The Follow-Up Playbook: Three Moments, and the Two Kinds of Work Each One Needs

The assumption that flattens three jobs into one

Ask most contractors what "follow-up" means and you'll get one answer: chasing a lead until they respond. Usually that means a CRM sequence that fires on a timer after an estimate goes unsold. A few templated texts, a couple of emails, maybe a call. If the homeowner answers, sales takes over. If not, the sequence runs out and the estimate gets written off.

That answer isn't wrong. It's just incomplete. It treats "follow-up" as a single job that happens at a single moment, when a stalled estimate is actually only one of three places a paid-for lead quietly loses value after the first contact.

The first leak happens before your team ever drives to the estimate: a booked appointment that should never have made it onto the calendar. The second is the one everyone already thinks of: the detailed proposal that goes quiet. The third happens after the invoice is paid, when a finished job that should have produced a review, a referral, or both just... ends.

Follow-up is three moments, not one job

Each of these moments asks a different question, and generic automation answers all three the same way: send another message and see what happens. That's why a single drip sequence, stretched across all three, ends up doing a mediocre job at each one.

What connects all three is the same missing ingredient: context that was captured earlier in the relationship and never carried forward. The marketing channel, the project, the quote, what the homeowner already said, what your team already knows. Qualification, rehash, and reviews and referrals all get sharper the moment that context is attached to the message instead of stripped out of it.

But look at the right-hand column. Carrying the context forward is necessary and it is not sufficient. Every one of these three moments ends in a question a record cannot answer on its own.

Neither software nor people can do this alone

Follow-up is one of the few problems in contracting where both of the obvious answers are wrong, and they are wrong in opposite directions.

Buy software and you get scale without judgment. A sequence can touch two thousand records this month, at three in the morning, without forgetting one. It cannot hear that the homeowner sounds annoyed. It cannot tell the difference between a buyer who is comparing bids and a buyer whose father just went into the hospital. Its answer to silence is always the same: send another one.

Hire people and you get judgment without scale, and usually without memory. An answering service or a part-time coordinator can have the conversation, but they open every call cold. They have not seen which ad the homeowner clicked, which project pages they read for eleven minutes, what the salesperson wrote in the CRM, or that this is the second time this household has inquired in two years. So they ask questions the homeowner already answered, which is exactly the signal that makes a premium buyer decide you are not organized enough to run their $80,000 project.

That division of labor is not a compromise between the two. It is what makes each side better than it could be alone. The system earns the call: it watches the record, applies the rules your team approved, and decides that this specific homeowner, at this specific moment, is worth a person's time. The person earns the answer: they open the call already knowing where the lead came from, what the homeowner researched, what they were quoted, and what they said last time, and they write back what they learn so the next decision is sharper than the last.

It matters who that person is. A trained agent who knows your trades, your pricing range, and your service area can tell the difference between a homeowner asking a normal scoping question and one describing a project you do not take. A generic call center cannot, which is why generic call centers produce transcripts instead of opportunities.

The boundary matters just as much. The agent gets the answers. Your salesperson does the selling. Scope, pricing, financing, objections, the walkthrough, and the close belong to your people, and every call an agent makes should end with your salesperson better prepared than they would have been without it.

Moment one: the estimate that should have been qualified before your team drove out

The instinct with a booked estimate is to treat the calendar as proof of readiness. A name, a phone number, and a time slot look like a qualified appointment. They aren't. They're a placeholder that hasn't been checked against anything yet.

Only 39% of businesses consistently apply lead qualification criteria before dispatch, which leaves roughly 55% of leads unworked or mishandled by the time a salesperson shows up (Source: MarketingSherpa). Separately, only about 25% of marketing-generated leads are judged high enough quality to go straight to a salesperson (Source: Gleanster Research). The gap between those numbers is the drive time, gas, and estimate hours your team spends on appointments that were never going to close, on top of the ones that will.

This is what qualification does: it pulls together where the lead came from, what they viewed, what they already told you, CRM notes, and project details before anyone drives anywhere. An approved data append can add useful outside information to fill in real gaps. The workflow checks that against the fit rules your team approves and flags exactly which answers are still missing.

Then someone has to go get those answers, and this is where most qualification setups quietly fail. A form asking a homeowner to self-report their budget gets ignored or gets a number they made up. A text asking about timeline gets a one-word reply that means nothing. The information that decides whether an appointment is worth a truck lives in a conversation, not a field.

A 15-minute question-and-answer call is enough. Scope, timeline, service area, budget range, and who is actually making the decision, asked by a trained person in the contractor's voice who already knows what the homeowner read on the site and does not make them repeat it. Homeowners rarely mind the call, because it is short, it is specific, and it demonstrates that the business is organized. Qualified homeowners get booked with the salesperson. The rest never take a slot. And nothing about that call involves quoting a price, because pricing is the salesperson's conversation.

Moment two: the estimate that goes quiet after the proposal

This is the leak most contractors already have a plan for, and it's still the one that costs the most, because the plan is usually a timer. Day 1, day 7, day 14. Three or four templated messages, then the estimate goes into a folder nobody reopens.

The quieter assumption behind that plan is that the problem is frequency, that the homeowner just needs to hear from you more often. For a buyer making a $30,000 to $80,000 decision, that assumption is often the reason they stop responding, not despite the follow-up, because of it. A templated "just checking in" message signals that the business doesn't remember who they are or what they discussed. For a homeowner who just sat through a two-hour consultation, that gap is felt even when it's never said out loud.

48% of salespeople never follow up after the initial appointment at all, not from indifference, but because they're already running the next estimate (Source: IRC Sales Solutions and HubSpot). Of the ones who do, 92% stop after four attempts, well before most homeowners on a large project have reached a decision (Source: Marketing Donut and The Brevet Group). Rehash starts only after your sales team hands the estimate off. From there, it keeps where the lead came from, the project, the quote, and prior messages together, watches for a quiet buyer suddenly returning to pricing or project pages, and checks appointments, quote changes, invoices, and payments before anything sends. Any reply stops automatic follow-up.

Then a person calls. This is the part that separates rehash that works from rehash that produces open rates, because the single most valuable thing you can get from a stalled estimate is not a response, it is a reason. Price. Timing. Another bid came in lower. A spouse who never weighed in. A permit that stalled. A kitchen that got postponed for a roof. None of those show up in a click, and a homeowner will say all of them out loud to a person who asks plainly and is not trying to sell them anything on that call.

The reason is worth more than the recovered job, because it is the only one of the three follow-up moments that tells you why you lost. Collected across a hundred stalled estimates, those reasons tell you whether your pricing is off, your timeline promises are unrealistic, or a specific competitor is undercutting you in one zip code. Recovery rates on unsold estimates run 10 to 15% when follow-up is personalized and matched to the buyer's decision process (Source: Wealthy Contractor, directional). On $2 million in unsold quotes, that's $200,000 to $300,000 sitting in a folder your business already earned the right to close.

Moment three: the finished job that never became proof or a referral

The final invoice gets sent, the job gets marked complete, and for most businesses that's where the relationship ends. Nobody asks for a review. Nobody asks for a referral. Not because the customer would say no, but because nobody owns the ask.

91% of customers say they'd give a referral if asked. Only 11% of salespeople ask (Source: The Brevet Group). Only 30% of companies have any formal referral program at all (Source: TrueList). Reviews follow the same pattern from the other direction: 71% of homeowners read reviews regularly before hiring, but only 45% wrote a Google review in the past year (Source: BrightLocal, 2025 Local Consumer Review Survey). That's not a shortage of satisfied customers. It's a shortage of asking.

The timing problem compounds the asking problem. A review or referral request tied to the sales stage rather than the actual finish line, something like "job sold" or "invoice sent," often goes out before the work is verifiably finished, or misses entirely if that trigger never fires. And a review ask and a referral ask are not the same conversation. One asks a customer to share an honest experience publicly. The other asks them to put their name behind an introduction to someone they know. Sending both from the same generic message, or skipping the check for an open service issue first, turns a finished $45,000 kitchen into an awkward moment instead of the next lead.

Connected follow-up ties the review and referral trigger to the system that actually knows the job is complete, not a rough sales date. Before anything sends, it checks whether the job was reopened, a service issue is active, the customer was already asked, or someone on your team is already handling the relationship. Reviews and referrals run on separate timing and separate rules.

That handles the customers who respond to a message. It does nothing for the larger group: the customers who were genuinely happy, meant to leave the review, and never got around to it. They are not refusing. They are busy. A text is easy to intend to answer later and never answer at all, and no amount of resending fixes that, it just turns goodwill into irritation.

A short call does fix it, and it does two things a message cannot. It gets the review while the customer is talking to a human who thanks them, and it surfaces the problems you would otherwise never hear about. A customer who is 80% happy will click nothing and say nothing. On the phone, that same customer will mention the trim in the hallway. That is a service call you can still win back, and it is a one-star review you just avoided by asking before the internet did. When the answer is genuinely good, that is the moment to ask for the referral, which is a different ask on a different day than the review, and never conditional on it.

What all three leaks are worth

None of this is abstract. Each moment has a number attached, and none of them require a dollar of new ad spend to fix, because the lead, the estimate, and the finished job are all things your business already paid for.

Why this doesn't work as three separate tools

Qualification, rehash, and reviews and referrals could each be bought as a standalone tool. A lead-scoring plug-in for the first. A drip-sequence app for the second. A review-request widget for the third. Plenty of businesses run exactly that setup, and each piece works in isolation, which is part of the problem: none of them know what the others already learned.

Bolting a call center onto that stack does not fix it either. Now you are paying people to dial from three disconnected lists, opening every call without the context that would have made it worth taking, and the reasons they collect land in a spreadsheet nobody reads instead of on the record that decides who gets called next.

The context that makes any one of these moments useful, the marketing channel, the project, the quote, what the homeowner already said, what happened in sales, is captured once and needs to travel with the lead through all three moments. Lead Intelligence is what carries it. The same record that tells qualification which appointments are worth field time tells rehash which buyer it's talking to, and tells the reviews and referrals workflow which finished job came from which marketing channel.

It runs in the other direction too. A cancelled appointment, a stalled estimate that never revives, a referral the customer declines, a job that reopens for a service issue, all of that feeds back into the same record as a signed job would. The system doesn't just learn from wins. Every outcome, including the quiet ones, sharpens what "worth qualifying," "worth rehashing," and "worth asking" mean the next time.

Run separately, better qualification, better rehash, and better review and referral requests each still improve their own moment. Connected to the same lead history, and staffed by people who can read that history before they dial, they add up to something else: a business that gets more revenue out of every lead it already paid for, not just the leads that were easy to begin with. See how the three run together on the Follow-Up page, or look at each piece on its own: qualification, rehash, and reviews and referrals.

Self-audit: what's true in your CRM right now

Before deciding whether any of this matters for your business, these are worth answering with real numbers, not instinct.

Before the estimate

• Does every booked estimate carry a project scope, timeline, and source, or does the calendar just show a name and a time?
• How often does your team discover a bad fit during the drive, or after the appointment has already started?

After the estimate goes quiet

• How many unsold estimates from the last 12 months are sitting in your CRM, and what's their combined value?
• Does the follow-up message change based on who the buyer is, or does the same sequence go to every unsold estimate regardless of project type or concern?

After the job is done

• What share of finished jobs from the last 12 months actually received a review request, a referral request, or both?
• Does anyone check for an open service issue before a review or referral message goes out?

Across all three

• When a follow-up message gets no response, does anyone pick up a phone, or does the record just go quiet?
• If someone did call, would they know what the homeowner already researched, was already quoted, and already told your team?