The demo always works. Break it before you sign.

Nobody has ever seen a bad demo.

Think about it. You’ve sat through pitches for CRMs, job systems, booking platforms, AI tools, whole ERPs. The mouse glides, the data is tidy, the numbers add up, the salesperson never once hits an error. Then you buy the thing, load your own records, and within a week you’re staring at a spinning wheel wondering where the smooth product from the meeting went.

The demo wasn’t lying, exactly. It was rehearsed. Someone chose the account, the search term, the invoice, the customer record. They chose the one path through the software that works beautifully, and they walked you down it. That’s not fraud. It’s sales. Your job as the buyer is to get off the path.

A demo is a magic trick, and the deck is stacked

A good salesperson runs the same demo forty times a month. They know exactly which button lags, which report takes eight seconds to load, which import chokes on a comma in the wrong place. So they route around all of it. The account they pick has clean data. The customer has one address, not three. The date range returns twelve rows, not forty thousand.

None of that matches your business. Your data is a mess, because everyone’s data is a mess. You’ve got duplicate customers, a supplier whose name is spelt three ways, an invoice from 2019 that never closed, a phone number field with “call the office, ask for Dave” typed into it. The demo never meets any of that. Which is exactly why the demo isn’t evidence.

The question isn’t “does the software work?” It’s “does the software work when I do the thing I actually do, with the data I actually have?” Those are different questions, and only one of them matters after the invoice clears.

Make them drive with your data, not theirs

The single most useful move in any software evaluation is also the simplest. Send the vendor a slice of your real data before the second meeting, and ask them to load it and run your workflow live.

Not a sample set they’ve pre-cleaned. Not a “sample import” they’ve tested overnight. Your export, warts and all. A few hundred real customers, a batch of real invoices, a month of real jobs. Watch what happens when it hits their system in front of you.

You’ll learn more in ten minutes than in ten glossy slides. Does the import map your fields, or does it drop the ones that don’t fit? What happens to the customer with no email? Does the duplicate get merged, flagged, or silently doubled? When you search for a name with an apostrophe, does it come back? These are the moments the standard demo is built to hide, and they’re the moments you’ll live with every day.

If a vendor won’t touch your data until you’ve signed, that tells you something. Sometimes it’s a fair caution about privacy, and you can solve that with a sample set and an NDA. Sometimes it’s because they already know what happens, and they’d rather you found out after the deposit.

The questions that make salespeople sweat

Anyone can answer “does it do invoicing?” The useful questions are the ones the script doesn’t cover. Ask these, and watch whether you get a straight answer or a redirect.

• Show me it failing. Ask them to trigger an error on purpose. What does the software do when someone fudges a required field, loses connection mid-save, or uploads the wrong file? A product with dignity handles failure gracefully. A fragile one shows you a stack trace or eats the data.
• Do that again, but slower, and let me drive. Take the mouse. The demo path is muscle memory for them and a maze for you. If you can’t fumble your way to the same result in a couple of minutes, your staff won’t either, and adoption is where most software quietly dies.
• How do I get my data back out? Not “can I export.” Everyone says yes to that. Ask to see the export, right now. Ask what format, whether it includes attachments and history, and whether you can do it yourself without lodging a support ticket. This is the vendor lock-in question, and it’s cheapest to answer before you’re a customer.
• What happens at 200 users, or 50,000 records, or end of month? The demo dataset is small. Your fifth year won’t be. Ask what the same screen looks like with a real volume of data behind it, and whether they can show you a live account that size.
• Who actually built this, and can I talk to them? Plenty of tools are a thin skin over someone else’s platform. That’s not automatically bad, but you deserve to know whether the people selling it can fix it.

You’re not being difficult. You’re doing the job the demo was designed to save you from doing.

Reference calls, done properly

Every vendor has a list of happy customers ready to say nice things. Call them anyway, but ask better questions than “are you happy with it?”

Ask what broke. Ask what the implementation actually took, in weeks, versus what they were told. Ask what they’d do differently. Ask about support: when something went wrong at a bad time, how long until a human who could help actually helped? Ask the question that separates a reference from a friend of the sales team: “If you were choosing again today, would you pick this?” The pause before the answer is often the answer.

Better still, ask to speak to a customer who left, or one who’s a similar size and industry to you rather than the flagship logo. The vendor won’t offer that. You can ask.

The demo is theatre. The trial is the truth.

Wherever you can, don’t buy off a demo at all. Buy off a trial where your own people do their own work for a fortnight.

Not a sales engineer configuring it for you. Your staff, your data, your Tuesday. That’s when you find out that the report everyone needs takes six clicks, that the mobile app logs you out on the ute, that the thing which looked effortless in the meeting takes a training session nobody has time for. A demo shows you the software at its best. A trial shows you the software as you’ll actually meet it.

If a trial isn’t on offer, a paid pilot with an agreed exit is the next best thing. Put a real but small piece of your business on it for a month, agree up front what “working” looks like, and agree what happens if it doesn’t. A vendor confident in their product will take that bet. One who won’t is telling you to trust the demo, and you already know what the demo is worth.

What we do differently, and what we’d tell you to watch for

We build custom business systems and integrate the tools businesses already run, so we’ve sat on both sides of this. When we show you something we’ve built, we’ll run it with your worst-case in mind, not our best-case. And when you’re weighing up an off-the-shelf product against something custom, the demo test cuts both ways: if we can’t show you the thing working the way you work, that’s on us to prove, not on you to assume.

The pattern to remember is short. The demo is choreographed. Your data isn’t. Get your data into the software, take the mouse, ask to see it fail, and don’t confuse a smooth performance with a product that fits.

If you’re mid-way through choosing a system and the demos are starting to blur, send us the shortlist and the workflow you’re trying to fix. We’ll tell you which questions to ask each vendor, and which answers should end the meeting.