Video service ROI: how to measure costs and results
The return on video customer service depends on what changes in the operation: costs, capacity, resolution and customer experience. Learn how to build a verifiable comparison without treating every time saving as a financial saving.
The key decision
Measure the cost of resolving the problem as well as the cost of the call. Use the same period and type of request, and assign financial value only to benefits you can demonstrate.
1. Define the question the pilot must answer
“Does video work?” is too broad. Prefer an operational question: “For installation support requests, does video resolve more cases with the same budget without increasing complaints?” This defines demand, outcome and a quality condition.
Choose the period, eligible population and unit of analysis. A call is an interaction; a case is a request that may require several interactions. Use a case identifier to group contacts, including when customers start on video and finish by phone.
- Before the pilot: record volume, costs, staffing, resolution and request types.
- During the pilot: keep definitions consistent and record changes in staffing, campaigns, products or processes.
- When comparing: use equivalent periods and separate simple from complex cases. Report sample size and missing data.
If the team selects only the easiest cases for video, a higher resolution rate does not demonstrate a channel advantage. Where feasible, assign eligible cases comparably between flows and retain necessary alternatives for customers.
2. Define numerator, denominator and observation window
Minimum measurement dictionary
Cost per resolution
- Proposed calculation
- Total service cost ÷ resolved cases.
- Interpretation caveat
- Include effort spent on unresolved cases and all channels within scope.
Resolution rate
- Proposed calculation
- Resolved cases ÷ eligible cases received × 100.
- Interpretation caveat
- Use the same outcome criteria and follow-up window.
First contact resolution (FCR)
- Proposed calculation
- Cases resolved at first contact, with no repeat contact within the window ÷ eligible cases × 100.
- Interpretation caveat
- Define how to link new contacts to the same reason.
Average handling time
- Proposed calculation
- Sum of handling times ÷ measured interactions.
- Interpretation caveat
- State whether after-call work is included; do not confuse handling time with waiting time.
Satisfaction (CSAT)
- Proposed calculation
- Responses classified as satisfied ÷ valid responses × 100.
- Interpretation caveat
- Report the scale, scores considered positive and response rate.
| Metric | Proposed calculation | Interpretation caveat |
|---|---|---|
| Cost per resolution | Total service cost ÷ resolved cases. | Include effort spent on unresolved cases and all channels within scope. |
| Resolution rate | Resolved cases ÷ eligible cases received × 100. | Use the same outcome criteria and follow-up window. |
| First contact resolution (FCR) | Cases resolved at first contact, with no repeat contact within the window ÷ eligible cases × 100. | Define how to link new contacts to the same reason. |
| Average handling time | Sum of handling times ÷ measured interactions. | State whether after-call work is included; do not confuse handling time with waiting time. |
| Satisfaction (CSAT) | Responses classified as satisfied ÷ valid responses × 100. | Report the scale, scores considered positive and response rate. |
These are proposed study definitions. Check the rules of source reports: two identically named metrics may measure different events. For waiting, separating camera setup from time waiting for staff helps locate the problem.
A seven-day repeat-contact window may suit an illustrative simple-support pilot; another service may require longer. Compare only groups that have completed the window. Mark cases still being observed as pending rather than prematurely resolved or failed.
If the denominator is zero, the metric is unavailable for that segment. If there are few cases or low survey response, show absolute counts. A satisfaction score based on five responses should not look as conclusive as a larger, representative sample.
3. Separate total cost, investment and realized benefit
For cost per resolution, add the entire operation's costs within scope: staff, employment costs and allocated supervision, platform, variable usage, equipment, training, integration and follow-up work. Identify monthly and one-off amounts and the basis for allocating shared costs.
For project ROI, use the incremental costs and the attributable benefits of the change within the same time horizon. Do not count the entire salary of an existing team as new investment if it was already being paid before the pilot.
- Realized savings: spending that ceased, documented in a contract, payroll or invoice.
- Freed capacity: hours available for other work. Record how they are used; they do not automatically become money.
- Commercial outcome: if measured, use incremental contribution margin after variable costs and returns, with demonstrable attribution. Gross revenue is not profit.
Avoid counting the same gain twice. If reduced rework already appears in lower outsourcing expense, do not also add the value of those same hours as another financial benefit.
4. Hypothetical example: from operational spreadsheet to ROI
All figures below are illustrative. They do not represent a customer, NEFT price, market benchmark or promised result. The example assumes 1,000 eligible cases per month in each scenario, with outcomes observed over the same window.
Example monthly recurring costs and resolutions
Team
- Before
- BRL 20,000
- Pilot
- BRL 20,000
Outsourced service
- Before
- BRL 8,000
- Pilot
- BRL 5,000
Other allocated costs
- Before
- BRL 2,000
- Pilot
- BRL 2,000
Incremental video cost
- Before
- BRL 0
- Pilot
- BRL 1,500
Total monthly cost
- Before
- BRL 30,000
- Pilot
- BRL 28,500
Eligible cases received
- Before
- 1,000
- Pilot
- 1,000
Resolved cases
- Before
- 800
- Pilot
- 950
Cost per resolution
- Before
- BRL 37.50
- Pilot
- BRL 30.00
| Item | Before | Pilot |
|---|---|---|
| Team | BRL 20,000 | BRL 20,000 |
| Outsourced service | BRL 8,000 | BRL 5,000 |
| Other allocated costs | BRL 2,000 | BRL 2,000 |
| Incremental video cost | BRL 0 | BRL 1,500 |
| Total monthly cost | BRL 30,000 | BRL 28,500 |
| Eligible cases received | 1,000 | 1,000 |
| Resolved cases | 800 | 950 |
| Cost per resolution | BRL 37.50 | BRL 30.00 |
Cost per resolution is 30,000 ÷ 800 = BRL 37.50 before, and 28,500 ÷ 950 = BRL 30.00 during the pilot. The table includes recurring expenses only. For a first-year view, add implementation and use resolutions actually observed during the period.
Now assume an initial investment of BRL 6,000 and a proven reduction of BRL 3,000 per month in outsourcing expense attributable to the project. Video's recurring incremental cost is BRL 1,500 per month. Staff costs remain unchanged; additional resolutions receive no separate financial value in this calculation.
With a recurring net benefit of BRL 1,500 per month, the simple payback period on the initial investment would be 6,000 ÷ 1,500 = 4 months. This assumes the full gain from month one and constant cash flows; it excludes the time value of money. With a gradual rollout, record each month's amounts and find when the cumulative balance stops being negative.
5. Show what could change the conclusion
A before-and-after comparison may mix video's effect with training, staffing changes or seasonal demand reductions. Record these factors and avoid attributing everything to the channel. Observational data supports decisions but does not isolate causality on its own.
Sensitivity of the same example over 12 months
Proven gross monthly savings
- Central scenario
- BRL 3,000
- Conservative scenario
- BRL 1,000
Incremental annual costs
- Central scenario
- BRL 24,000
- Conservative scenario
- BRL 24,000
Net annual benefit
- Central scenario
- BRL 12,000
- Conservative scenario
- −BRL 12,000
Annual ROI
- Central scenario
- 50%
- Conservative scenario
- −50%
Simple payback period
- Central scenario
- 4 months
- Conservative scenario
- Not reached under these conditions
| Assumption | Central scenario | Conservative scenario |
|---|---|---|
| Proven gross monthly savings | BRL 3,000 | BRL 1,000 |
| Incremental annual costs | BRL 24,000 | BRL 24,000 |
| Net annual benefit | BRL 12,000 | −BRL 12,000 |
| Annual ROI | 50% | −50% |
| Simple payback period | 4 months | Not reached under these conditions |
In the conservative scenario, BRL 1,000 in monthly savings does not cover the BRL 1,500 recurring cost. Calculating a positive payback period by dividing investment by a negative balance makes no sense. The decision may be to reduce costs, adjust the workflow, extend observation or end the pilot.
Track quality alongside finances. An operation can become cheaper while leaving more requests pending. Define in advance which resolution, complaint and access outcomes prevent expansion, even if estimated ROI looks attractive.
6. Turn measurement into a decision routine
Maintain a spreadsheet with period, service, cases received and resolved, interactions, costs by category, data source, owner and notes. Retain approved-version assumptions; do not retroactively change criteria to favor results.
The NEFT reports support analysis of volumes and operational times. To assess resolution and financial returns, combine records with case tracking and finance information. Confirm which fields are available in your configuration.
- Operations checks record quality and exceptions.
- Finance validates costs and actually realized savings.
- The pilot owner compares groups and records limitations.
- Leadership decides to expand, adjust or stop, with a reassessment date.
The useful outcome is a decision with enough evidence for the next step. An ROI figure without a period, denominators and assumptions does not support that assessment.

