Case Study · 2023–2024
When the Phone Is the Business
Full project lifecycle — from a vague problem to a production IVR system, a vendor feature request that shipped, and a business that outgrew its own phone line.
The Problem
My client ran a tech services business serving elderly clients — teaching them to use modern technology safely and dispatching technicians for in-home repairs. About 50 calls a day came in. Every single one hit the owner's personal cell phone.
If he was in a meeting, callers went to voicemail. Most of them hung up instead of leaving a message. He'd end the meeting, scroll through missed calls, and start calling people back. Some picked up. Most didn't. Callbacks could take days. There was no routing, no professional greeting, and no way to tell an urgent repair request from a class inquiry.
He came to me with the problem, not a solution. He said he needed a better way to handle phone calls. What that actually meant — what tool, what architecture, what it should cost — was mine to figure out.
What the Phone Situation Looked Like
- • ~50 inbound calls/day to one personal cell phone
- • No greeting, no routing, no prioritisation
- • Busy? Callers hit voicemail — most hang up
- • Callback time: hours to days
- • Urgent repair calls treated the same as class inquiries
- • Owner's personal number publicly visible to all clients
How I Built It
Research and vendor selection (Jun 2023)
- Started by consulting someone who runs phone operations professionally — learned what tools serious call centres actually use.
- Researched Nextiva, mapped its capabilities to the client's needs, and built the business case.
- Negotiated pricing with Nextiva's sales team and closed the deal myself.
Requirements gathering and call flow design (Jul – Aug 2023)
- Drafted call flows, presented, incorporated feedback, revised — multiple iterations to reach the final design.
- Final flow: (1) voicemail for classes, (2) reach the owner directly for urgent matters, (3) make an appointment, (4) automated cybersecurity tips.
- Key design decision: Option 2 built triage into the system. Callers self-selected into urgency categories. The menu became a filter, not just a routing tool.
Configuration and build (Aug – Sep 2023)
- Built everything in the Nextiva portal: admin account, user accounts with role-based permissions, complete call flow, all audio files.
- Ported the client's existing phone number — personal number went private permanently.
- Zero-downtime constraint: the existing line had to stay live throughout the entire build and test cycle. No touching it until the new system was fully validated.
Testing and go-live (Sep – Oct 2023)
- Routed test flow to my own number and a colleague's — never the client's live line.
- Tested: routing for each menu option, audio quality, system stability, and loop behaviour.
- Found a bug: the system hung up after voicemail instead of looping back. Worked with Nextiva support to resolve.
- Scheduled cutover and went live after all test cases passed.
Production Incident: The Saturday Call
About a week after launch, the client called me in a panic on a Saturday. He'd called his own line and the audio was breaking up — choppy, distorted, barely intelligible.
Root cause: the audio file format and compression settings weren't optimised for Nextiva's codec. I hadn't caught it in testing because the degradation was subtle in a quiet desktop environment. Calling from a car with road noise and mobile compression made it obvious. Reprocessed the audio files and resolved it same-day.
Lesson learned: test under the conditions your users actually experience, not the conditions your desk provides.
Influencing the Vendor's Product Roadmap
From the start, we identified a feature gap: Nextiva had no call transcript functionality. The client needed the ability to read — not just listen to — what clients said in voicemails and recorded calls.
I submitted a formal feature request and then followed up. Repeatedly. Over eight months I maintained contact with the Nextiva team, provided use cases, clarified requirements, and pushed for prioritisation. By February 2024, Nextiva shipped call transcripts as a production feature.
The downstream impact was unexpected. The client now uses those transcripts to generate content: a customer calls asking about phishing emails, that conversation gets transcribed, and the transcript becomes the basis for a web page or guide. The phone system became a content production pipeline nobody planned for.
Results
| Before | After |
|---|---|
| 50 calls/day to one personal cell phone | Professional IVR with automated routing |
| Callers wait hours to days for a callback | Every call gets an immediate response path |
| No way to distinguish urgent from routine | Callers self-triage via menu (urgent = press 2) |
| Owner's personal number public to all clients | Personal number fully private |
| Zero staff dedicated to call management | Hired dedicated staff to manage incoming calls |
| No record of call content beyond voicemail | Full call transcripts → content pipeline |
↑ Growth
Business growth directly enabled by the system
The owner credits the IVR as the infrastructure that let the business scale. Customer confidence grew, call volume increased, and the company hired its first dedicated call management staff.
✓ Quality of Life
"A great life improver"
His words, not mine. No more personal cell phone ringing at all hours. Every call that reaches him has passed through the triage filter — he knows it's urgent before picking up.
↗ Unplanned
Transcripts became a content tool
The feature we spent 8 months pushing Nextiva to build is now used daily to generate web content and client guides. The IVR feeds the business's content strategy.
Exact call metrics are measured by the client and are confidential.
What I'd Do Differently
Test under real conditions from day one. The audio file issue was avoidable. I should have confirmed Nextiva's recommended formats before recording and tested with mobile calls in noisy environments.
Build monitoring into the go-live plan. The client's concern about silent failure was legitimate. I should have had a monitoring solution ready before cutover — not after.
Document the call flow as a living artefact. Menu options changed over time. Without a maintained diagram, each change required reconstructing the current state from memory.
Create a formal handoff and training plan. When the client later hired staff, there was no structured onboarding for the system. User guides and a troubleshooting runbook should have existed from day one.