Nobody’s data is messy in a software demo. Every customer has one record. Every employee follows the process. Every integration works.
Then Monday morning happens.
Your best salesperson keeps notes on their phone. A customer is in the system three times. The person who knows how billing works is on vacation. Someone changes an appointment after the schedule has already gone out.
That’s the business your new software has to support. When I evaluate a technology vendor, I want to see what happens there.
Start with the problem you would pay to make disappear
“We need a new CRM” isn’t specific enough to guide a purchase. “We lose track of inquiries when they move from the office to sales” gives you something to test.
Write down the work that’s causing trouble, who does it, and what a better result would look like. Bring in the employee who will use the system every day. If the person buying it and the person living in it never compare notes, you’re taking a pretty expensive guess.
Also ask whether the tool you already own can do the job. A missing feature and an unused feature can feel exactly the same when nobody has time to investigate. Before you replace the platform, find out which problem you have.
Ask the vendor to show you a bad Tuesday
Give each finalist the same ordinary workflow, with a few ordinary complications. Use made-up or appropriately protected sample records.
For example: a prospect fills out your website form and asks for a callback next Thursday. The assigned salesperson is out that day. The prospect already exists under a slightly different company name. They call the office before anyone follows up.
Now ask the vendor to show you the whole thing. Where does the inquiry land? Who owns it? Who covers the absence? Can the office see the original request? What prevents two people from calling with different information?
Then have one of your employees try the workflow. Watch where they hesitate. A process that looks effortless with a practiced salesperson driving may be awkward for the person who has twenty other things to do.
“Yes, we can do that” should start a demonstration. Ask whether it works in the package you’re buying, needs an add-on, requires custom work, or is planned for a future release. Those are four different answers.
Pay attention to the word “integration”
Two products appearing on the same slide doesn’t tell you how they work together.
Suppose the vendor says the platform connects to your accounting system. Ask them to create a customer, update that customer’s details, and send an approved invoice through. Where do changes appear? How long do they take? What happens if the connection fails? Who sees the error?
Find out who maintains the connection and who you call when each vendor says the problem belongs to the other one.
Apply the same scrutiny to your old records. Which notes, attachments, and history will move? Who cleans duplicates? Who checks the import? A migration isn’t finished just because a progress bar reaches 100 percent.
Have them demonstrate permissions, too. Can employees see only what they need? Can you remove access promptly? Ask how your information is stored, retained, exported, and used by any AI features. Get answers specific to the product and package you’re considering.
Price the whole first year
A monthly subscription is easy to compare. Getting the business ready to use it takes a little more digging.
For a simple hypothetical, software priced at $500 a month costs $6,000 a year. Add $3,000 for setup, $2,000 for migration, and $1,000 for training, and the first-year commitment is already $12,000. Your employees’ time, integrations, and extra usage charges could add more. These are example numbers, not market prices.
Ask every finalist to price the same users, work, support, and expected usage. Separate one-time costs from recurring charges. Include what it takes to get your information back if you leave.
Then ask what improvement would make that cost worthwhile. If you’re buying it to reduce duplicate entry, measure the duplicate entry. If you’re buying it to improve follow-up, check whether the team can actually follow up more reliably during the pilot.
A lower subscription price doesn’t help much if the missing pieces become your team’s second job.
Someone on your side needs to own the rollout
Even a good product needs decisions: how the process will work, which records matter, who gets trained, and when the old system can be retired.
Name one person on your side to coordinate those decisions. Give them time to do it. A project owner who is already fully booked can’t make an implementation happen by being copied on every email.
Before committing, get a written scope that identifies deliverables, responsibilities, timing, additional charges, and what counts as completed work. Clarify support and renewal terms. Ask who takes over after the salesperson hands you off.
I would also ask for a conversation with a customer whose operation resembles yours. “How did implementation go?” is a start. “What did your team have to do that you weren’t expecting?” is more revealing.
A short list to take into your next demo
You don’t need fifty questions to change the quality of the conversation. Start with these five:
1. Can our employee complete our most important workflow? Include an exception, not just the ideal path.
2. What’s included, and what still needs to be built or bought? Get the distinction in writing.
3. Who owns migration, training, and problems after launch? Name people or accountable teams.
4. What will the first year really cost? Include your own team’s effort.
5. What must the pilot prove before we move ahead? Agree on success before everyone falls in love with the product.
A vendor who can discuss limitations clearly gives you something useful to work with. I’d rather hear “That requires an extra step; let me show you” than discover the extra step after signing.
Left Brain provides Technology & Digital Solutions, including vendor advocacy and implementation oversight. If you’re comparing systems or trying to untangle a rollout, start with a free discovery call. We can talk through the decision and where you need support.
Evaluating AI features? Read AI Can Draft the Email. Who Should Hit Send? for the approval questions worth asking.