Feature lists are close to useless for choosing between serious platforms, because at this point most of them have most of the features. The two things that decide whether you regret the purchase are whether the integrations hold up under real load, and how good the vendor’s support is once you have already paid. Neither appears in a demo.
I run a company as well as advising on them. At PhoneStack I have looked at well over a hundred platforms across AI, marketing, email, content, CRM and dialling, and picked the handful that fit. That is not a boast, it is the reason I have opinions about this: I have made these calls with my own money often enough to have got some of them wrong.
Here is the one that cost me most, and what it taught me to check first.
I bought it for the integration. The integration was what broke.
One disclosure before the story, because it is the obvious objection: I am co-founder of a company that sells an AI phone service, so I am writing about buying a dialler while being in the business of phone software. I have named no product here and Blue Harbour sells none. Read it with that in mind.
We needed a dialler. There was an obvious candidate: one built into the email platform we already used. One system, email and calls running together in a single sequence, no glue code, no second vendor, no data sitting in two places disagreeing with itself.
In practice it broke constantly. And when it broke, support was slow enough that I could never get it working properly. I killed it and we built our own instead.
Building our own was available to me because we build software. It will not be available to you, and it should not need to be. The realistic version of the same move is smaller: run the two systems separately and accept the manual step between them, which is what the integration was supposed to remove. Worse than the promise, better than paying for something that does not work.
I chose it for how well it connected. Connecting was the thing that failed.
Why demos cannot show you either one
A demo is a controlled environment running clean sample data at low volume, driven by someone who uses the product every day. Nothing about that resembles your Tuesday afternoon.
Integration reliability
“Native integration” means the vendor built the connection themselves. It says nothing about whether it holds. Native means a connection exists. It does not tell you whether the API holds at your volume, what happens when it fails, whether it fails loudly or silently, or how long it takes to notice. A sync that quietly drops records for three weeks is worse than no sync at all, because you will trust the data in the meantime.
Support quality
This is the one nobody scores and everybody eventually cares about, and it is scoreable. Ask for the service level agreement in writing and read three things in it: the response time for a system-down issue, whether that clock runs outside business hours, and what happens when they miss it. A vendor whose SLA has no consequence attached has not committed to anything. Ask separately how support is staffed, because “24/7” and “someone answers at 3am who can actually fix it” are different products.
Software vendor evaluation: what to check instead
Six questions that find what the demo hides
- Ask for the integration to be demonstrated failing. Not working, failing. What does the product do when the connection drops? Does it queue, alert, retry, or silently lose the record? If nobody can answer, assume the worst one.
- Ask what their support response times are, in writing. Not “we pride ourselves on support”. Hours, in a contract, with what counts as a business day and what happens outside one.
- Ask to speak to a customer of roughly your size who has had a problem. Reference calls with delighted customers tell you nothing. A customer who had a bad week and stayed tells you everything.
- Run one real workflow end to end during the trial. Your actual data, your actual volume, your actual edge case. The one weird thing your business does is the thing that will break, and it is never in the sample data.
- Find out who owns the integration. Sometimes the vendor built it. Sometimes a partner did. Sometimes it is a third-party connector both of them will blame. Knowing who fixes it before you need it fixed is worth more than the feature itself.
- Ask what it takes to leave. Can you get your data out, in what format, and how quickly? A vendor confident in the product answers this easily. The reaction to the question is informative on its own.
The wider point about feature lists
Serious platforms in a mature category converge. By the time you are comparing two credible options, they will both do most of what you need, and the comparison chart in the sales deck exists to highlight the handful of places where one of them is ahead this quarter.
Those differences are usually the least durable thing about the decision. The other vendor ships the missing feature next year. What does not change on that timescale is whether the company answers the phone, and whether the plumbing holds.
So weight the boring criteria far more heavily than feels natural. Reliability, support, data portability, and who is accountable when something breaks. Those are what you live with for the next five years. The feature comparison is what you feel clever about for the first five weeks.
What I do now
I select on whether tools connect to each other, not on what each one does alone. An owner-led business does not have a systems integrator on staff, so every tool that does not talk to the others becomes somebody’s manual job forever. That is the cost that never appears on the invoice and never goes away.
And I assume I will be wrong about something. I had not asked question one on that list, which is the one that would have caught it: I never asked what happened when the integration failed, because it had not occurred to me that the reason I was buying it was the thing that would break. That is why the list exists.
The questions narrow the odds. They do not eliminate them, so the second job is making a wrong answer cheap: shorter commitments, exportable data, and knowing before you sign what it takes to reverse the decision.
If there is a shortlist on your desk
Blue Harbour Navigator is a fixed-fee independent assessment for owner-led businesses. Seven to ten days, and you get a read on which problem is worth solving first and what would genuinely fix it. Sometimes that is a platform. Often it is not.
I sell no software and I do not implement it, and I take no commissions from any vendor. $5,000 flat, in your own currency.
Book a 20-minute fit call Or tell me what you’re facing, no call needed
Two minutes to write, I read them myself. Or see how Navigator works first.
Sources. This one is entirely first-hand. The dialler decision was made and reversed at PhoneStack in 2026. No vendor is named, deliberately: the product has since had time to improve, the lesson does not depend on which one it was, and a page that takes no vendor money should not be settling scores with vendors either.
Trevor Jamieson is the founder of Blue Harbour Solutions in Halifax, Nova Scotia. He has spent twenty years across the vendor and buyer sides of technology, at Dayforce, The Hercules Group, TELUS and Xerox. He currently holds a contract role in partner marketing at Amazon Web Services, and is co-founder and CEO of PhoneStack. Neither pays Blue Harbour anything inside a client assessment, and both are declared in writing on the site. Blue Harbour sells no software and takes no commissions from any vendor.