How to Calculate Software ROI for a Small Shop
Vendor return-on-investment claims are unusable, not because they are dishonest but because they come without methodology and assume nothing goes wrong. Here is how to build a figure you can actually defend.
Four returns can be measured honestly: faster collection, more jobs per day, recovered agreement renewals, and recovered unsold estimates. Everything else is real but not quantifiable, and should be described rather than costed.
And the calculation only works if you record a baseline first. You cannot compute a return without a before.
Why vendor ROI figures are not usable
You will see claims of large percentage improvements in scheduling time, drive time or revenue per technician. Three problems with all of them.
No methodology. Measured how, over what period, against what baseline, at how many shops? Without that, a percentage is a marketing figure rather than a finding.
Survivorship. The businesses producing the case studies are the ones for whom it worked. Shops that abandoned the platform in week three are not in the sample.
They assume clean data. Most of the quoted gains depend on accurate job durations and reliable status updates. A shop with neither will not see them, and most shops arriving at a platform have neither yet.
None of this means the returns are imaginary. It means you have to calculate your own.
Record the baseline first
This is the step that gets skipped and it makes everything afterwards unprovable. Spend an afternoon capturing six numbers before you change anything:
| Baseline metric | How to get it |
|---|---|
| Average days from job completion to payment received | Sample 30 invoices from last quarter |
| Average jobs completed per technician per day | Total jobs ÷ technician-days over a month |
| Maintenance agreements lost to non-renewal last year | Count. Most shops find this higher than expected |
| Value of estimates issued and not closed, last 90 days | Total the open and expired ones |
| Office hours per week on scheduling and invoicing admin | Ask, honestly, and do not round down |
| Overtime hours per month | Payroll |
Six numbers, an afternoon’s work, and without them you will spend the next two years unable to tell whether the platform paid for itself.
The four returns you can actually quantify
1. Faster collection
The clearest and the fastest to appear. Moving from paper invoicing to on-site payment compresses collection from weeks to the same day, as covered in how on-site payment capture works.
This is a one-off working capital release rather than recurring profit, and it should be described as such. If you invoice $120,000 a month on a four-week cycle, roughly a month of revenue is permanently outstanding. Collecting same-day frees most of that once.
Value it at what the money is worth to you — the interest on a line of credit you no longer need to draw, or simply the cash flow relief. Do not count it as annual profit.
2. More jobs per technician per day
The largest recurring return when it materialises, and the slowest to arrive. It comes from better routing, less time on the phone, and less admin per job.
Half an additional job per technician per day, across six technicians, at a $340 average ticket and 40% gross margin, is roughly $400 a day in additional gross profit. Over 250 working days that is a six-figure figure, which is why vendors lead with it.
Be conservative here. This gain depends entirely on accurate durations and reliable status updates, and it typically appears in months four to nine rather than immediately. Model a quarter of a job per technician, not a whole one, and treat anything more as upside.
3. Recovered agreement renewals
Straightforward arithmetic and quite reliable. Count agreements lost to lapse last year, multiply by your agreement fee, and assume automation recovers a meaningful share of them.
Forty lapsed agreements at $220 is $8,800 of annual recurring revenue. Recovering half is $4,400 a year, and it compounds because those customers stay in your service cycle — which also protects the replacement conversation described in service agreement management.
4. Recovered unsold estimates
Potentially the largest of the four and the hardest to be honest about. If you leave $250,000 a month in unsold proposals, systematic follow-up recovering even 3% is $7,500 a month.
The caution: this return depends on someone actually writing and running the follow-up sequence, not on the software existing. Attribute it to the process rather than the purchase, and only count it if you commit to the work — the detail is in why unsold estimates go cold.
What you should not put a number on
These are real benefits and quantifying them produces figures nobody believes, including you:
- Fewer “where is my technician?” calls. Genuinely the fastest visible improvement, and impossible to value honestly
- Reduced dispatcher stress
- Better customer experience — real, and it shows up in retention over years rather than in a spreadsheet
- Professional appearance
- Owner visibility — knowing what is happening without asking
Describe these in words alongside the calculation. A business case with four defensible numbers and five honestly-described qualitative benefits is far more credible than one with nine numbers, four of which are invented.
The shape of the return
Vendor ROI charts start at zero and rise. Real ones start below zero, because implementation absorbs staff hours and the first weeks run worse than before. Building the dip into your expectations is what stops a shop concluding at week three that the software failed.
A worked example
Six technicians, coming off paper, moving to a mid-tier platform.
Costs, year one
| Item | Amount |
|---|---|
| Subscription, $320/month | $3,840 |
| Implementation and setup | $1,200 |
| Internal hours, 90 at $35 | $3,150 |
| Hardware, readers and chargers | $900 |
| Total year one | $9,090 |
Returns, year one, deliberately conservative
| Return | Basis | Amount |
|---|---|---|
| Recovered renewals | 20 of 40 lapses at $220 | $4,400 |
| Recovered estimates | 2% of backlog, 40% margin | $14,400 |
| Additional jobs | 0.25/tech/day from month 6 | $12,750 |
| Admin hours saved | 5 hrs/week at $22 | $5,720 |
| Total year one | $37,270 |
Net year one: roughly $28,000, with breakeven somewhere around month four or five.
Those figures are constructed to show the method, not to predict your outcome. Every line depends on assumptions you have to replace with your own numbers.
Two of the four returns — recovered estimates and additional jobs — depend on work you have to actually do. Buying software does not deliver them. If you are not going to write the follow-up sequence or fix your durations, remove those lines from your calculation entirely, and the picture changes considerably.
Payback period is the more useful figure
Annual ROI percentages sound impressive and are easy to inflate. Payback period — how many months until cumulative return exceeds cumulative cost — is harder to argue with and more useful for a decision.
For a small shop coming off paper, four to eight months is a realistic range when on-site payment is part of the change. For a shop already on a competent platform, upgrading to something more capable, payback is considerably longer and harder to justify — which is itself useful information.
Frequently asked questions
Can I trust vendor ROI claims?
What is the first thing to do before calculating ROI?
Which return appears fastest?
Should I include things like reduced stress in the calculation?
What is a realistic payback period?
Why does the return start negative?
What to do next
- Capture the six baseline numbers this week, before changing anything. This is the step that makes everything else provable.
- Build your calculation with only the four quantifiable returns. Describe the rest in words.
- Remove any return that depends on work you will not do. Follow-up sequences and duration fixes do not happen by themselves.
- Use payback period rather than an ROI percentage when you present it to yourself or a lender.
- What HVAC software costs — the cost side of the calculation
- How on-site payment capture works — the fastest return
- Moving from paper to software — why the dip happens and how long it lasts
All figures in the worked example are constructed to demonstrate a method and are not predictions, benchmarks or measurements from any business. Every line depends on assumptions specific to your operation. Replace them with your own baseline data before drawing any conclusion, and treat returns that require ongoing process work as conditional on doing that work.