Thirty-Seven Problems from CHS Manchester, Answered
On 30 September at CHS Manchester, you typed 37 problems into your phones in about three minutes. I said on stage I'd answer every one of them. This is me keeping that promise.
If you weren't there: session one was a QR code and a simple question. What slows you down? The thing you rebuild every week, the system that doesn't talk to the other system, the bit of Tuesday afternoon you'd pay someone else to do. Everything went up on screen anonymously, grouped itself into themes, and the room voted.
The room picked spreadsheets that don't talk to each other (13 votes) and everything buried in the inbox (12 votes), with the same report every time close behind. In session two I built Proposal X-Ray live, for the venue and supplier problem (the best of it is heading into Zoby Desk), and in session three we launched the CHS Innovation Hub, which is where you're reading this.
But one tool doesn't answer 37 problems. So below is every single one, grouped the way the room grouped them. For each, a few honest lines on what's actually going on, and where to go next: a course, a ready-to-copy prompt, a tool, or a straight answer that AI isn't the fix.
Everything linked lives here on the hub. Find your problem, follow the link, and you'll have something to try on Monday.
1. Spreadsheets that don't talk to each other
The room's number one, with 13 votes. Eight of you described some version of hand-copying the same figures between systems that have no idea the others exist.
The pattern I showed in session two still stands. Most of these aren't apps waiting to be built. They're connectors: things that already exist, talking to each other. Give your AI access to the files (or just upload them), save the rules once, and have it tell you what's new, changed or missing.
"Too many spreadsheets. It would be great if they all spoke to each other"
The fastest win isn't making them all speak. It's making AI compare two of them for you: new rows, changed values, missing entries, one clean file out. Do that once by hand in a chat, then save it so it runs the same way every week.
Go here: Our spreadsheets don't talk to each other, the training I promised in session two. Then Claude Cowork for Events Professionals or Copilot Cowork, Researcher and Analyst, depending on what your organisation pays for.
"I move only specific data columns from one spreadsheet to another every week"
This is the perfect first build, because it's identical every week. Describe it once (these columns, from this sheet, into that one, matched on this field), test it on last week's files, then turn it into a saved skill or a scheduled task. Honestly, it might also be a ten-minute Power Query or a lookup formula, and AI can write that for you too.
Go here: From Asking to Running: Skills, Scheduled Tasks and Loops, and the Copilot in Excel course if you'd rather keep it inside Excel.
"Data entry is time consuming"
Most event data entry is copying something that already exists somewhere else: a PDF, an email, a form, a phone call. AI is very good at reading the messy version and filling the tidy one. Your job becomes checking, not typing.
Go here: Automate the Boring Stuff and What to Upload to AI. If a lot of it starts as phone calls, Zoby Voice captures the call and pushes the details straight into your CRM.
"Different event tech that isn't connected e.g reg system, abstracts and app"
I'll be straight with you: AI won't magically wire three platforms together. What actually helps is picking one system as the master, then using its integrations or an automation tool (Zapier, Make or n8n) to push changes out from it. AI's job is writing the automation for you in plain English, and checking the outputs match.
Go here: Our reg system, abstracts and app don't talk to each other, then n8n, Make and Zapier: AI Workflow Automation.
"Manual programme updates on event app and website separately (all from multiple spreadsheets…)"
Same root cause as above. The programme lives in five places, so every change happens five times. One master programme file, one automation that publishes it, and AI checking the versions against each other before the doors open.
Go here: the same answer covers this, and the single source of truth lesson in Function Sheets, BEOs and the Handover applies directly, even though it's written for venues.
"Web scraping websites of member organisations to put on the umbrella organisations site to publicise their events"
Very doable, and you're right that it's the only route that works, because members won't fill in forms. The tool I'd point you at is Firecrawl. Give it a list of member event pages and it pulls out the fields you choose (name, date, town, link) so every site comes back in the same shape, ready for your listing. There's a free tier to try it on, and it connects to Claude and other assistants so the AI can run it for you. Put it on a weekly schedule and you only see what's new or changed. It won't be perfect: some sites block automated visits (we covered this on 7 October) and redesigns break things, so a person checks before anything's published.
Go here: I want to pull our members' events off their websites, then Always-On Content for the monitoring side.
"Uploading sponsorship inventory to Salesforce/subsequent amends"
Salesforce now has its own connector for Claude and ChatGPT. Once your admin switches it on, the AI can check what's already in Salesforce and make the changes itself, with your say-so. No import file, and no Zapier or Make needed. Without the connector, AI can still take your inventory in whatever format it's in, produce a clean import file matched to your fields, and compare an amended version against the last one so you only upload what changed.
Go here: two of you asked, so we wrote it: I need to get data in and out of our CRM without retyping it. It covers Salesforce, HubSpot and what to do if you have neither connector.
"Checking data and updating reports accordingly (specifically for me within Salesforce, such as allocating hotel accommodation to buyers)"
Allocating rooms is a rules problem: who's eligible, what's available, who asked for what. AI is good at applying rules you write down and flagging the ones that don't fit. With Salesforce connected to your AI, it can read the records itself, propose the allocation and the exceptions, and update Salesforce once you've checked them.
Go here: the CRM answer for the allocation prompt, and Group Bookings, Rooming Lists and the Detail That Bites for the rooming side.
2. Building the same report every time
Seven votes, eight problems. Budget reports, board reports, sales reports, ROI, trustee papers. Different audiences, same pain: pulling numbers out of scattered places to rebuild something that looks identical to last month.
The pattern is simple. Put your template and your best past report into a project, so the AI knows exactly what good looks like. Point it at the data. Run it when you need it, or put it on a schedule. Then check it before it lands anywhere important, because AI should never be the thing that decides a number.
"Reporting up to senior management around risk, resource issues, event progress etc."
This is the most automatable report there is, because senior management wants the same five headings every time. Give AI your project tracker and a template (a simple red, amber, green layout works well) and have it draft the update on a Friday morning for you to sense-check.
Go here: I build the same report every month, the training I promised in session two. Then Your Morning Briefing, which is the same idea pointed at you rather than your boss.
"Event budget reporting (taking numbers from various spreadsheets to create a report)"
AI is genuinely useful here for structuring the report, spotting missing cost lines and explaining variances in plain English. It must not be doing the maths you sign off on without you checking it.
Go here: Budgets, Quotes and Reconciliation, especially the lesson on tracking spend and variances.
"Sales reports that don't detract from sales time"
Good salespeople hate reporting, which is exactly why it should be automated. If your pipeline lives in a CRM or a sheet, AI can turn it into the weekly update without you touching it, and flag the deals that have gone quiet.
Go here: the "Write your weekly events pipeline update" prompt in Automate the Boring Stuff, then Conversion: Turning Enquiries into Confirmed Business for win/loss analysis that actually tells you something.
"Writing post event reports is boring!! Data is the same, trying to track and present it"
Agreed, on both counts. The trick is capturing as you go (five minutes on the day) so the report writes itself from your notes, your numbers and your best previous report as the template.
Go here: Post-Event Reporting and Debriefs, plus The Post-Event Debrief Prompt That Actually Gets Used. In a rush? I have to write a post-event report and I have no notes.
"Reporting my ROI for my board review"
ROI falls apart when the objectives weren't written down at the start. AI can help you work back to sensible measures, calculate cost per outcome, and write it two ways: one for the board, one for the team.
Go here: The client wants ROI in the post-event report and My CEO wants to know which channel filled the room. To make next year's ROI easier, tag every campaign link with Zoby UTM (free).
"Partnerships reporting. We have no way of properly tracking or measuring results."
The reporting is the easy bit. The hard bit is agreeing what you're measuring before the partnership starts, and tracking it consistently. Start with tracked links and a simple deliverables tracker per partner, then the report becomes a summary of data you already have.
Go here: the sponsor reporting and renewal lesson in Sponsorship and Exhibitor Sales with AI, and Zoby UTM for the tracking.
"Writing council papers for trustees"
Trustee papers have a fixed shape, a cautious audience and a lot of source material. AI is good at turning a pile of updates into that shape and at reading the draft back as a sceptical trustee would. It can't decide what the council needs to know, so that part stays with you.
Go here: I have to write the council paper for trustees, plus the "Read a document as a sceptical stakeholder" prompt in the Prompt Library.
"I'm trying to transform my client - building dashboards etc for them. They struggle with copilot so it's a tough one. Need integrate AI into our events / registrations acceptances arrive a via emails but get lost in other conversations - people do not register via Eventbrite etc and find it a barrier"
Three problems in one, so three answers. On Copilot: struggling is normal, and it's usually a confidence and permissions issue rather than a skills one. On the emailed acceptances: AI can sweep an inbox for replies and build the registration list for you. On the barrier: a simpler sign-up page might beat Eventbrite for this audience.
Go here: Registrations arrive by email and get lost covers all three, including the dashboard. Try Zoby Tickets (free) as a lighter registration option, and When People Do Not Want To: AI and a Reluctant Team for the client.
3. Everything's buried in the inbox
Twelve votes, one behind the winner, and home to the single most upvoted problem of the session. Bookings, supplier replies, registrations and invoices all arrive as email and disappear into the pile.
The good news: this is where the AI tools have moved fastest this year. Claude, ChatGPT, Copilot and Gemini can all now read your inbox (with your permission), sort it, and draft replies for you to send. The rule I'd hold to is sorted and drafted, never sent. You press send.
"Manage my inbox - emails from suppliers, promotional marketing, replies to suppliers about event status, best filing practise, flagging, categoriing, searching for"
The most upvoted single problem, so it gets the most honest answer. Half of this is AI and half isn't. The non-AI half is a filing system you'll actually keep: a handful of folders or labels, rules that strip marketing out before you see it, and one flag that means "waiting on someone". The AI half is a daily sweep that sorts what's left, files it, drafts the obvious replies, and tells you what's been waiting longest.
Go here: My inbox is out of control, with a route for Outlook, Gmail, Claude and ChatGPT. Then Your Morning Briefing.
"Email inbox management"
Same answer as above. Start with the morning briefing course this week. It's the one build that pays you back every single day.
Go here: Your Morning Briefing.
"Chasing sponsors and exhibitors"
Chasing is two jobs: knowing who owes you what, and writing the chase without sounding like a robot. Keep the first in a tracker (AI can build it from the contracts), and let AI draft individual chasers that reference the specific thing that's late.
Go here: the exhibitor documentation pack and tracker lesson in Sponsorship and Exhibitor Sales with AI, and I need to chase twelve suppliers and I have no time.
"Receipt reconciliation"
AI can read a photo of a receipt or a PDF invoice and pull out the supplier, date, amount and VAT into a table. Then it can match that table against your budget or purchase orders and show you what doesn't line up. Dedicated expense apps do the capture side well too, so check what your finance team already uses.
Go here: I've got a pile of receipts and the expenses are due, and for supplier invoices, Five supplier invoices, one closed budget.
"Managing invoicing for commison. Getting a venues accounts details to send commission to, getting the events final spend, sending the venue the invoice, receiving payment and tracking it"
This one's close to home, because it's exactly how we work at the agency end too. It's a five-step chase that lives in email, which is why it slips. Put it in one tracker (venue, event, final spend, commission due, invoice sent, paid), let AI pull the final spend out of the venue's email and draft the invoice and chasers, and have your morning briefing flag anything unpaid after 30 days. If you use Xero or QuickBooks, both now connect to AI assistants too.
Go here: I invoice venues for commission and keep losing track, then Budgets, Quotes and Reconciliation for the money discipline.
4. Getting straight answers out of venues and suppliers
Only three votes, but this is the one I built live, because it was the clearest problem to show end to end. Chasing venues for the details you asked for, then trying to compare what comes back in a dozen different formats.
The result was Proposal X-Ray: four proposals in, any format, one like-for-like table out. True cost rather than headline price, red flags in plain English, and a chaser email in one click. In session three we added what you asked for: your brand, commission and fixed costs, exports, venue photos and a map. It took eight replies to Claude and two Lovable prompts.
Rather than launch it as a separate tool, we're adding the best of its features to Zoby Desk, our venue sourcing tool, which already captures the brief, matches venues and sends a branded proposal.
"I want venue data in one consistent format?"
This is what Zoby Desk is for, and we're adding the X-Ray comparison features to it. If you'd rather do it yourself in a chat window, the trick is defining your columns first and making AI fill them from each proposal, leaving a cell blank rather than guessing.
Go here: Zoby Desk if you source venues regularly. For a free route, the case study Eight Quotes, Eight Formats and I need to compare five quotes that are all in different formats.
"Comparisons on supplier costs that are intuitive to the differences between key elements"
The key word there is differences. A comparison table where every row looks the same is useless. Ask AI to put the cheapest option in bold, flag anything excluded or charged extra, and explain in one line per supplier why the totals differ.
Go here: Zoby Desk, where we're adding this side-by-side comparison. Or do it yourself with Making quotes comparable and the comparing quotes lesson in Venue Sourcing and Chasing Suppliers.
"Getting the correct information from a venue when I send them an enquiry"
Most wrong answers start with a vague question. A tight brief with the exact fields you need back (and a request to reply in that format) gets you far better answers first time. When they still miss something, AI can spot what's missing and draft a polite, specific chaser.
Go here: Venue Sourcing and Chasing Suppliers with AI, especially writing a venue brief and chasing without being a nuisance. If they've gone silent: A venue will not reply to me. And Zoby Desk sends the brief in a consistent format every time.
"A tool to create partner rate sheets from information stored in various teams (pricing, commission, products) into exportable pdfs that can be shared with partners."
This is the reverse of comparing proposals: instead of reading lots of documents into one table, it's turning one set of data into lots of documents. Very buildable. The starting point is getting pricing, commission and product into one sheet that each team owns a column of. After that, generating a branded PDF per partner is a small build, and Lovable is where I'd build it.
Go here: I need partner rate sheets built from three teams' data, then Vibe Coding: Build Without Coding if you want to try building it the way I did on stage.
"Saying no to clients, constantly having to find the bits in various contracts to determine what we can't/shouldn't do"
Put your contracts into a Claude or ChatGPT project and ask it questions in plain English: "Can the client bring their own catering?" It answers with the clause it's quoting, so you can check it in seconds rather than re-reading 40 pages. It's an assistant for finding the clause, not a lawyer.
Go here: I need to know what the contract actually says we can't do, then Claude Projects: Deep Dive.
5. Delegate lists, room counts and table plans
No votes in the final round, but five of you described hand-building lists from data that already lives somewhere else. These are some of the best jobs to hand to AI, because they're rule-based and the rules are already in your head.
"Creating table plans manually"
A table plan is a puzzle with rules: these two together, those two apart, VIPs near the stage, dietary needs grouped for the kitchen. The newer reasoning models are genuinely good at constraint puzzles like this, as long as you write every rule down. Expect a strong first draft you then tweak, not a finished plan.
Go here: I'm doing the table plan by hand, then ChatGPT Reasoning Models.
"Pulling lists of guest and data."
If the data's in an export, AI can clean it, de-duplicate it, split it and give you exactly the list you need (dietary list for catering, name badges, a VIP list) in seconds. Just strip out anything personal you don't need before you upload it.
Go here: ChatGPT Data Analysis: Spreadsheets, Charts and Files or Copilot in Excel, and the "Remove personal data before pasting" prompt in the Prompt Library.
"Room counts"
Room counts go wrong when the rooming list arrives late and in the wrong format. AI can normalise a messy rooming list into standard columns, count room nights by type and date, and flag the errors before the hotel sees them.
Go here: Group Bookings, Rooming Lists and the Detail That Bites and The rooming list has arrived and it is a mess.
"How do I monitor dietary needs/accessabikty needs"
Two parts. First, ask the same questions the same way every time, so the answers are comparable. Second, keep them in one place that feeds the catering sheet and the venue brief automatically. One caution: health and access information is special category data under UK GDPR, so be careful which AI account you put it into.
Go here: Accessible and Sustainable Events with AI, the special requests lesson in Group Bookings, and AI at Work: Accounts, Data and Getting Permission for the data side.
"Building unique confrence programmes."
AI is a great sparring partner for programme design: themes from your audience research, session formats you haven't tried, and checking the draft for clashes, gaps and too many panels in a row. It can't tell you who's a good speaker. That bit is still your network.
Go here: I'm building a conference programme from scratch, then Brain Dump to Plan.
6. Answering the same questions again
Three of you are typing near-identical replies over and over. The fix isn't a chatbot on day one. It's writing your best answers down once, so AI can draw on them instead of guessing.
"I answer the same 15/25/50 questions from delegates, every event"
Your sent folder already contains the answers. AI can mine it for the questions that keep coming up and your best reply to each, giving you an FAQ and a reply library. From there you can draft replies in seconds, or put the FAQ in front of delegates so they stop asking.
Go here: the handling volume lesson in Delegate and Attendee Communications, I keep answering the same twenty questions, and Zoby Knowledge, which does the inbox mining for you.
"Responding to enquiries in innovative, engaging ways that isn't fully copy and paste for each one"
The sweet spot is a library of strong paragraphs you trust, plus AI personalising the opening and the bits that genuinely depend on their brief. That way it reads as written for them, without you starting from a blank page.
Go here: Responding to Enquiries and RFPs: The Venue Side, especially personalising properly, and Writing That Sells for the engaging part.
"Making agendas for the same meetings every week"
Save the agenda structure once in a project, feed it last week's notes and actions, and have it produce this week's agenda with outstanding items carried forward. Put it on a schedule and it'll be waiting for you on Monday.
Go here: the "Draft a meeting agenda" and "Prepare materials before meeting" prompts in Automate the Boring Stuff, then Skills, Scheduled Tasks and Loops to make it run on its own.
7. Don't know where to start
I saved these for last on purpose, because they're the most important. If you can feel time being wasted but can't work out which bit to tackle first, you're in the majority, not behind.
"Not really sure where to start with it all!"
Start with one task you did more than twice last week. Not a strategy, not a tool comparison. One boring, repeated task, done with AI once, this week. Confidence comes from a single win.
Go here: the New to AI? Start Here path, and AI for Beginners: Your First 30 Minutes.
"Don't know where to start in how ai can help my business, how do I ask the right questions?"
This is exactly why I gave away the Friction Audit prompt in session two. It flips the usual thing round: instead of you asking AI questions, it interviews you about your week. Fifteen minutes later you've got a map of where your time actually goes, and the three things worth fixing first.
Go here: The Friction Audit: Let AI Interview You for Once, then Can AI Help? A Decision Framework and Prompting Like a Pro.
"Identifying sponsorship leads"
AI is good at the research legwork: who sponsors similar events, who's targeting your audience, who's just launched something relevant. It's bad at knowing who'll actually pay, so treat its list as a starting point to qualify, never a mailing list.
Go here: prospecting without spraying in Sponsorship and Exhibitor Sales with AI, and Perplexity: AI Search That Cites Its Sources for research you can check.
What happens next
That's all 37. Some of you got a course, some got a tool, and a few got an honest "AI isn't the fix here, but here's what is". Where we didn't have a good enough answer, we've written a new one, fourteen of them in all, each with a route for whatever tools you already have.
The point of session one was never really the software. It was getting you to describe what's broken, clearly, because that's the whole bottleneck. You've already done the hard part.
Three things you can do this week:
1. Sign up to the hub. It's free, and you'll need an account to read the full answers and copy the prompts.
2. Run the Friction Audit. Fifteen minutes, and you'll know which problem to fix first.
3. Post your next problem on the Idea Board. Anyone can post, everyone can vote, and the ones that get voted up are the ones that get built.
Thank you to everyone who scanned the code, typed something honest into their phone and stuck around for all three sessions. And thank you to the team at CHS for letting me build software on a show floor with no idea what it would be.
See you at the next one. Bring a problem.