AI for vetting acts — spotting bad bookings before they happenThe 4-step triage flow UK venue bookers use to vet 50+ unsolicited pitches a month in 3 minutes each — with the 6 red flags AI catches in 60 seconds. Field-tested. Annual refresh.

This is the fourth post in our AI for UK music venues series. The first three covered 12 inbox-reply prompts, how to use AI to fill your venue and the venue tools stack. This one solves a specific upstream problem: most UK venue inboxes don’t fail at writing replies — they fail at knowing which pitches to reply to in the first place.

We’ve seen the same pattern across UK grassroots venues we work with on GigXchange: a one-person booking team gets buried under unsolicited pitches, can’t triage them properly, and books on gut feel from the most recent or most polished email. AI vetting fixes the upstream problem.

Why bad bookings are expensive — the £200-£800 number

UK grassroots venues operate on a 2.5% average profit margin (Music Venue Trust 2025) — the thinnest in any UK hospitality category. With more than half (53%) of grassroots venues posting no profit at all in 2025, the operational reality is that a single bad booking can wipe out a week’s profit, and a run of three can wipe out a month’s. The MVT report is explicit: 30 venues closed forever in the year to July 2025, with another 48 dropping live programming entirely. Booking risk is no longer abstract; it’s the difference between staying open and not.

Here’s the all-in cost of one bad booking at a typical UK grassroots venue (capacity 80-150, weekend evening, £8 door):

Verification note: figures reflect real-cost benchmarks from UK venues we work with on GigXchange and the cost models published in the MVT annual report. Your numbers will vary by capacity, ticket price and bar margin; substitute your own door + bar takings for accuracy.

The pattern: one bad booking erases the profit from 4-6 good ones. AI vetting doesn’t need to be perfect to pay for itself. Catching even 1 bad booking a quarter pays back the entire £20/month subscription many times over.

The size of the inbox problem

Across the UK grassroots venues we work with, the volume of unsolicited pitches per booking inbox falls into three brackets:

The realistic-time-available column is the critical one. Booking is rarely someone’s full-time job; it’s ~20% of a venue manager’s week, alongside rota, stock, compliance and the actual gig nights. The triage gap (theoretical time vs available time) is where bad bookings are made.

The 4-step flow below collapses the per-pitch time from ~30 minutes to ~3 minutes, which is what closes the gap. That ~85% reduction is in line with Anthropic’s 2025 productivity research, which finds Claude speeds up individual knowledge-work tasks by approximately 80% on average.

Step 1 — The 30-second triage prompt

The single highest-leverage AI use is the inbox-triage prompt. Paste a band’s pitch email plus their two most-shared links (Spotify, Bandcamp, Instagram, Linktree) into ChatGPT or Claude. Ask for a one-paragraph verdict.

The prompt template:

You are a UK grassroots venue booker. The venue is [VENUE NAME, CITY, CAPACITY, GENRE FOCUS]. Below is an unsolicited pitch from a band. Vet it in 4 lines:

1. Genre and culture fit (yes / no / borderline) for this venue.
2. Realistic UK draw (low / medium / high) based on social and gig history.
3. Top red flag if any (inflated claims, dead engagement, no real gigs, paid followers, recycled content, professional pitch / amateur act).
4. Verdict: yes-shortlist / maybe-follow-up / polite-no.

Use British English. Be blunt. Don't be encouraging. Default to no when in doubt.

[PASTE PITCH EMAIL]
[PASTE BAND LINKS: Spotify, Bandcamp, Instagram, Linktree]

Run this on every unsolicited pitch. The output is a 30-second read. The AI will get ~70% of borderlines right; you check the “yes-shortlist” and “maybe” outputs manually, polite-no the rest.

Critical rule: never auto-send the no-thanks reply. Draft with AI, send manually. UK artists talk to each other; a venue that sends obviously bot-rejected emails gets a reputation in months. Save the AI for triage and drafting; keep the send-button human.

Step 2 — The 3-minute social-presence summary

For acts that pass the triage step, run a deeper check. Paste the band’s Instagram URL, Spotify URL, and most-recent live-gig poster (or photo) into the LLM. Ask for the social-presence summary.

Give me a 6-line UK booker's read on this act, based on the public-facing assets:

1. Real-vs-paid follower ratio (use comment-to-follower ratio and follower spike patterns as the proxy).
2. Engagement quality: are followers actually fans, or is this looking-busy / ghost-town?
3. Last 12 months: how many real, ticketed UK gigs can you find evidence for?
4. Is the EPK / press kit consistent with the social presence?
5. Genre and aesthetic fit for [VENUE NAME, CITY, CAPACITY].
6. Verdict: yes / maybe / no — with the single most important reason.

Be blunt. Default to no when in doubt.

The LLM can’t actually fetch live URLs in most setups, so paste the visible text and key numbers (follower count, post frequency, recent show photos with venue tags). For the live-fetch version, use ChatGPT’s browse feature or paste a screenshot. Both work for vetting purposes.

What the AI is good at here: pattern-matching across hundreds of similar profiles. A booker reviewing their 50th pitch at 11 PM will miss the “follower count went 2k → 12k in March, then flat” pattern. AI catches it on every read.

Step 3 — The draw-verification cross-check

Inflated draw is the most expensive bad-booking type because the venue pays the act in full and gets an empty room. The single fastest AI vetting check is cross-referencing claimed gig history with public sources.

The prompt: paste the band’s claimed past gigs (from their pitch or website), then ask AI to verify which can be confirmed against public listings.

Here are the past gigs this act claims:
[PASTE 5-10 RECENT GIGS WITH DATES + VENUES]

For each, tell me:
1. Is this a UK grassroots venue / mid-size / festival / pub?
2. Realistic typical draw at that venue size (low / medium / high).
3. Cross-reference: can this gig be confirmed via a poster, social post, ticket-platform listing, BBC Introducing page, or local press archive? (You won't have live web access; flag the ones that need a manual check.)
4. Pattern: does the gig list look like a real working band, a hobby band overstating, or a paid-festival-slot fabrication?

Be blunt.

Then spend 90 seconds googling the 2-3 gigs the AI flagged for manual check. Posters, ticket links, BBC Introducing pages and local-press archives confirm or deny within seconds. If you can’t find evidence of any of the “recent” gigs, that’s the single biggest red flag in this whole flow.

Step 4 — The final 60-second human check

AI is good at pattern-matching. It’s bad at three things that actually matter for venue bookings: cultural fit, local relationships, and gut feel from a video clip. The final step in every vetting pass is a 60-second human check.

The minimum check:

• Watch 30 seconds of one live video. Not the studio recording. Live. Look for: stage presence, sound, audience response, whether the singer can actually pitch when they’re not double-tracked.
• Check one reference. If they say they played The Camden Underworld, message a booker you know there. The UK live-music network is small; one DM gets you the truth in 24 hours.
• Read the pitch email out loud. If it sounds like a template, that’s data. Acts who write a venue-specific pitch are more likely to honour the booking.

This is the layer AI cannot replace. Don’t skip it.

The 6 red flags AI catches in 60 seconds

Across our network, these are the six patterns AI vetting catches consistently and a tired human booker doesn’t. Memorise them; ask the AI to flag them explicitly.

Why fake-follower checks matter: across all influencer categories, 25-28% of followers are fake or inactive on average (HypeAuditor / Modash 2024-25 audits), and the influencer-fraud market loses an estimated $1.3 billion globally each year to inflated metrics. Music-industry accounts often sit at the higher end of that range. Pattern-matching the ratio between follower count and comment-to-follower ratio is the single most reliable AI signal.

None of these flags is automatic disqualification. They’re prompts to ask one more question (“your last verifiable gig is 2023 — what have you been doing since?”) before booking. Most acts with one flag are fine. Three+ flags is when bad bookings happen.

What AI can’t see — the 3 limits of vetting prompts

Three things AI vetting will not give you, even on a good day:

1. Cultural fit. Whether the band’s vibe matches your venue’s regulars is a judgement call. AI gives you the data; the call is yours. A polished blues-rock four-piece looks identical to another on paper, and one of them will fit your Friday night and the other won’t.
2. Real references. A 30-second WhatsApp to a booker you trust at another venue beats any AI summary. The UK grassroots circuit is small enough that a phone call still works. Use it for the bookings that matter.
3. The live performance gap. Some bands sound great recorded and weak live. Others sound rough recorded and pull the room together when they play. AI can’t close that gap; only watching the live video can.

Use AI to triage 80% of pitches and cut the time on the obvious yes-or-no decisions. Spend the time you save on the human checks for the bookings that matter.

The legal layer — what NOT to do with AI vetting

Three rules to stay clean:

1. Don’t auto-reject. Use AI to draft the no-thanks reply, but always have a human send it. UK acts spot bot-rejections in days and the venue reputation hit is permanent.
2. Don’t share AI-generated assessments publicly. If your prompt produces a withering one-liner about a band’s draw, it’s a private working note — don’t post it to a forum or share it with another venue. The ICO guidance on UK GDPR covers personal-data processing; assessments of identifiable artists count.
3. Don’t train AI on bands without permission. Pasting one band’s pitch into ChatGPT to vet that booking is fine. Building a database of band profiles you’ve fed into AI is data-processing territory; you’d need a privacy notice and a lawful basis. Almost no UK venue is doing this; if you are, get advice.

The default safe pattern: vet one act at a time, in your own LLM session, for your own decision-making, and don’t share or store the output.

The £20/month vetting stack

Everything in this post can be run on:

The take-home stack: ChatGPT Plus alone, £20/month. Don’t add specialist “AI for venues” platforms — they’re wrappers. We unpack this in the venue tools post.

The annual refresh commitment

This post is refreshed every May. AI tools and the patterns of bad pitches both shift fast. We re-test every prompt and update the red-flag catalogue once a year. Last refreshed at the date stamped above; next scheduled refresh is May 2027. If you’ve seen a new bad-booking pattern AI is or isn’t catching, we’d genuinely like to know.

Where AI vetting ends and GigXchange begins

AI vetting is a triage layer. It tells you which pitches deserve your time. It doesn’t solve the upstream question: why are 50 unsolicited pitches landing in your inbox in the first place, and can the right ones land in front of you instead?

That’s where we plug in:

• Browse vetted UK acts: the GigXchange artist directory shows verified UK artists, agents and bands with real gig history attached. Search by city, genre, capacity-level. No unsolicited pitches; you reach out to the ones you want.
• See what’s actually playing: the live UK gig directory across 40+ cities is the fastest way to see which acts are actively gigging in your region right now. Real, ticketed gigs — not pitch claims.
• Inbox-reply prompts: our 12 ChatGPT prompts for venue bookers covers the yes / no / follow-up reply layer. Pair it with the vetting flow above and a typical 50-pitch month becomes a 90-minute job, not a 25-hour one.
• The full venue cluster: see AI for UK Music Venues (2026) for all 9 posts in the series.

Vet with AI. Book direct on GigXchange. £20/month, properly deployed, catches 80% of bad bookings before they ever reach your calendar — and saves the £200-£800 hit each one would have cost.

Have a vetting workflow we’ve missed, or a red flag pattern we should add? We refresh this post once a year and rely on UK venue feedback to keep the catalogue current. Last refreshed at the date stamped above.