B2B database reactivation case study: AI won back $200K+ in lost deals
A B2B fleet telematics company used AI database reactivation on its closed-lost deals and won back more than $200K, about 35 times the cost.
A car finance broker's written-off leads were converting at zero. An AI agent turned them into more than $300K in settled loans and a 2.8x return.
A Sydney car finance broker had a pile of enquiries its brokers had tried up to five times and written off. They were converting at zero. An AI agent went back to them by text, and those leads have since produced more than $300,000 in settled car loans. On the broker’s own revenue per loan, the campaign returned about 2.8 times what it costs to run, and loans started settling within the first month.
Car finance enquiries arrive fast and go cold fast. People enquire while they are looking at a car, and if a broker does not get through quickly, they often finance it somewhere else.
This brokerage was losing hundreds of leads a month to “exhausted”. Its team had called each one up to five times, and when nobody answered, the lead was marked dead and never touched again. Every one of those leads had been paid for, and none of them was producing a loan.
We built an agent that texts each exhausted lead under the brokerage’s own persona. The first text names the month they enquired and asks whether they are still looking. If they reply, it finds out what they need one question at a time (the car, the timing, their work situation) and books a call straight in with their broker. People who did not reply got two more rounds of follow-ups.
Texts suit this kind of lead. Someone who never picked up an unknown number will often answer a message they can read when it suits them, which is part of why SMS gets read when email does not.
This lead had enquired months earlier. His first reply was “No thanks”. Names, the brand and the car are hidden.
Agent: Hi, it’s Maria from [broker], is that the same [name] who enquired in [month] about a Car Loan?
Lead: No thanks
Agent: No worries, [name]. If you ever want another look at your options, just reach out. Take care.
Lead: Quick question whats your interest rate
Agent: Rates have moved around a bit lately. It really comes down to your profile and which lenders are actually a fit hard to compare apples to apples until a broker checks the full picture. If you are open to a quick review, we can see if your [car] rate stacks up. Want to chat tomorrow or later in the week between 9 and 5:30?
Lead: It’ll have to tomorrow as im about to lock it in
The call was booked for 9am the next day and moved once to 4pm. The client reported the loan approved the following day and settled within a week of the call. More threads from this campaign are on our real conversations page.
| Result | Figure |
|---|---|
| Car loans settled from leads the broker had written off | More than $300,000 |
| Return on the campaign | About 2.8 times what it costs to run |
| Written-off leads who replied | 29% |
| What these leads converted at before | Zero: they had been written off |
| Brokers’ time spent chasing them | None: the agent did all of it |
How these are worked out: the settled loans come from the broker’s own settlement tracker, November 2025 to June 2026. The return compares the loans that settled while the latest campaign ran, at the broker’s average revenue per loan, against what that campaign costs to run. It does not count the leads that kept coming back afterwards.
Yes, for most brokerages. Every brokerage has a pile of leads its team tried and gave up on. They asked for finance, so the interest was real at some point, and a text that remembers what they asked for finds out who is still in the market.
The maths is simple: one extra settled loan a month more than covers what the agent costs to run, and every loan after that is margin from leads you had already paid for and given up on.
Three things made the difference here: the first text named their original enquiry, the agent kept going after a “no thanks” instead of arguing with it, and every interested reply went straight into a booked call with their broker rather than a “someone will be in touch”.
You can see how this works on our re-engagement page. If you have your own exhausted leads, book a free revenue engine audit and we will look at what is sitting in your CRM.
For this Sydney car finance broker it was. Leads its team had tried and marked exhausted were converting at zero. An AI agent texted them, and the loans that settled came to about 2.8 times what the campaign costs to run, based on the broker's own revenue per settled loan.
A lead your team has already tried to contact several times without getting through, so it gets marked as dead and left alone. In this case the brokers had tried each lead up to five times before the agent was switched on.
It reminds them when they enquired, asks whether they are still looking, and asks one question at a time about what they need. If they are interested, it books a call with their broker. If they say no, it leaves the door open politely.
We map where your leads go quiet, then show you the conversations an agent would have had. Thirty minutes, no obligation.