Two Leads, Two Clocks: Building a Measurement System for Service Demand

Short answer: Measure service demand on two clocks, not one. Repair and emergency calls run on a fast clock measured in minutes and hours: was the call answered, was it booked, did a technician run it, did it stick. Replacement, install and other big-ticket decisions run on a slow clock measured in days and weeks: was an estimate presented, was there follow-up, did it close, and was the customer still with you months later. Tag every lead with its clock at intake, start and stop each clock at the right moment, and report the two side by side with their own targets. One blended "lead conversion" number punishes the slow decisions and hides the slow answering.

Back in May I wrote a short piece called Stop measuring all leads the same way. Two kinds of leads, two different timelines, one measurement that hurt both. It is still one of the lessons I repeat most to other owners.

What that piece did not give you was the system. Owners keep asking me the practical version: what do I actually put on the report on Monday morning?

This is my answer, written for owners in the $1 to $5 million range. No software to buy, no company numbers.

The mistake I made first

Let me start with my own mistake, because the whole framework comes out of it.

We had two types of leads in one funnel. One type closed fast but churned more. The other took longer to close but stayed much longer and was worth much more over time.

For months, the team kept optimizing for speed. Faster closes made the weekly number look great. Nobody was doing anything wrong on purpose. The report rewarded speed, so the team chased speed.

Then we looked at 90-day retention, and the picture flipped completely. The leads that took longer to close were the ones that stayed. They were not slow. They were careful. They had done their research before they ever got on a call.

Our reporting could not see any of that. It treated both kinds of leads the same way: "unconverted after 14 days." A careful buyer on day 14 who was still deciding sat in the same bucket as a buyer who had ghosted us on day 3. One was still researching. The other was already gone.

So we separated the reporting. Different funnels, different benchmarks, different coaching. Once we stopped comparing them, we stopped optimizing the wrong things.

That story was not about service calls. But the lesson travels perfectly to home services, because service demand comes in on two very different clocks.

Two leads, two clocks in home services

The first homeowner has no cooling, it is hot, and there are kids in the house. They are calling three companies at once. Whoever answers, books a time today, and shows up wins the job. This decision is made in minutes.

The second homeowner has a fifteen-year-old system that still runs, but they are tired of the repair bills and want to talk about replacing it. They will get two or three estimates, talk it over at dinner, maybe wait for a rebate. This decision is made in weeks.

Both show up in your system as "a lead," maybe from the same ad. Measure them on the same clock and you will get both wrong.

Clock 1, the now clock. Repairs, emergencies, no-heat and no-cool calls, leaks, anything where the customer wants help today or tomorrow. This clock runs in minutes and hours. The customer's patience is short and the competition is one phone call away.

Clock 2, the decision clock. Replacements, new installs, bigger repairs that need a quote, maintenance plans and memberships. This clock runs in days and weeks. The customer is comparing, thinking and often involving someone else in the decision.

The rule is the same one from my May piece, translated for service demand: conversion speed is not the same as conversion quality. The now clock is about speed. The decision clock is about qualification, follow-up and how long the customer stays.

What goes wrong with one blended number

When a company reports one number, usually something like "booked rate" or "lead conversion within 7 days," three things go wrong at once.

The slow decisions look like failures. A replacement estimate still open on day 10 counts against the team, even though that customer may sign on day 25. So people push. They discount too early, call too often, and pressure a family that just wanted a few days to think. Many of those families go quiet.

The slow answering hides. On the now clock, the damage happens in the first minutes: the call that rang too long, the voicemail returned the next morning. A weekly blended number cannot show that, especially if the install pipeline had a good week.

Marketing gets blamed for operations. If a lead source sends mostly decision-clock demand, it looks "slow" on a blended report, so it gets cut. If the after-the-call work breaks down in one market, the whole channel looks bad, and the ads get changed when the ads were never the problem.

I learned that last one the hard way too. Earlier this year, return on ad spend dropped in one of our markets. The marketing team looked at the ads and tested new campaigns. Nothing moved. Then I asked a different question: how are our technicians performing in that market? Same ads, same budget, same targeting. The only real difference was what happened after the call came in. We invested in technician training, close rates went up, and ROAS followed. I wrote the full story in Your ROAS problem isn't a marketing problem.

A blended lead number would never have shown that. Two clocks, split by market, would have pointed straight at it.

Step 1: Tag the clock at intake

Whoever answers the call, a person or an AI receptionist, should label the lead with its clock right then: now or decision.

This sounds obvious, and almost nobody does it well. The usual problem is that nobody writes it down.

I saw this exact problem in my own company this week, in a different form. When I asked my AI COO to sort a year of our HVAC sales by job type, more than half of the jobs could not be classified, because the booking notes never said whether the home had AC before. I wrote about it in The Agentic CRM. The lesson for lead measurement is the same: if the clock is not captured at intake, no report and no AI can rebuild it later with confidence.

So make it a required field, not a sentence in a note. Two values are enough to start. If you want a third, add "unclear" so people do not guess, and review the unclear ones every week.

A few simple rules help:

• Anything where the customer wants a technician today or tomorrow is on the now clock.

• Anything that needs an estimate the customer will compare or think about is on the decision clock.

• A now-clock repair visit can create a decision-clock lead. The technician finds an old system and the customer wants to hear about replacement. That is a new lead with its own clock, not an extension of the repair.

That last point matters. Many of the best replacement opportunities start as a repair call. If you do not open a new lead at that moment, they disappear inside the repair ticket.

Step 2: Start and stop each clock at the right moment

The now clock starts when the phone rings or the form is submitted. Not when someone gets around to opening the lead. It has a few checkpoints:

1. Answered live, or called back.

2. Booked.

3. Technician arrived.

4. Job completed.

5. Did it stick? No callback for the same problem, no cancellation, no complaint in the next few weeks.

The decision clock starts when the estimate is presented, not when the first call comes in. The time between the call and the estimate visit belongs to scheduling, and you should watch it, but the customer's decision really begins when they have a price in their hands. Its checkpoints:

1. Estimate presented.

2. First follow-up done.

3. Decision: signed, or lost with a reason.

4. Installed.

5. Still a customer at 90 days, and ideally on a maintenance plan.

Notice the stop rule on the decision clock: signed, or lost with a reason. Never "unconverted after 14 days." Every decision-clock lead stays open until someone closes it one way or the other and writes down why. Price, timing, chose a competitor, decided to repair, could not reach them. That list is one of the most valuable things in your business, and most companies throw it away.

Step 3: Give each clock its own targets

Owners often ask me for "the right numbers." I am not going to invent benchmarks. Every market, trade and season is different, and a target copied from someone else's slide is worse than none. Here is how to set your own.

On the now clock, targets are about time and completion. How fast should a ringing phone be answered? How quickly must a missed call get a callback? What share of urgent calls should be booked the same day? Pick numbers your team can hit on a normal day, measure every day, and tighten them over time.

On the decision clock, targets are about follow-through and quality. How soon after an estimate does the first follow-up happen? How many touches before a lead can be closed as lost? What share of estimates reach a clear decision, yes or no, within your normal cycle? What share of new install customers are still with you, and on a plan, after 90 days?

Then define what "normal" means for each clock, using your own history. If your replacement decisions usually take about three weeks, judging them at day 7 is meaningless. If your urgent calls usually book in minutes, judging them weekly is far too slow.

That is what I meant in May by measuring each offer against its own baseline, not against each other.

Step 4: Put both clocks on one page, side by side

The goal is one page with two honest columns. Here is the template I would use for a $1 to $5 million company. The numbers below are made up for illustration only. They are not our numbers and not an industry benchmark.

Weekly demand report (illustrative example)

In this example, 18 urgent calls were not answered or called back in time. That is a staffing or process problem, visible this week, not at month end.

On the decision side, 9 open leads are fine if they are inside your normal cycle. But 6 of 24 estimates missed the first follow-up, and 4 leads are past the normal cycle. That is a coaching conversation, not a reason to change the ads.

Blend the two columns into one "conversion rate" and all of that disappears.

Step 5: Split by source and by market

Once the two clocks are in place, cut each one two more ways.

By source. Referrals, past customers, paid search, your website, the sign on the truck: each sends a different mix of now and decision leads. A source feeding mostly decision-clock leads looks slow on a blended report and great on a 90-day view. A source sending lots of urgent calls looks busy, so check how many became completed jobs that stuck. Judge each source on the clock it mostly feeds.

By market, technician and team. This is the lesson from my ROAS story. If one market is weak on the now clock, look at answering and dispatch. If it is weak on the decision clock, look at who is presenting estimates and how they follow up. If both are weak in one place and strong everywhere else, the problem is almost certainly what happens after the call, not the marketing.

Step 6: Coach each clock differently

Now-clock coaching is about speed and reliability. Answer live. Call back fast. Book while the customer is still on the phone. Confirm the appointment. Show up when you said you would. The people who win this clock are organized and fast.

Decision-clock coaching is about patience and follow-through. Present a clear estimate with options. Agree on a next step before you leave the house. Follow up when you promised, not five times a day. Respect that the customer is comparing. Close the lead honestly, with a reason, when it is over. The people who win this clock are trusted, not pushy.

In May I wrote that we stopped pressuring careful buyers to close faster just to hit weekly numbers, and stopped letting fast buyers sit for weeks. That is the balance here: each clock moves at its natural pace.

Where AI helps, and where it does not

AI makes this system much easier to run. An AI receptionist can suggest the clock while the call is happening and flag the unclear ones. Agents are good at noticing that an urgent call waited too long or an estimate has gone quiet, and alerting the right person. They can read every lost lead each week and tell you which reasons are growing.

What AI does not do is set your targets, decide who owns each clock, or have the conversation with the customer. And none of it works if the clock label, estimate date, follow-up and lost reason are not real fields people fill in. The record comes first, then the reports, then the AI on top.

How to start next Monday

If you run a $1 to $5 million company and you want this in place quickly, here is the order I would follow.

1. Add one required field at intake: now or decision, plus "unclear."

2. Write down the start and stop rules for each clock on one page, and share it with everyone who touches a lead.

3. Make "lost with a reason" the only way to close a lead that did not convert. No more silent expiry.

4. Open a new decision-clock lead every time a repair visit turns into a replacement conversation.

5. Pull your own history and write down what a normal cycle looks like for each clock.

6. Build the one-page weekly report with two columns, then split it by source and by market.

7. Name an owner for each clock. One person is accountable for answering and booking. One person is accountable for estimate follow-up.

8. Review it every week for a month before you change any ads or budgets.

Most owners skip straight to changing the ads. Do these eight steps first. You may find, like I did, that the ads were fine.

The bigger point

Every owner wants one number that says whether marketing is working. But service demand is not one thing. It is a customer who needs you in the next hour and a customer who will decide over the next month, often in the same afternoon, from the same ad.

Two leads, two clocks. Measure each on its own clock, put them side by side, and you will stop pressuring the careful buyers, stop missing the urgent calls, and stop blaming marketing for what happens after the phone rings.

Conversion speed is not the same as conversion quality. Build a measurement system that can tell the difference.

FAQ

How should home service companies measure lead performance?

Split leads by clock. Measure repair and urgent calls on speed and completion: answered, booked, run, and whether the job stuck. Measure replacement and install leads on follow-through and quality: estimate presented, follow-up done, signed or lost with a reason, and 90-day retention. Report both side by side with their own targets.

When should the clock start on a replacement lead?

Start the decision clock when the estimate is presented, because that is when the customer's real decision begins. Track the time from the first call to the estimate visit separately, as a scheduling measure.

Why is a single lead conversion rate misleading?

Because it mixes leads that should be judged in minutes with leads that should be judged in weeks. Slow, careful decisions look like failures, slow answering on urgent calls gets averaged away, and good sources of big-ticket demand can get cut for looking "slow."

What should a weekly lead report for a home service business include?

For each clock: new leads, the right checkpoints (answered and booked for urgent calls; estimates presented and follow-ups for big-ticket leads), completed or decided, leads stuck past the normal cycle, a quality check, and the top lost reasons. Then split it by source and by market.

Can AI help measure leads?

Yes. AI can suggest the clock at intake, watch timers and alert people when a call or follow-up is late, and summarize lost reasons each week. It cannot set your targets or own the customer conversation, and it only works if the key fields are captured.