SquintPRO · Business case
Video analytics ROI calculator: work out the return with your own figures
Video analytics ROI is easy to promise and hard to check. Most business cases start from someone else's saving per camera, and fall apart when finance asks where the number came from. This page sets out a calculation you can defend: cameras in scope, times use cases per camera, times the hours of watching and review each one replaces, less the time spent checking alerts, set against the cost in your quote. Every input is yours, every assumption is visible, and the uncertain ones are measured in a pilot rather than guessed.
Published 10 min read
In short, for managers
- The method: hours saved = cameras × use cases × hours of watching each replaces, less the hours spent checking alerts. Value = hours saved × your loaded hourly cost, set against the licence, services and one-off costs.
- The inputs: all yours. Headcount and rates come from HR and finance, costs from the quote, and the two uncertain figures from a pilot on your cameras.
- What we do not give you: a standard saving per camera. Nobody can know it before your cameras and your shifts have been measured.
- Safety value: counted beside the money, as events found and response times, not converted into riyals unless your own records support it.
What a video analytics ROI calculation should measure
A camera network costs money every day and, on most sites, is watched only after something has gone wrong. Analytics changes that by watching each stream continuously and raising an alert with video evidence when a condition you define occurs. The return comes from three places, and an honest case keeps them apart:
- Time. Hours that people spend today watching screens, walking patrols to check zones, or scrolling through footage after an incident, which alerts and footage search can replace in part.
- Events seen. Conditions that happen today without anyone noticing: a person in a restricted zone at night, a missing hard hat, a vehicle in a walkway. These are counted, not priced.
- Avoided cost you already record. Only where your own records hold a cost per event, such as a repair, a delay or a claim, and only for events the pilot shows the system catches.
The calculator below prices the first, counts the second, and adds the third only from your own data.
The inputs, and where each figure comes from
Each input has a letter so the formula below can be copied into a spreadsheet.
| Input | Letter | Where your figure comes from |
|---|---|---|
| Camera streams in scope | C | The camera survey: only streams it qualifies for the use case count |
| Use cases per stream | U | Your priorities; a licensed stream can carry as many as the site needs |
| Hours of watching or review per stream, per use case, per month today | H | Rosters, patrol logs and incident-review records; zero if nobody does it |
| Share of those hours the alerts replace | R | Measured in the pilot, never assumed to be all of it |
| Hours spent checking alerts per month | V | Measured in the pilot from real alerts, true and false |
| Loaded hourly cost of the people involved | W | HR and finance: salary, allowances and overheads for that role |
| Yearly licence, support and operation cost | L | The quote, which is built per camera stream and deployment model |
| One-off cost | I | The quote: camera work, on-site hardware for on-premise, integration |
| Events found per month | E | Counted in the pilot against a person's review of sampled hours |
The formula, step by step
Five lines, using only the letters above, so finance can follow each one and change any input.
-
Hours replaced per month = C × U × H × R.
The watching and review time the alerts take over. If H is zero because nobody watches those cameras today, this line is zero, and the value sits in E instead.
-
Net hours per month = hours replaced − V.
Someone opens each alert and checks the clip; leaving V out overstates the saving.
-
Yearly value = net hours per month × 12 × W.
Hours become money only if they are used for other work or no longer paid for.
-
Net yearly value = yearly value − L.
The running cost of the licence, support and operation, from the quote.
-
Payback in months = I ÷ (net yearly value ÷ 12).
If the net yearly value is zero or negative on time alone, the case rests on E and on avoided costs, and should say so plainly.
Why use cases per camera matter more than camera count
SquintPRO is licensed per camera stream, with unlimited use cases on a licensed stream. In the formula, that means U raises the hours replaced without raising L: the camera that watches a loading bay for vehicles in the walkway can also watch for missing protective equipment and for people in the bay after hours. Adding cameras adds streams; adding use cases on a licensed stream does not add a licence.
Read more
Each use case still has to earn its own H and R: one that nobody would have watched for is worth nothing in hours. Start with the one or two that matter most, measure them, and add the rest once they show results on your footage. How the per-stream licence, deployment model and integration combine into the quote is set out in how a SquintPRO quote is built.
The worksheet: fill it with your own figures
Copy these lines into a spreadsheet or the message box of the form below. A blank line becomes a question for the survey or the pilot.
| Line | Calculation | How to fill it honestly |
|---|---|---|
| 1. Hours replaced per month | C × U × H × R | Use the survey's camera count, not the whole estate; take R from the pilot |
| 2. Net hours per month | Line 1 − V | Include the time spent on false alerts, not only true ones |
| 3. Yearly value | Line 2 × 12 × W | Count only hours that are redeployed or no longer paid for |
| 4. Net yearly value | Line 3 − L | Use the yearly figure in the quote for the deployment you chose |
| 5. Payback in months | I ÷ (line 4 ÷ 12) | Include camera work and hardware the survey found necessary |
| 6. Events found per month | E | Report beside the money, with the response time for each type |
| 7. Avoided cost per year | Your cost per event × events caught × 12 | Only from your own records, and only for events the pilot caught |
Lines 1 to 5 are the financial case. Lines 6 and 7 are the safety and operations case. Keep them as two totals so a reader can accept one without having to accept the other.
Safety monitoring ROI: when the value is not hours
On many industrial sites nobody watches the cameras in real time, so H is close to zero and the time case is thin. The value is in what is now seen: unsafe acts and near misses that used to go unrecorded, found with a clip, a time and a zone. That is real value, and it is measured differently.
- Count events by type and shift. Missing protective equipment at zone entry, people in forklift-only aisles, entry into restricted areas at night.
- Measure the response. The time from alert to someone acting on it, which depends on your routing and staffing, not only on the software.
Turning an avoided injury into riyals needs assumptions about severity and probability that finance rightly questions. Report the counts as they are, and add a money line only where your own incident costs support it.
Checks before the business case goes to finance
The usual reasons a video analytics case is sent back, and how to avoid each.
- R is taken as the whole of the watching time; use the pilot's measured share
- V is missing, so the time spent checking alerts is counted as saved
- Every camera on site is counted, including views the survey did not qualify
- Hours are priced even though nobody will be redeployed or paid less
- An event is counted twice, once as hours saved and again as avoided cost
- One-off costs leave out camera adjustment, lighting or on-site hardware
Where video analytics pays back, and where it does not
The calculation tends to work where there are many cameras, people already spend real hours watching or reviewing them, the conditions can be written as rules per zone and shift, and someone acts on each alert. It tends not to work where:
Read more
- nobody will act on the alerts. An alert that reaches no one returns nothing, whatever the detection rate;
- the cameras cannot see the condition. Distance, angle or night lighting rule some views out, which the camera survey for existing CCTV establishes before any money is committed;
- the goal is to identify people. SquintPRO detects what happened, not who it was; it does not use face recognition;
- the case rests on one rare event. A pilot cannot measure an incident that happens once in years.
How Interkey helps on Saudi sites: measurements, not estimates
Estates spread across Riyadh, Jeddah, Dammam and remote project sites often have many cameras and few people watching them, which makes E and footage search weigh more than live-monitoring hours. Sites that run night shifts need H and V measured on every shift, not only by day.
The deployment model also moves L and I. With on-premise deployment, SquintPRO runs on servers at your site and footage stays there, in the Kingdom, which usually adds processing hardware to I. The cloud option, with 3, 6 or 12 months' retention, moves that cost into the yearly line instead. Footage in which people can be identified can be personal data under the Personal Data Protection Law, overseen by the Saudi Data and AI Authority (SDAIA), so decide retention for policy reasons first and price it second. Interkey supports SquintPRO from Riyadh, with Arabic-speaking engineers who can sit with your HSE and finance teams while the figures are gathered.
Read more
Four steps replace the estimates in the worksheet with measured figures:
- Camera survey. Each camera is sorted, per use case, into usable as is, usable after adjustment, or needing new hardware. This fixes C and the camera part of I.
- Pilot with written criteria. One site, one or two use cases, success criteria agreed before switch-on, following how to scope a pilot that proves something. The pilot measures R, V and E.
- Quote. Built per camera stream and deployment model, with integration and any camera work listed. This fixes L and I.
- Decision. Proceed, extend or stop, with the worksheet filled from measured figures rather than estimates.
Questions buyers in Saudi Arabia ask about the business case
How do we calculate video analytics ROI for a site in Saudi Arabia?
Multiply the camera streams in scope by the use cases per stream and by the hours of watching or review each one replaces, subtract the hours spent checking alerts, and price the result at your own loaded hourly cost. Set that against the yearly and one-off costs in the quote.
Is there a standard saving per camera we can put in our business case?
No, and we do not publish one. The saving depends on how many hours your people spend watching or reviewing each view today, which share of that the alerts replace, and how long checking alerts takes. Those come from your rosters and a pilot.
Does running more use cases on the same camera raise the licence line in the calculation?
No. SquintPRO is licensed per camera stream, with unlimited use cases on a licensed stream, so extra use cases raise the hours replaced without raising the licence cost. Count each one only once it has measured hours or events.
How long until we have measured figures instead of estimates?
The camera survey gives the camera count first. The pilot then measures the replaced share, the checking time and the events found, over long enough to cover shift patterns and at least one tuning cycle: usually weeks, not days. The exact duration is set at scoping.
Should avoided incidents count as savings in a safety monitoring ROI case?
Only where your own records hold a cost per incident and the pilot shows the system catches that kind of event. Otherwise report the events found, the response time and the trend as counts, so the financial case does not depend on assumptions about severity.
Does keeping footage on-premise in the Kingdom change the payback?
It moves cost between lines. On-premise deployment keeps footage on servers at your site, which usually adds processing hardware to the one-off cost. The cloud option, with 3, 6 or 12 months' retention, puts more of the cost in the yearly line. Choose for policy reasons first.
Next
Related pages
Next step
Check our ROI figures
Paste the worksheet lines you have, even if some are blank, with the site and roughly how many cameras it has. We will tell you which inputs a survey and a pilot would measure.
Or directly
+966-11-2180999 [email protected]Tawuniya Towers, North Tower, 7th Floor, King Fahad Highway, Olaya, P.O. Box 56835, Riyadh 11564, Saudi Arabia