedding league, down from £523
- 407Published rate cells, up from 280 on the new uniform grid
- £350UK median pub fee, 3–4 piece, the third fall in a row
- 160Seed rows withdrawn from the verified-report source on 8 August
All figures are artist take-home (net of commission). Built from the GX Index on the 3 September 2026 05:00 UTC refresh, six live weighted sources. A cell’s n counts weighted samples applicable to it, not unique gigs, and from this issue a city n includes evidence that applies nationally. Full report: Issue 05. Previous issue: July 2026.
1. The league that flattened, and why we did it on purpose
If you opened the August city table expecting the usual order, you will think something broke. Edinburgh top, Manchester bottom, and twenty-two pounds between them. For four issues I ranked British cities against each other on wedding fees, and for four issues that ranking rested on something I am not proud of: a national price list counted towards the country and towards no city inside it.
That meant every city was being built from a thinner, cheaper pool than the nation it sat in. A union minimum that applies from Truro to Thurso was evidence for the UK figure and evidence for nowhere in particular. Once you see it, you cannot leave it alone. So we changed the model, and the league collapsed. That is the number we can defend, and I would rather print it than dress up the old one.
Thirteen cities inside £22
Here is the full wedding league for a 3–4 piece band, ranked by median. Read the n column with care: it now counts every observation applicable to the city, national rows as well as local ones, which is why it reads over a thousand everywhere.
| City | p25 | p50 (median) | p75 | p90 | n |
|---|---|---|---|---|---|
| Edinburgh | £750 | £1,067 | £1,480 | £1,997 | 1,062 |
| Brighton | £750 | £1,065 | £1,467 | £1,975 | 1,093 |
| Cardiff | £750 | £1,065 | £1,472 | £1,995 | 1,074 |
| Glasgow | £750 | £1,065 | £1,480 | £1,995 | 1,070 |
| Leeds | £750 | £1,065 | £1,469 | £1,995 | 1,084 |
| Newcastle | £750 | £1,065 | £1,475 | £1,995 | 1,060 |
| Nottingham | £750 | £1,065 | £1,469 | £1,995 | 1,068 |
| Sheffield | £750 | £1,065 | £1,472 | £1,995 | 1,094 |
| Bristol | £750 | £1,064 | £1,465 | £1,975 | 1,116 |
| Birmingham | £750 | £1,057 | £1,465 | £1,975 | 1,112 |
| Liverpool | £750 | £1,057 | £1,465 | £1,975 | 1,095 |
| London | £745 | £1,053 | £1,447 | £1,955 | 1,148 |
| Manchester | £750 | £1,045 | £1,447 | £1,975 | 1,136 |
City league · Wedding fees, 3–4 piece band · Ranked by p50 · Source: GX Index Issue 05, 3 September 2026 refresh.
Seven cities print the same £1,065. London, which in July was the widest market we published and sat fourth, now lands at £1,053 with the lowest p25 in the table at £745. None of that is a story about London. It is what happens when 1,052 shared rows sit behind every cell and the local evidence on top of them is thin: London has 96 local rows, Newcastle has 8, and neither is enough to pull its median far from the national centre.
If you are a function band pricing a wedding this autumn, the honest reading is that the country-wide 3–4 piece median of £990 is your anchor, and your city adds or subtracts less than the price of a set of strings. That was probably always closer to the truth than the £523 spread we printed in July.
Six rows in ten have no city
Roughly six in ten observations we hold carry no location at all. Agency rate cards quote a price for the country. The Musicians’ Union national gig rates apply everywhere by definition. Until August those rows fed the UK cells and nothing else, so a city with little reporting of its own was answered from a handful of local rows and a regional adjustment.
From this issue a location-less observation counts as evidence for every city it could apply to. Each cell records the split between local and national rows, so the blanket layer is measurable and, if we ever need to, removable. The full mechanism is written up in the methodology, and nothing about region or national cells changed.
How to read a city cell now: 176 of 299 rest on zero local rows
This is the part I want quoted if anything is. Across the city grid, 176 of the 299 cells we publish have no local evidence behind them at all, and only 95 stand on three or more local rows. Those 176 are the national answer wearing a city name. The cell says so, and I would rather you knew that before you put one in a quote or a news story.
Cities will separate again when musicians in them report real fees, city by city, at /rates. Not through a cleverer model. We tried the clever version for four issues and it produced a league that looked precise and was not.
2. Where the market settled: August medians by gig type
Every figure in this section is performer pay. Ticket money, door splits, gear hire and travel are refused at the database layer, so a fee here is what the act took home, net of any commission. The window runs from the July issue’s close on 31 July to 28 August, four weeks rather than a calendar month, and the cells are the 3 September refresh, the last before publication.
Wedding £990 to pub £350: the 3–4 piece table
Six gig types clear the sample floor at 3–4 piece this month. Theatre does not: the 3–4 piece theatre cell dropped out of the national table, and theatre now appears only at solo, at £175 on six observations. Here is where the rest landed.
| Gig type | p25 | p50 (median) | p75 | p90 | n |
|---|---|---|---|---|---|
| Wedding | £717 | £990 | £1,367 | £1,880 | 1,588 |
| Private party | £500 | £866 | £1,292 | £1,808 | 525 |
| Corporate event | £669 | £812 | £892 | £1,082 | 6 |
| Festival | £200 | £669 | £866 | £892 | 5 |
| Club (ticketed) | £450 | £600 | £1,463 | £1,463 | 11 |
| Pubs & bars | £250 | £350 | £600 | £700 | 187 |
UK national medians by gig type · 3–4 piece band · All figures artist take-home (net of commission). Source: GX Index, 3 September 2026 refresh.
Weddings still set the ceiling at £990, but that is £52 under July’s £1,042 and the middle half of bookings now runs from £717 to £1,367. I do not read it as couples paying less. A deeper pool with a different mix pulled the median towards the centre of the distribution, and only the five-plus end of the wedding curve rose. Treat the £990 as the better-founded number.
Corporate at £812 and festival at £669 both rest on single-digit samples at this lineup size and carry a low-sample flag in the report. They are bearings, not benchmarks. Check the n before you quote either in a negotiation.
Pubs at £350: the third fall in a row, and not a withdrawal artefact
A 3–4 piece pub gig pays £350, with most fees between £250 and £600. That is £25 under July and the third issue running in which the figure has eased. July’s drop was a quarantine artefact. This one is not: the rows withdrawn in August came out of the verified-report source, which feeds pub cells almost nothing. Whatever moved the pub figure this time, it was not our housekeeping, though a pool that added 288 rows in four weeks can shift a median without a single room paying less.
The rest of the pub curve: solo £250, duo £375, five-plus £525. Pub money barely flexes by city, because door take, bar margin and a fixed entertainment budget set the price in Newcastle exactly as they do in London. It was the one part of the table the August model change did not disturb. If you gig pubs for a living, our guide to getting paid as a musician covers how to hold a fee once you have agreed it, and the Gig Directory lists what is on near you.
Private party £866, corporate £812: function work stops reading as one market
For four issues a private party and a wedding paid almost the same at 3–4 piece. In July the gap was £42. In August it is £124: private parties at £866, weddings at £990, corporate a further £54 below at £812. That is the first real separation the two function markets have shown at this size.
What it means if you are quoting: the wedding rate is no longer a safe default for a 40th birthday. Price the party against its own cell, and if you book through a third party remember these figures are net of commission, which our post on commission models explains. Venues and planners building a budget can sanity-check any quote against what you should pay a live band.
3. What band size is worth: the wedding curve from £330 to £1,489
The band-size curve is the most practical thing the Index publishes, because it is the question every booker actually asks: what does one more musician cost me? Here it is for weddings at UK national level, with the sample behind each figure.
Four sizes, one curve
Each card is the August median for that lineup, with the weighted sample it rests on.
Per chair, the big band has stopped being the cheaper deal
The duo-to-band cliff got steeper: £464 this issue against £429 in July. That step is where a client stops buying background music and starts buying an event with a rhythm section, and the market prices the difference accordingly.
Above four players something unusual happened. A quartet at £990 works out at roughly £248 a musician. The five-plus median of £1,489, priced as six players (the lineup we assume for that cell), comes to about £248 as well. In July the bigger band still cost less per head. Now the two are level, which I read as the van, the PA and the admin being shared efficiently at both sizes rather than a bargain at either. Model your own lineup and city on the free rate calculator, or compare asking rates on artist profiles.
4. 407 cells from 280: a grid change, not a data surge
The second model change is duller than the first and I think just as important. The published cell count rose from 280 in July to 407 in August, and if you read that as a 45 percent jump in evidence you would be wrong. The pool grew by 141 rows. The grid grew by 127 cells. Those are different things.
Every city now answers the same 23 questions
Each city used to publish whatever cells its own data happened to support. One city answered eleven questions, another six, and the places with the least evidence offered the fewest answers, which is the exact failure the Index exists to fix. From this issue every city publishes the same 23 cells: 13 cities, 299 city cells, plus the regional and national layers on top.
So 407 is a coverage decision. Every one of the new cells is answerable, and every one carries its local count so you can see how much of the answer was collected in that city. That is the trade: more questions answered, more honesty required about how.
The window: 31 July to 28 August, four weeks
This is the first issue compiled on the settled cadence, overnight on the 28th, so its window runs from the July issue’s close on 31 July to 28 August. In those four weeks 288 rows were approved, one public submission cleared the review queue, and 156 previously published rows were withdrawn: a net 133, taking the pool to 4,423 at the window close. Eight more arrived before the 3 September refresh the issue is built from, which is how the cover reads 4,431.
Re-sighting did its quiet work too. The weekly scan recognised 636 listings it already held in the week to 28 August and bumped a sighting counter on each rather than inserting a twin. The monthly artist panel added 201 snapshots on the 28th, the largest single intake of the window.
5. The last of the seed data: 160 rows withdrawn
July’s issue was the one where we took 225 rows out. I hoped that was the end of it. It was not, quite. On 8 August a sweep of the verified rate-import source found 160 rows that were never observations at all: auto-generated seed records from launch, fanned across four variance bands, sitting in a source we publish as externally sourced rate schedules, checked on import.
All 160 are withdrawn. 156 of them had reached print; four were still waiting in review. Nine more rows, none of them yet in print, went the same day on implausible amounts: six agency extractions between £2,437 and £11,091 with no city attached, and three artist-panel entries including a pub gig recorded at £500,150. The test-account check that July added to two sources now runs on all six, and fails closed when it cannot tell. That is how these were found rather than missed. Every figure July printed stays as printed. We correct forward.
What the source table shows now: how much of a cell any one source can own
The visible effect is one line in the source table. That source is externally sourced rate schedules checked on import, not post-event confirmations — there is no post-event verification flow — and what survived the sweep is the 39 verified rate-import rows that remain. The column beside it is a ceiling, not a per-row score: no source class may own more than a fixed share of any published percentile, and it reaches that ceiling only as it earns credibility with volume. Earlier issues printed a per-observation trust weight; that described a model this series no longer runs, and this issue corrects it. Here is the full mix at the window close.
| Source | Observations | Share | Max share of a cell | Status |
|---|---|---|---|---|
| Agency rate cards | 3,496 | 79.0% | 0.10 | Weekly re-scan of three UK agencies; 636 re-sights to 28 August |
| Artist profile rates | 536 | 12.1% | 0.13 | Monthly panel; 201 snapshots on 28 August |
| Web-extracted rates | 173 | 3.9% | 0.05 | Pulled from public pages, human-reviewed |
| MU / Equity recommended | 143 | 3.2% | 0.10 | Union floors, refreshed annually |
| Verified rate imports | 39 | 0.9% | 0.18 | 160 seed rows withdrawn 8 August; 39 rows remain |
| Anonymous submissions | 36 | 0.8% | 0.20 | Public form; rate-limited and reviewed |
| Venue gig budgets | 0 | · | · | Still withdrawn since July: the source stands empty |
| Ticket-derived club estimates | 0 | · | · | Still withdrawn since July: a shadow model, never a fee |
Observation breakdown by source · August 2026, at the 28 August window close (4,423 rows) · Max share = the ceiling on how much of any one cell that source may own; the nine ceilings sum to 1.00. Source: GX Index Issue 05.
Rate cards are four rows in five and can still never own more than a tenth of any cell. That is the point of the quota: volume alone cannot buy an answer, so when confirmed GigXchange bookings, the reference signal, start flowing, a few hundred settled fees are not drowned by thousands of asking prices. They are wired in and still contribute zero rows, for the reason July gave: the bookings we hold are dominated by our own test accounts, and August was spent removing that class of data, not adding to it.
The range check July restored has turned away 169 rows on implausible amounts and 29 more for a gig type it did not recognise. Both numbers are in the report’s quality section, alongside the regional adjustment layer, which blends ONS regional household income and family spending data and only ever fills a gap where no real observation exists.
6. Scoring July’s promises
Every issue scores the commitments the previous one made, and the rule is that nothing gets quietly dropped. July listed three, one of them a restatement of a promise first made in May, and carried one open item still standing since June. Here is how they stand.
| What we said in July | What actually happened | Verdict |
|---|---|---|
| The four city cells promised since May (Nottingham and Sheffield corporate, Brighton and Edinburgh festival) stay on the board and publish only when three trusted observations exist | All four print this issue, but on zero local observations, carried entirely by the blanket layer. The report refuses to score that as delivery, and so do I. | Printed, not delivered |
| The quarantined 225 stay quarantined and are never quietly readmitted | None readmitted. The test-account sweep was extended from the two sources July fixed to all six, and found 160 more rows to withdraw. | Kept |
| The cadence moves to the 28th: data window closes on the 28th, report compiles that night, publication follows within days | Compiled overnight on 28 August, published 3 September. The first issue on the settled window. | Kept |
| First confirmed-booking observations (carried since June) | Still zero rows, still deliberately. Counting test-account bookings would repeat the exact mistake this issue spent a section undoing. | Still open |
July’s promises, scored in August · Source: GX Index Issue 05.
One of those verdicts needs a sentence more. The three-observation rule for a city cell was true at city scope for four issues and is not true now: under the blanket layer a city cell can print with no local rows at all. We chose disclosure over withholding, so the cell prints and says how it was built. If you think that is the wrong call, tell me. It is the decision in this issue I am least certain about.
7. What we honestly cannot tell you yet
Same policy as every issue: the gaps, stated plainly, before anyone cites this data somewhere it does not belong.
- Local evidence, above everything. 176 of 299 city cells rest on no local observations and only 95 stand on three or more. The blanket layer makes those cells answerable. It does not make them local.
- Confirmed bookings: zero rows. Every published cell is still built from asking prices, union floors and verified reports rather than money that provably changed hands on the platform.
- Venue budgets restart from nothing. The first genuine venue budget the Index ingests will be the first row in that source.
- Theatre has thinned to one cell, solo at £175 on six observations. The 3–4 piece theatre cell no longer clears the floor.
- Club and corporate stay directional. Club duo rests on three observations, club solo and corporate duo on four, festival 3–4 piece on five.
- No genre-level cells yet. Every observation carries the tags; publication waits on volume, not engineering.
8. What comes next
Issue 06 covers September, compiles overnight on 28 September and publishes within days. The commitments, deliberately modest:
- Every city cell keeps printing its local count next to its national one, so the blanket layer stays visible and removable.
- Cities separate only through local reporting. No modelling shortcut will be used to manufacture a league. If the spread is still £22 in September, we print £22.
- The 225 and the 160 stay out. If any subset is ever rehabilitated it gets its own section explaining why.
- Theatre returns to the 3–4 piece table only when the cell clears the sample floor on its own.
You never need to wait for the report. The live index rebuilds nightly at 05:00 UTC, rate cards re-scan every Monday, and the artist panel snapshots on the 28th. Back issues stay online at the reports hub, and the evergreen reference for the whole series is the GX Index data reference.
9. How to contribute
This issue is a long way of saying we do not know enough about your town. Eight local observations stand behind Newcastle’s wedding figure. Ten behind Edinburgh’s. A handful of real fees from real gigs would move either of them, and no amount of modelling will.
Report a fee anonymously at the UK Gig Pay Survey: no email, no signup, about sixty seconds. Three approved reports open every cell, trend line and city drilldown for 30 days. If you gig regularly, keep your GigXchange profile rate current, because the monthly panel reads it on the 28th. Venues can help too: the venue-budget source has been empty since July, and the first genuine budget a venue shares becomes its first row.
If you are a journalist or researcher, the Index is free to cite and free to build on under CC BY 4.0. Attribution: “GigXchange Index, gigxchange.app/rates”. The August report page carries the PDF and the 407-row CSV today. The dataset mirrors on Hugging Face, Kaggle and Zenodo (series concept DOI) follow once this issue’s version is minted; July’s are already up.
Frequently Asked Questions
Annual refresh commitment
This guide was published on 4 September 2026 and is refreshed every September. UK live music rate data moves monthly, so annual verification matters. We re-verify every reference, recommendation, and data point once a year. Next scheduled refresh: September 2027. If any claim is outdated before then, email support@gigxchange.app and we will update it within 24 hours.
