1. The month nothing moved, and why that is the finding

I have spent five issues explaining why a number changed. This one is the opposite problem. Every gig type that clears the sample floor at 3–4 piece printed the same median in September as it did in August, to the pound, six for six. The first time I read the diff I assumed the build had reused the previous month’s cells.

It had not. The pool grew, the cells rebuilt nightly as they always do, and the medians landed in the same place. That is worth more to you than any of the movement I have reported before, because a benchmark that jumps every month is not a benchmark. It is noise with a publication schedule.

All six 3–4 piece medians printed twice

Here is the whole headline lineup with August next to it. The p25 and p90 columns are the quarter of the market that pays least and the premium end, and those are the numbers to argue from rather than the midpoint.

Two of those rows are thin enough that you should read the n before you quote them. Corporate at this lineup size rests on six observations and festival on five, and both carry the low-sample flag in the report. The wedding and pub rows are the two you can lean on, and they are the two that did the most interesting thing this month.

The pub cell got 46 per cent deeper and did not budge

A pub 3–4 piece pays £350. It also paid £350 in August, but in August that figure stood on 187 weighted samples and now it stands on 274. Adding 87 samples to a cell and watching the median hold still is the first properly tested pub number the Index has published, and it ends a three-issue slide at that lineup size.

What did move is the top of the cell. The premium end went from £700 to £750 while the bottom quartile stayed at £255. Read that as the better pub bookings pulling away from the ordinary one rather than the ordinary one improving. If you are playing pubs for £255 a night, nothing in this issue says the floor is rising.

£375 to £325: the one tile that fell

The pub duo is the exception, and it is a real fall: £375 in August, £325 in September, with the lower quartile down from £300 to £250 on a sample that grew from 45 to 66. The new rows priced pub duos below the old ones. That is as much as I can honestly say. Whether it is a market moving or a small cell finding its level, one month cannot tell you, and I would rather print the drop and report it again in October than smooth it into a trend I have not evidenced.

The per-musician arithmetic is the part that stings. The Musicians’ Union national gig rate recommends a floor of £167.16 per musician, and a £325 duo splits to £162.50 a head, which lands just under it. The £375 it replaced cleared the floor at £187.50. This one does not, and that is the sharper way to read the fall. How to set your gig price covers arguing from that floor rather than against it.

2. What band size is worth: £330 solo to £1,489 at five-plus

Weddings are the deepest cell in the Index, which makes the wedding band-size curve the most reliable thing we publish. In September it is almost exactly what it was, and the one figure that moved on it moved by £7.

The curve, and the £7 that moved on it

Solo, duo, 3–4 piece, five-plus. Three of the four are unchanged to the pound, and the five-plus median has now printed identically for two issues running.

Per chair, a bigger band is the same band economics repeated

The steps between sizes are the useful part. Solo to duo is £203. Duo to a 3–4 piece is £457, a shade gentler than the £464 August printed. Quartet to five-plus is steeper still at £499. Both of the upper steps dwarf the first one, which tells you the market prices a band as a band and a soloist as something else entirely.

Divide by heads and the curve almost flattens: around £267 a musician for a duo, £248 for a quartet, and between £248 and £298 across a five-plus lineup depending on whether that means five players or six. That is close to flat rather than a clean discount: a quartet musician takes around £19 less than a duo musician, while a five-piece pays about £298 a head, more than either. A bigger band is not reliably a cheaper band per musician, which is why what UK gigs pay reads so differently depending on how many people you are splitting it between.

3. Thirteen cities, fourteen pounds apart

Since August a city cell is fed by its own observations plus every observation that carries no location, on the reasoning that a national price list is a price being charged in that city too. The consequence is a league table fourteen pounds wide, down from twenty-two in August and five hundred and twenty-three in July. August promised that if the spread was still flat in September we would print it flat. It is, so here it is.

The league table, and the local counts under it

The local column is the one to read. It counts observations actually collected in that city, as against the 1,069 location-less rows that sit behind all thirteen cells identically.

Fourteen pounds across thirteen cities is not a discovery about Britain. It is a description of how little local evidence we hold, and the cells say so on their face. Treat the ordering as our evidence talking rather than the market, and use the national figure when someone asks what a wedding band costs in the UK.

Why the cities we know best sit at the bottom

This is the uncomfortable part. Manchester is last on 84 local rows and London is joint second-last on 96, and those are our two deepest local samples. Edinburgh and Glasgow top the table on ten and eighteen rows respectively.

The mechanism is visible once you look at the two medians involved. The 1,069 location-less rows have a median of £1,079, and each city adds only its own 8 to 96 local rows on top, so every city lands within a few pounds of that shared baseline. A city with more local evidence gets pulled further off it, and this month local evidence happened to sit below the national price lists. The national cell is the counter-example. It is built from the same 1,069 location-less rows plus all 536 located rows from every city at once, rather than one city’s handful of them, and those 536 have a median of £850. Carrying the full weight of the cheaper located evidence is why the national figure prints £990, below every city in the table. Bristol, third on 64 local rows, breaks the pattern in the other direction. I am printing a hypothesis, not a result.

4. The most instructive number in the issue is 213

For two issues running this section reported faults in our own machinery: 225 rows quarantined in July, 160 seed rows withdrawn in August. This month it reports the gates working, which is a duller read and a much better one.

550 rows in, 213 turned away, nothing withdrawn

Across the window 550 new observations were approved and 213 more were refused at the door by the range check, which bounds what a gig can pay per musician per hour for its type. That is 28 per cent of everything presented this window stopped before it could vote, and 62 per cent of the agency rows offered. The refusals break down as 148 artist-panel entries and 65 agency extractions, and every one is logged with a reason, because the audit trail is part of the product rather than an internal convenience.

Cumulatively the range check has now refused 391 rows since it was restored in July. More importantly, not one row had to be corrected or withdrawn after approval this window. That is the first clean window in three, and it is the difference between a gate that catches things on the way in and a sweep that catches them after they are already in print. The 225-row quarantine July opened is unchanged and still out of the pool, as is August’s 160.

Where the 550 came from, and the 632 re-sights that did not count

Two sources supplied every new row: 511 from the monthly artist panel and 39 from the weekly agency rate-card scans. The anonymous public form took nothing at all this window, which is the gap I keep asking readers to close. The live mix at the window close looks like this.

Read the first and last rows together and you have the whole design argument. Rate cards are three rows in four and may still never hold more than a tenth of any published cell. Anonymous submissions are 36 rows in 4,772 and carry the highest ceiling in the system, because a fee someone was actually paid beats a thousand asking prices. Not one of those ceilings has changed since 8 August, so September is the August model run on a deeper pool, and the methodology page carries the full nine-source table.

The re-sights matter too. A listing still on an agency’s site next Monday tells you its price is holding, but it must never vote twice, so a recognised listing bumps a sighting counter instead of inserting a twin: 632 re-sights this window, 560 of them in the final week, none of them widening a sample. The adjustment layer that fills genuine gaps blends ONS family-spending data with public industry price benchmarks, and it only ever fills gaps. Where real observations exist, they win outright.

5. How to read a cell before you quote it

There is no single right fee for a gig, so we never print one. Every cell gives you four points along the distribution instead, and the whole value of the Index is in using them rather than grabbing the median and running.

Four percentiles, one decision

Take the wedding 3–4 piece cell as the worked example, since it is the deepest one we hold: p25 £717, p50 £990, p75 £1,370, p90 £1,880. Here is how I would use those four numbers in an actual negotiation.

1. Find your cell, not your gig type. A fee is the intersection of three things: where, what kind of gig, and how many musicians. A quartet playing a wedding and a quartet playing a Sunday pub slot are different cells with £640 between their medians, so start at the live lookup and pick all three.
2. Check the n before you trust the number. Our wedding 3–4 piece cell has 1,605 weighted samples behind it and our corporate 3–4 piece cell has six. Both print a median. Only one of them is a benchmark, and the report flags the thin ones with a dagger for exactly this reason.
3. Open at p50, justify at p75. The median is a defensible market rate that is hard to argue down, because it is where half the bookings sit. Push toward p75 when you have real audience pull, a repeat-booking history, or a date the client cannot move. The rate calculator models your own lineup and date.
4. Treat p25 as the alarm, not the floor. Anything below the lower quartile is a sign you are being undercut rather than priced. Cross-check it against the union recommended minimum of £167.16 per musician, and if an offer fails both tests it is not a gig, it is a favour.

6. Scoring August’s promises

Every issue scores the commitments the previous one made, and the rule is that nothing is quietly dropped. August listed four and carried one open item that has been standing since June. Here is where they land.

One correction belongs in this section rather than buried in the methodology. August said roughly six in ten observations carry no location. The exact figure is 54 in every 100, and the September issue prints that instead. It is a small enough difference that nobody would have caught it, which is precisely why it is worth saying out loud.

7. What we still cannot tell you

Same policy as every issue: the gaps stated plainly, before anyone cites this data somewhere it does not belong.

• Local evidence, ahead of everything else. 174 of the 299 city cells rest on no observations collected in that city, and only 99 stand on three or more. Both numbers improved on August, by two cells and four cells, which is movement you would need a microscope to see.
• Confirmed bookings: zero rows, fifth issue running. Every published cell is still built from asking prices, union floors and verified reports rather than money that provably changed hands.
• Venue budgets restart from nothing. The first genuine budget a venue shares becomes the first row in that source.
• Theatre is a single national cell at £175 solo, a rung below pubs, on six observations.
• Clubs stay directional. Club duo rests on three observations and club solo on four. They are bearings, not benchmarks.
• No genre-level cells yet. Every observation carries the tags, so publication is waiting on volume rather than engineering.

8. What comes next

Issue 07 covers October, compiles overnight on 28 October and publishes within days. The commitments, deliberately modest:

• The pub duo gets reported again, whatever it does. A £50 fall on a 66-row cell is not a trend. October will print the figure on whatever sample it has by then and say which of the two readings the extra evidence supports.
• Local counts keep printing, and cities still separate only through local reporting. If the league is fourteen pounds wide again, it gets printed fourteen pounds wide again.
• The 225 and the 160 stay out, and the model stays where it is unless there is a reason to move it that I can explain on the page it affects.
• Confirmed bookings and venue budgets stay at zero rather than being filled with something weaker to make the source table look complete.

You never need to wait for the report. The live index rebuilds nightly at 05:00 UTC, agency 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

Eight local observations stand behind Newcastle’s wedding figure and ten behind Edinburgh’s. The public form took no reports at all this window, and the league compressed instead. A handful of real fees from real gigs would move either city, 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 as well, since the venue-budget source has been empty since July. If you think a decision in this issue is wrong, tell me and I will print the argument.

Journalists and researchers: the Index is free to cite and free to build on under CC BY 4.0, with the attribution “GigXchange Index, gigxchange.app/rates”. The September report page carries the PDF and the 412-row CSV, and this issue’s dataset is mirrored on Hugging Face and Kaggle with a citable Zenodo DOI. For the wider sector picture UK Music’s This Is Music 2025 is the standard reference, and the gig directory is where the shows themselves are listed.

Annual refresh commitment

This guide was published on 6 October 2026 and is refreshed every October. 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: October 2027. If any claim is outdated before then, email support@gigxchange.app and we aim to update it within one working day.