Case studies
Built in a roofing company first.
Versafield wasn't dreamed up in a boardroom.
Brennen Harris runs Roof Boyz (opens in a new tab), a roofing company in the Phoenix Metro, and built these systems for his own business first:
- Phones answered after hours — no more morning voicemail triage
- Every lead followed up in under a minute
- Follow-ups that send themselves, on schedule, every time
- Invoices that go out like clockwork
- Materials ordered and labor crews scheduled without lifting a finger
The result, in his words: time to work on the business, not just in it. Versafield exists to install that same setup in yours.
Built by a tech-forward home service business owner. Not a guru. Not a tech bro.
Client case studies: published when they're real
Versafield AI is new, and we won't fill this page with invented logos, testimonials or percentages. Here's the format every client case study will follow, and below it, exactly how each result will be measured, from a baseline taken from your own data to a report that includes what didn't work.
Template preview, not a client result.
[Result] for a [Trade] company in [time]
Example layout: The Receptionist at a home service company. Placeholders only.
- Baseline
- 30 days of call and lead data before launch: unanswered calls, after-hours calls and median callback time.
- The hire
- [Which AI employee, and the hard limits it works within]
- Shadow mode
- [Shadow-mode length]
- Measured result
- [Baseline vs. after clock-in day, using the same definitions]
- Caveats
- [What needed correcting, what was escalated, and why the result may not transfer]
- Client quote
- [Shown only with the client's written permission]
Where our proof stands today
We'll move each rung to 'available' only when it's real. Until then, the example workflows on our trade pages are illustrations, not results.
Founder experience
Status: Available
Documented process
Status: Available
Example workflows
Status: Available
Working AI demonstration
Status: Building
Client metrics
Status: Building
Client testimonials
Status: Building
Full case studies
Status: Building
Reference calls
Status: Building
How we measure: baseline first, then shadow mode
One change at a time, measured the same way before and after. See the full process.
Pull the baseline from your own data
Before anything is built, we pull about 30 days of history from your phone system, CRM and invoicing: total, answered and unanswered calls, after-hours calls, callback time, web-lead response time, open estimates and days to payment.
Agree on the definitions
We write down how each number is calculated, such as what counts as a qualified call or a useful response, so 'before' and 'after' measure the same thing.
Start in shadow mode
Your employees start in parallel while your current process still runs. You see exactly what they would have said and done. We tune until you'd trust them live.
Measure the same way again
After clock-in day, we pull the same numbers with the same definitions, plus every exception, correction and escalation.
Report wins and boundaries
You get the result in plain language, including what needed fixing. We publish it only with your written approval.
The numbers we baseline
Each hire tracks the numbers that fit its job, using these definitions.
- AI
Answer rate
Qualified inbound calls answered ÷ qualified inbound calls.
- Automation
Median first-response time
Median time from inquiry to a useful response, not just an auto-reply.
- AI
Qualified booking rate
Booked qualified inquiries ÷ qualified inquiries.
- Automation
Missed opportunities recovered
Previously unanswered inquiries that became appointments.
- Automation
Estimate follow-up coverage
Open estimates with a recorded next step ÷ open estimates.
- Automation
Days to payment
Average days from invoice to payment.
- Automation
Review request coverage
Eligible completed jobs that received a review request ÷ eligible completed jobs.
- Human
Transfer success
Appropriate handoffs to a person completed ÷ handoffs attempted.
- Automation
Automation exception rate
Automated actions that needed a correction ÷ all automated actions.
- Automation
Hours saved
Baseline task time minus task time after launch, measured, not estimated.
Trade-specific metrics
For example, roofing clients also track lead-to-inspection rate, inspection no-show rate and time from inspection to estimate. See roofing.
What we will and won't publish
Real clients, with permission
A client's name, quote or logo appears only with their written approval.
No anonymous miracle numbers
We won't post an unnamed company's revenue jump or any other headline figure unless the result is documented and the reason for anonymity is explained.
Demos labelled as demos
Test calls and demonstrations are always marked as demonstrations, never presented as client results.
Boundaries included
Every case study reports caveats, corrections and escalations alongside the wins.
The same definitions every time
Each result states how it was measured and over what period.
Questions about results
Why don't you have case studies yet?
Versafield AI is a new practice, and we'd rather show an empty shelf than a fake one. The employees were built and run in a roofing company first. When our first clients go live, their results, good or mixed, will be published here with their approval.
Will my company's name be published?
Only if you say yes in writing. We can publish an anonymized result and explain why it's anonymized, or not publish at all.
Who approves the numbers?
You do. We share the baseline, definitions and results with you before anything is published.
Do you guarantee results?
No. We guarantee the process: a baseline from your data, shadow mode before anything goes live, monitoring and an honest report. Results are measured against that baseline, not promised in advance.
Can I talk to a current client?
Not yet. Reference calls are on our proof ladder, and we'll offer them as soon as clients agree.
Which hat comes off first?
Book a 30-minute fit call. We'll map where your hours are going and tell you honestly which employee would pay for itself first.