10,000 Contacts. 250+ Appointments. 42 Days.

The question isn’t always “how do we get more leads.” Sometimes it’s “how many of the contacts we already have are we actually reaching?”

For most teams, the honest answer is uncomfortable.

Levi Rodgers Real Estate Group in San Antonio and the Duncan Duo Team in Tampa Bay both asked it. They came back with different answers — and used Felix, Fello’s AI teammate for real estate teams, to test them.

LRG: 10,000 “trash” leads and a 40% stick rate

Levi Rodgers built one of the largest real estate organizations in Texas. Roughly 350 agents. Deep portal partnerships. A database of over 450,000 contacts.

And a pile of leads his team had labeled trash.

Not because the contacts were actually worthless. Because they were old. The leads from 2020 and 2022 had been triaged to the bottom so many times they’d effectively been written off. Every time a fresh inquiry came in, the older contacts fell one spot further down the list. That’s not a failure of discipline. It’s just how prioritization works at scale. Agents and ISAs are rational people. They call what seems most likely to convert right now. The contacts that have been sitting for two years rarely make that cut.

The consequence is a category of leads that never gets a real look regardless of actual intent. The contact’s situation may have changed. Their timing may be perfect. Nobody knows, because nobody called.

Rodgers knew there was value in that older data. He also knew his team would never get to it on their own. So he ran a different play.

He loaded Felix, Fello’s AI teammate, with 10,000 of those deprioritized contacts — deliberately starting with the leads his team had effectively given up on — and let Felix work them. In six to seven weeks, Felix set over 250 appointments. LRG held roughly 125 of them, a stick rate above 40% from contacts that had no recent engagement with the team.

Rodgers put it directly: “These are people that our agents wouldn’t have hit and that our ISA teams would not have hit.”

That last part is the point. These weren’t contacts the team tried and failed on. They were contacts the team never reached because something newer always took priority. The leads weren’t trash. The prioritization system just made them invisible.

The operational shift reinforced that. Felix didn’t replace LRG’s ISA team. It changed what the ISA team was responsible for. With Felix handling first contact on the database’s coldest segments, the human team moved toward post-appointment accountability: everything between the set appointment and the signed contract. The team got more specialized, not smaller.

Duncan Duo: A database too big for humans to work

Andrew Duncan built one of the most recognized real estate brands in Tampa Bay. He also built a database of 130,000 contacts. And knew his team wasn’t reaching most of them.

“Even when they were calling that enormous database, they weren’t getting the impact that I expected,” Duncan said.

A 130,000-contact database isn’t a list problem. It’s a coverage problem. Even a strong ISA team has a realistic ceiling on how many contacts they can touch in a week, a month, a quarter. Adding headcount gets expensive fast when the job is simply creating more coverage. When the database outgrows the team’s capacity for outreach, the gap between what’s possible and what’s happening quietly widens.

This is a different challenge than LRG’s. LRG had a prioritization problem. Duncan Duo had a scale problem. The contacts weren’t being deprioritized because of triage logic. There just weren’t enough hours in the day to reach them all.

Duncan was one of Felix’s earliest beta adopters. He started deliberately small: 5,000 older, unassigned leads the ones his ISAs had never prioritized because they were managing inbound. “I’ll launch tomorrow,” he told Fello’s team when the product was still being refined. “I don’t care. It’s better than no call at all.”

Once the model proved out, Duncan made a deliberate choice about how to expand it. Rather than rolling Felix out across the team unilaterally, he gave every agent a decision: opt into Felix-sourced leads at an adjusted commission split, or opt out. The structure gave agents ownership of the decision instead of forcing another tool or lead source into their workflow. Not a single agent opted out.

That detail matters. Agent adoption is where a lot of new tools become shelfware. Duncan’s opt-in model turned a potential friction point into a commitment — agents chose in, which meant they worked the handoffs.

The results aligned with the team’s best three-month stretch in three years. Felix generated 323 handoffs, contributed to multiple new pendings, and within 49 days of first outreach, a $390,000 listing was active.

What it comes down to

Both teams faced the same underlying constraint: more database than their teams could realistically work. The traditional fix — adding headcount — gets expensive fast when the job is simply creating more coverage. What both found is that AI can handle the coverage work at a unit economics that hiring can’t match, which frees the human team for the part that actually requires a human.

The larger lesson isn’t really about AI. It’s about what your team is actually built to do well, and whether the work you’re asking them to do matches that.

Learn more

How much of your database can your team realistically work this quarter — not in theory, but given actual bandwidth and the reality that newer leads always win?

If you’re asking the same question, see how teams like LRG and Duncan Duo are answering it with Felix.

[See Felix in Action →]