Mortgage broker technology ROI

Software vendors selling into the intermediary market are fond of return-on-investment claims, and almost all of them are built on assumptions the vendor chose. The only figure worth acting on is one you calculate yourself, using your own numbers.

This is the arithmetic, in the order you need it.

Establish the true cost, not the headline price

The subscription is rarely the whole cost. Build a full annual figure covering:

• Licence fees, per seat and per month, multiplied by the number of people who will genuinely need access including administrators and paraplanners.
• Implementation, configuration and any setup charge.
• Data migration, whether you pay a supplier or absorb the internal hours.
• Training, counted as hours multiplied by a loaded hourly cost rather than as a course fee.
• Integration work, including anything a third party has to build.
• The lost productivity of the transition period, which for a case management change is usually several weeks of slower work.
• The cost of the system you are keeping in parallel until the old one is switched off.

Add these into a first-year cost and a steady-state annual cost. The gap between the two is often large, and comparing a vendor's annual price against your existing annual price ignores it entirely.

Value the time honestly

The usual justification is time saved, and the usual mistake is to price that time at an adviser's hourly value while the time saved is actually administrative.

Work out a loaded hourly cost for each role: salary plus employer national insurance, pension, software, workspace and management overhead, divided by realistic productive hours in a year. Productive hours are not contracted hours; take out holiday, sickness, training and the general friction of a working week.

Then measure the task before you change it. Time an administrator doing the rekeying for three days and take the average. Count how many times a week someone asks where a case has got to. Log how long a fact find takes now. You cannot claim a saving against a baseline you never recorded, and a vendor's estimate of your current cost is not evidence.

The two ways time converts into money

This is the distinction most calculations skip, and it decides whether the investment is real.

Saved time only becomes a return if it is either removed from the cost base or refilled with revenue-generating work. Two administrative hours a day saved across a team is worthless if everyone simply goes home slightly less exhausted. It becomes real if it lets you take on more cases without hiring, or lets you delay a hire you were about to make, or lets an adviser hold more appointments.

So write the conversion explicitly. If the saving is capacity, state how many additional cases per adviser per month it supports and whether you have the enquiry flow to fill them. If you do not have the demand, the tool has not produced a return, it has produced slack, and the constraint is in marketing rather than operations.

Revenue effects, calculated conservatively

Some tools claim to increase income rather than reduce cost. Those claims are testable.

A retention or maturity system should show up as a higher proportion of eligible clients refinanced through you. Calculate the base rate first: of the clients whose products ended in the last twelve months, what percentage did you actually refinance? A tool that moves that figure by even a few percentage points on a book of several hundred clients is straightforwardly valuable, and you can measure it.

A lead response tool should show up as a higher proportion of enquiries converting to booked appointments. Again, measure the current figure before buying.

Be sceptical of anything that promises more leads without changing what you do. And do not model revenue increases at the vendor's suggested rates; model them at half, and see whether the case still holds.

Payback and the comparison that matters

With a first-year cost and an annual benefit, payback period is the total first-year cost divided by the monthly net benefit. Under eighteen months is generally a comfortable case for a small firm. Over three years, the market will have changed before you break even.

Then run the comparison people forget: what does doing nothing cost, and what does the cheapest alternative cost? Sometimes the honest answer is that a better-configured version of your current system, or one part-time administrator, delivers most of the benefit for a fraction of the outlay. A vendor will never present that option and it is frequently the right one.

Specific cautions for AI tools

Priced by usage rather than per seat, AI tools have costs that scale with volume, so model your busiest quarter rather than your average one.

Two hidden costs recur. Checking time: if an adviser has to read and correct every generated summary or draft, the net saving is the drafting time minus the checking time, which is a much smaller number than the demonstration suggested. And error cost: a mistake in a suitability record or a client communication is expensive in a way that a mistake in a marketing email is not.

Structure the calculation so the human review point is included rather than assumed away. Advice remains a regulated judgement and the review cannot be removed to improve the numbers, so any model that only works without it does not work.

Review it against the baseline you recorded

Put a date in the diary at six and twelve months to measure the same tasks you measured at the start, using the same method.

Expect the honest outcomes to be mixed: some savings will be larger than modelled and some will have evaporated because a workflow changed. What you are looking for is whether the specific number you built the case on has moved. If administrators were spending ninety minutes a day rekeying and now spend seventy, you have most of a return. If they spend ninety minutes doing something else that is equally unproductive, you bought an interface.