AI business cases can become complicated very quickly.
They don't always need to be.
Before discussing models, platforms, integrations or licence fees, there is a much simpler number to establish:
What does this process cost us today?
THE BASELINE COMES BEFORE THE BUSINESS CASE.
If you cannot answer that, it is difficult to know what an improvement is worth.
01 / BASELINE
Start with the baseline
Take a recurring process and establish four numbers:
Time per occurrence × frequency × people involved × relevant cost
Imagine a professional process takes five hours and happens 200 times each year. That is 1,000 hours per year.
If a new way of working reduces the process to one hour, the annual effort falls to 200 hours.
CURRENT PROCESS
5.0 H× 200 / YEAR= 1,000 H / YEAR
IMPROVED PROCESS
1.0 H× 200 / YEAR= 200 H / YEAR
CAPACITY RELEASED
1,000 H − 200 H
800 H / YEAR
Annual value depends on both time per task and frequency.
The difference is 800 hours of professional capacity released each year.
That's the beginning of the business case. Not the end.
02 / CAPACITY
Time saved is not automatically money saved
This is where ROI calculations often become misleading.
If a solution releases 800 hours, it does not necessarily mean the company has put the cash equivalent of 800 hours into its bank account.
You need to ask what happens to that capacity. There are several possibilities.
CAPACITY RELEASED
800 H
ILLUSTRATIVE- More fee-earning work
- Avoided / delayed recruitment
- Faster turnaround
- Less rework
- More time for professional work
Possible uses, not five benefits to add together. Commercial value depends on what the organisation can actually do with the released capacity.
More fee-earning capacity
In a professional services business, released time may allow people to undertake more client work.
If demand exists and that capacity can genuinely be sold, the commercial value could be significant.
Avoided recruitment
A growing team may be approaching the point where another hire is required simply to cope with workload.
Increasing the capacity of the existing team could delay or remove that requirement. That has a different but very real financial value.
Faster turnaround
Some improvements create value through speed rather than direct cost reduction.
A report delivered in a day instead of a week may improve client experience, shorten project timelines or allow a business to handle more concurrent work.
Better consistency
Reducing repetitive manual activity may also reduce omissions, formatting inconsistencies and rework.
Those benefits are harder to represent in a simple ROI percentage, but they should still be understood.
A better use of expensive expertise
Sometimes the benefit is simply that a £60,000 professional is spending less of their week on work that doesn't require a £60,000 professional.
That matters even when the saving cannot immediately be converted into revenue.
Time saved is a measurement.
Value created is a business outcome.
03 / VALUE OF TIME
Use the right value for an hour
This is another common source of inflated business cases.
There are several different ways to value time: salary cost, fully loaded employment cost, charge-out rate, and realisable revenue are not the same thing.
- Salary cost
- ≠Loaded employment cost
- ≠Charge-out rate
- ≠Realisable revenue
The right measure depends on what actually happens to the time released.
If an hour is released, you cannot automatically value it at the full client charge-out rate unless there is a realistic route to selling that hour.
For internal efficiency calculations, loaded employment cost may be more appropriate.
For a capacity-constrained consultancy with more demand than it can currently serve, recoverable fee-earning capacity may be relevant.
The assumptions should be explicit.
“A conservative business case that survives scrutiny is far more useful than a spectacular ROI number nobody believes.”
04 / COST OF OWNERSHIP
Include the full cost of the solution
Now look at the other side of the equation.
Depending on the solution, costs might include:
- Software or platform fees
- Design and implementation
- Integrations
- Data preparation
- Internal staff time
- Training and adoption
- Ongoing support
- Maintenance and improvement
- Governance and security requirements
This gives you a more useful calculation:
Measurable annual benefit
− Annualised cost of ownership
= Net annual benefit
Start with the current process baseline: time per occurrence, frequency, people involved and the appropriate value of that time. Then compare the measurable annual benefit of the improved process against the total cost of ownership.
From there you can calculate payback period and ROI if those measures are useful for the investment decision.
But there is an even more important step.
05 / SUCCESS CRITERIA
Define success before you build
Suppose a process currently takes five hours.
Before developing anything, agree what the new process would need to achieve to justify proceeding.
Perhaps reducing five hours to four would not be enough. Reducing it below two hours might be.
That becomes part of the success criteria.
You can do the same with quality, accuracy, human review, security or other important measures.
- 01 / BASELINE5.0 H
- 02 / SUCCESS CRITERIA< 2.0 H
- 03 / PROOF OF CONCEPT
Test the proposed improvement
- 04 / MEASURE RESULT
Compare against the agreed criteria
- 05 / DECISION
The measured result determines the next step. Neither outcome is pre-selected.
Then a proof of concept has something objective to prove. It is no longer “Look what the AI can do.”
It becomes “Did this improve the process enough to justify further investment?”
Those are very different questions.
WRONG QUESTION
“Can the AI do it?”
BETTER QUESTION
“Did this improve the process enough to justify the investment?”
Proceed when the measured improvement meets pre-agreed success criteria and the resulting value justifies the implementation and ongoing cost at an acceptable level of risk.
06 / FROM OUR WORK
An example from our work
We have seen why baselining matters in our own work with property professionals.
On one Schedule of Condition process, the existing workflow required around ten hours of professional time per report. The redesigned process reduced that to 30 minutes while keeping the professional responsible for reviewing and approving the output.
- BEFORE / PER REPORT
- 10H
- WITH GO LEFT / PER REPORT
- 0.5H
- ANNUAL CAPACITY RELEASED
- Around800H / YEAR
Professional review and approval remain part of the process.
At the client's annual volume, that translated into around 800 hours of potential professional capacity.
The important part was not simply that the technology worked. It was that there was an existing process against which the new approach could be measured.
Without that baseline, “faster” is just an impression. With it, you can have a commercial conversation.
07 / INVESTMENT DECISION
What if the business case isn't strong enough?
Then don't proceed.
This is an important part of sensible AI adoption.
An opportunity can be technically possible without being commercially worthwhile.
A proof of concept can work without producing enough improvement to justify deployment.
An existing tool may solve 80% of the problem at a fraction of the cost of building something bespoke.
Those are successful conclusions too.
A GOOD PROCESS CAN END WITH
Both are useful decisions.
Do not proceed when the measurable improvement is too small, the commercial value is insufficient, the risk is disproportionate, or a simpler, cheaper existing solution solves the problem adequately.
The objective should not be to get every AI idea into production.
The objective should be to allocate investment to the opportunities capable of producing meaningful value.
08 / BUSINESS CASE
A simple AI business-case template
Before approving an AI initiative, leadership should be able to answer:
GO LEFT / COMMERCIAL ASSESSMENT
AI business case
10 QUESTIONS BEFORE YOU INVEST
- Current state
- What happens today?
- Volume
- How often does it happen?
- Effort
- How much time does it consume?
- People
- Whose time does it consume?
- Problem
- What specifically are we trying to improve?
- Target
- What would the new process need to achieve?
- Value
- What happens to the capacity, cost or time released?
- Investment
- What will implementation and ongoing operation cost?
- Risk
- What needs to remain human-reviewed or otherwise controlled?
- Evidence
- How will we prove whether it worked?
If those answers are unclear, the business probably isn't ready to approve the solution yet.
That is why measurement sits at the heart of our BUILD™ methodology.