If you are thinking about switching trust accounting or trust administration software, the hardest part is not comparing products. It is knowing what to compare.
Most evaluations start with a feature grid. Vendors across the top, capabilities down the side, and whoever collects the most checkmarks wins. It feels rigorous. It almost never predicts how the software feels in year two.
What does? The questions you ask before you sign.
This post wraps up a series where we looked at the decision from four angles: building your own tools with AI, assembling a stack of specialist platforms, comparing purpose-built platforms for corporate trustees and individual trustees, and the risk of relying on a few key people to hold it all together.
Below are the seven questions you should be asking if you are re-evaluating your trust software. And if you want to go deeper, the Fiduciary Software Buyer’s Scorecard walks through each criterion and how to weigh it for your firm.
What are your options: Build, Stack, or Buy?
Building your own can make sense.
AI coding tools have made the first version of internal software genuinely cheap. You describe what you want and AI writes it. Firms in other industries have replaced six-figure CRM contracts with apps that cost a few hundred dollars a year to run.
The first version was never the expensive part.
The expensive part is the rounding behavior nobody specified, the permissions gap nobody tested, the backup routine that was never actually running, and the week your operating platform becomes unreadable because the person who wrote it took another job. In most businesses those are disruptions. In a fiduciary practice they are potential breaches of duty, and nobody reviewing a breach grades on effort.
Building makes sense when your firm runs specialized workflows no vendor serves and you have an engineering team that can own the software for as long as your firm exists.
A specialist stack can make sense.
This is the most common setup in the market, and the logic behind it is sound. Use the strongest tool for each job. A document system for version history and retention. A CRM for referral sources and attorneys. A reporting tool for leadership visibility. A trust accounting system for the books.
It works in other industries because the connective tissue already exists. A customer, an invoice, a payment, a ticket. Those objects mean roughly the same thing from one system to the next, so integrations get built once and reused by thousands of firms.
Fiduciary operations never got that foundation. A beneficiary is a contact in your CRM, a payee in your accounting system, and a folder name in your document system. A distribution is a task in one place, a ledger entry in another, and an email approval in a third. Something has to translate between them, and that something is usually a person at your firm.
The difference between a stack that works and one that does not is whether you are managing it as infrastructure or holding it together with institutional memory. A stitched stack with a named owner for every integration, a defined system of record for every critical field, and documented controls at every handoff is a legitimate operating model. A stitched stack held together by “ask them how we do that” is a single-person dependency wearing a software budget.
An all-in-one platform can make sense.
It fits when the fiduciary work itself is the business, when the same facts keep getting entered in more than one place, and when you want the ledger, the administration record, the approvals, and the billing to be the same system rather than four systems and a person in the middle.
There are still some trade-offs. Migrating years of history takes real work. You will adapt some workflows to the platform instead of the other way around. And an all-in-one system may not match a dedicated CRM or document tool in every corner. The real question is whether those trade-offs cost less than the translation work your team is doing today.
Whichever path you lean toward, the same seven questions will tell you whether it holds up in year two.
What are the seven questions you should ask any vendor?
Ask these of every vendor you are considering. Ask them of whoever is proposing to build something in-house. And ask them of the system you are on today.
1. For any given fact about a trust, which system is the record?
Pick one specific fact and ask about it. A beneficiary’s mailing address. The current fee schedule. The date of the last annual review.
What you want is a single named system, and a clear account of what happens everywhere else when that fact changes. What you do not want is “it syncs.” Sync describes a direction of travel, not an owner. Ask what happens when the two sides disagree, because eventually they will, and the trustee owns the answer no matter which vendor’s software created the discrepancy.
A useful version of this test costs you nothing: ask two people on your own team the same question separately and see whether you get the same answer.
2. Can you show me principal and income on a single transaction?
Not on a slide. In the product, on a real receipt, with the rule that decided the classification visible on screen.
This is the fastest test in the whole evaluation, because it separates software built around fiduciary objects from software built around generic ones. Either the classification happens as the transaction posts, under rules you can configure, or it happens later, by hand, in a month-end cleanup that somebody has to remember to do.
If the answer involves a class, a sub-account, a tag, or a spreadsheet, you have learned where the accounting actually lives.
3. What data will my team still have to manually enter?
Every demo shows the system full. Ask how it got full.
The answer you want is specific in both directions: which custodians and banks feed automatically, how often they sync, and which ones still require a file import or manual entry. “We integrate with everything” is not an answer. Push until you get a number and a list of exceptions.
This question maps more directly onto hours than any other one on this list. It’s the difference between a team that spends month-end reviewing numbers and a team that spends it entering them.
4. Three years from now, a beneficiary’s attorney asks who approved a distribution and why. What do you show them?
Your duty runs to the beneficiary, and it does not adjust for the fact that the person who made the decision no longer works there. “The prior administrator had a good reason” is not an answer you can give a court.
What you want to see is the request, the review, the approval, the supporting document, the payment, and the resulting accounting entry, in one place, with timestamps and names attached. Reconstructing that from three audit logs, a folder, and an email thread is possible. You will just be doing it under scrutiny, after something has already gone wrong.
Then ask the follow-up that matters more. Does the system capture the reasoning, or only the transaction? It’s easier to document why a distribution was discretionary when the note sits next to the distribution. When it lives in a separate document someone has to remember to open, it often never gets written.
5. How does your AI handle client data, and how does it ensure accuracy, confidentiality, and security?
Ask this of every vendor now, including the ones whose AI you are excited about.
Four things, and a vendor should answer all of them without flinching. Which models touch client data. Where that data is processed. Whether it is retained. Whether it trains anyone else’s system. Then one more: where does a human sit in the workflow before AI output becomes part of the record?
This is also the hardest question to answer about software you build yourself. Deciding which models see beneficiary data and under what terms is a job that security teams at large vendors spend careers on.
6. Who are you not built for?
The honesty test, and the one that tells you the most.
Vendors who have made a real architectural choice know exactly who they gave up, and they will tell you. Vendors who say they serve everyone have usually traded away the specific thing you need, and you will pay for that later in configuration time, consulting fees, or headcount by teaching a general-purpose tool how trusts work.
7. What does year three look like, and how do I leave?
Even building your own requires an answer. The build takes months. The ownership lasts as long as your firm does.
Ask who maintains the system and what happens when the person who knows it best leaves. Ask what the release cadence actually looked like over the last two quarters, not what the roadmap slide promises. Then ask how you get your data out, in what format, on what timeline, if you decide to leave.
A vendor confident in the product answers the exit question immediately and specifically.
The firms leaving their current stack are not wrong that something is broken. Paying for software that does part of the job, then paying again in hours to connect the rest, is a bad deal whether the software is generic, stitched together, or written in-house.
But the fix is not finding the longest feature list. It is knowing which seven answers you cannot do without.
To take a deeper dive on these seven questions and how to weigh them based on your firm’s unique needs, download the Fiduciary Software Buyer’s Scorecard here.
Schedule a call today to see how ProTrustee would work for your firm’s unique needs.