How to Follow Up on Estimates With AI, and Close the Jobs You Already Quoted

The hard part is already done. You took the call, drove to the house, looked at the job, and handed over a fair estimate. Then the homeowner said the six words that quietly kill contractor revenue: "let me think about it." And then nothing. No follow-up, because you were on the next job, and the estimate you already earned went to whoever did call back. This is the leak that costs home service businesses more than missed calls, because these are not cold leads. They are people who invited you out, saw your price, and were one conversation away from saying yes. This article is about closing that gap with AI, honestly: what to automate, and the one follow-up touch every playbook prescribes and almost nobody makes. Pesta handles that specific touch, and we will get to where it fits.

The unsold estimate is the cheapest job you are not closing

An unsold estimate is not a lost cause; it is a job you already paid to create. You spent the ad dollars to get the lead, the time to visit, and the effort to quote. Everything up to the signature is done. Recovering even a fraction of those quotes costs nothing in new marketing, which is why systematic estimate follow-up is the single highest-return sales activity a contractor has, and the most neglected.

‍

The neglect is well documented, and the numbers are not close. Most sales close only after several follow-up touches, research on sales generally puts it around five, yet a large share of salespeople give up after one attempt or never follow up at all. For contractors the reason is not laziness; it is that follow-up requires time on a day already full of jobs. So the estimate sits, the homeowner collects a competing quote, and the job is decided by whoever stayed in touch, not whoever quoted best.

Text first, then the call: the cadence that closes estimates

The follow-up pattern that actually recovers estimates is multi-touch and cadenced, and the contractors who use it agree on its shape. It is not one text and a shrug.

Text leads because it is fast and low-friction, and it closes the easy yeses. But notice the third touch. Day 3 is when the homeowner has a competing quote in hand and a real question that is keeping them from deciding, and answering that question takes a conversation, not a template. That call is where the estimate is actually won or lost. And it is the one almost nobody makes.

Every estimate playbook prescribes the Day-3 call. Almost nobody makes it

Read any serious guide to closing unsold estimates and you will find the same instruction: the sequence needs a real phone call around Day 3, because a text can nudge but it cannot handle the objection that is actually stalling the sale. "Is the tile included?" "Can you do better on price?" "What about financing?" Those are conversations, and the homeowner who asks one is close to signing, if someone picks up and answers.

Then the same guides admit, plainly, that contractors do not make that call. They describe the reality: sticky-note follow-ups, a technician who chases some customers and forgets others, a sequence that fires texts flawlessly and quietly skips the one human touch that requires a human's time. The call is prescribed as essential and left undone, because the person who would make it is under a house on the next job. The text handles the easy ones. The call that would close the hard, high-value ones never happens.

That is the gap. Not the strategy, which contractors understand, but the labor: the one touch that needs a conversation, and the one resource a busy contractor never has, someone to have it.

How an AI voice agent makes the follow-up call

This is the touch an AI voice agent is built to cover, and it is worth being precise: it is the voice follow-up the playbooks prescribe, to the homeowners who requested a quote and did not respond to the text, not cold outreach. It calls the prospect, has a real conversation about the estimate, answers the common questions that stall a decision, and books the job or the start-date call while the homeowner is on the line. Done this way, it turns the pile of unsold estimates from a spreadsheet nobody works into follow-up that actually runs.

Two capabilities make this work on an estimate call specifically. First, the conversation is a negotiation, not a script: the homeowner pushes on price, asks what is included, wants to know about timing, and a rigid system stalls exactly where the sale is decided. Because Pesta answers from a knowledge base rather than a fixed script, it handles that back-and-forth, the same reason it books the call that turns into a job instead of a voicemail. Second, Pesta's call analysis reads every follow-up call, so you can see which objections come up most, which estimates are worth another touch, and where deals stall, which turns follow-up from guesswork into a system you can tune. Powered by Deepdub, it does this across a wide range of languages. Recovering these jobs also keeps your calendar full without the churn, and the same system can handle the reschedule calls that reshuffle the day once the job is booked.

None of this replaces your sales judgment on the big, complex bids. It makes the routine follow-up calls that currently do not happen, so the estimates you already earned stop dying in silence.