# Marcus Delgado, Agency Principal at Delgado Insurance Group — read of Insurance Lapse Recovery Voice Agent, June 7 2026

> 14 years running a 7-person independent shop in Fresno, CA. P&C and commercial lines. I coach my kid's Little League on Saturdays so anything eating Monday morning is a problem.

## How I got here

Lost three auto renewals in one week back in May to Farmers direct. All three had emails sitting in their spam folders. Went to Google and typed "automated renewal calls for insurance agents" and this came up around result five or six. I'd already opened tabs for two other tools before I landed here, so I was already in comparison mode when I read this.

## What I clicked first

The hero number stopped me: "Recover 18-22% of at-risk policies without adding headcount." That range is suspiciously specific in a way I've learned means either it's real or it's made up to sound real. I genuinely couldn't tell which. I kept reading to find out where it came from.

The line that actually pulled me in was: "Your prospects have already switched carriers or been contacted by three competitors." That's not generic. That's the actual sequence I watch happen every quarter.

## Where I paused

The four-step workflow section. Specifically step one: "Upload your at-risk list CSV of policyholders 30-60 days from renewal lapse. We match to phone numbers and validate contact data."

I run Applied Epic. My data is not clean. It never is. The phrase "validate contact data" is doing a lot of work there and I have no idea what it actually means. Does it hit a phone validation API and flag bad numbers? Does it skip cells? Does it call anyway and burn a $2.50 charge on a disconnected line? That one sentence left me with three questions and answered zero of them.

## What I distrusted

Two things.

First: the 18-22% recovery rate appears in the stats section with no sourcing at all. Then I get to the bottom of the page and find: "Honest disclosure: we don't have live customers on this idea yet." So where does 18-22% come from? It's either a Fermi estimate they're passing off as a metric, or it's lifted from industry research they didn't cite. Either way I'm not supposed to notice that it was presented as a real outcome before being walked back as a projection. That's a credibility hit.

Second: the Wishdeal Factory scoring section at the bottom rated their own landing page quality 3/10. I respect that they published it. But now I'm staring at a product I was considering buying that the people who built it rated 3 out of 10 for landing page quality. That's either admirably honest or a sign they know it's half-baked and shipped it anyway. I can't figure out which.

## What would convince me

One real recorded call. Not a demo they scripted. An actual call where the prospect pushes back, gets confused, says "wait who is this?" and the AI handles it gracefully or awkwardly. I want to hear the failure modes, not the best run. If the call holds up under a confused 68-year-old who thinks it's a scam call, I'm in.

Also: one agent's actual numbers. Not a range. One shop, one month, how many calls placed, how many completed conversations, how many calendar books, how many converted. I don't need a case study PDF. I need a screenshot of a spreadsheet.

## What I'd ask in an email reply

1. Where does the 18-22% figure come from? Is that modeled, or did someone run this and actually track it?

2. What happens when the AI hits a cell with a bad number or a Spanish-speaking policyholder? Does it bail gracefully, leave a message, charge me anyway?

3. My state (California) has specific telemarketing consent rules for insurance calls. When you say "compliance-ready" and "DNC list integration," does that mean I'm still on the hook for getting prior express written consent, or does your system handle that documentation?

## Verdict: on-the-fence

The problem framing is the best I've read on a tool like this. The product logic makes sense. But I'm being asked to trust a recovery rate that contradicts the "no live customers" disclosure at the bottom of the same page, and I don't know if the AI can survive a real call with a skeptical policyholder. The trial offer is good enough that I'd probably upload 30 names and see what happens.

---
*Memo by skeptic persona, generated 2026-06-07. Studio breaks own self-grading loop.*