Lead intent scoring ranks every inbound lead by how likely it is to become a booked job, so your team calls the homeowner with a burst pipe before the one "just looking at options." You score two things: how well the lead fits what you sell and how the person behaved. Then you act on the number fast, and check it against real booked jobs every month.
Most of what ranks for this topic is written for software sales teams. This guide translates it for HVAC, plumbing, roofing, electrical, landscaping and turf companies, where the lead is a phone call or a form fill and the clock matters more than a nurture sequence.
TL;DR
- Intent scoring combines fit (who the lead is and where they are) with behavior (what they did, and how recently). Either one alone misleads you.
- For home services, the strongest signals are usually a call, a service-specific page or ad, an urgent request and a service-area match. Weight them by what actually booked in your own data.
- Start with a simple points model in your CRM, then move to a machine-learning model once you have enough closed jobs to learn from.
- A score only pays off if it changes what happens next: who calls, how fast, and which leads Google Ads learns from.
- Review the score monthly against booked and paid jobs. A score nobody checks drifts.
Table of Contents
- What lead intent scoring actually measures
- The signals that matter for a home service lead
- Build a simple scoring model first
- When rules stop being enough: predictive models
- Feed the score back into Google Ads
- Keep the score honest
- Our perspective
- Where Leapify Media fits
- FAQ
- Sources
What lead intent scoring actually measures
Lead scoring in general is the practice of assigning a value to each prospect that reflects how likely they are to become a customer. HubSpot's explanation describes the point plainly: it lets a company focus its energy on the leads most likely to buy. Intent scoring is the part of that system built on behavior: pages viewed, calls made, forms submitted, how recently, and how often.
The distinction matters because two different questions get blended together:
- Fit: Is this someone you can serve and want to serve? Right service, right ZIP code, right property type, right job size.
- Intent: Is this person trying to buy now, or just browsing?
Salesforce's engineering team makes the same point from the machine-learning side: a model that looks only at engagement or only at profile data leaves sales teams guessing and spending time on prospects who are not actually qualified. A homeowner who clicks six of your ads but lives two counties outside your service area is high engagement and zero fit. A homeowner in your area who calls once at 7 a.m. about a dead furnace is high fit and high intent. Your score should tell those two apart without a human reading each lead.
Why "high intent" does not always mean "ready to book"
A common misconception is that a high intent score equals a ready buyer. Salesforce's training material notes that real-time intent scoring works because it follows current behavior, such as visits to pricing pages, rather than chasing prospects who merely look perfect on paper. But interest is not the same as a purchase decision. In home services, a homeowner reading about full roof replacement costs may be a year from signing. The score tells you who to call first. It does not guarantee the outcome.
The signals that matter for a home service lead
Software companies score webinar attendance and whitepaper downloads. You do not have those. Your leads arrive through different doors, and each door carries a different amount of intent. Here is a practical way to think about them.
| Signal | What it usually suggests | Typical direction |
|---|---|---|
| Phone call from a paid ad or Google Business Profile | Wants an answer now | Strong positive |
| Call that lasts more than a minute or two | Real conversation, not a hang-up | Strong positive |
| Form fill asking for "emergency" or "today" service | Urgent need | Strong positive |
| Form fill on a service-specific page (e.g., water heater replacement) | Specific job in mind | Positive |
| Lead in your service area | Fit | Positive; outside area is often disqualifying |
| Repeat visit within a few days | Comparing and narrowing | Positive |
| Visit to pricing, financing or "request a quote" pages | Evaluating cost | Positive |
| Blog-only visit, no contact | Research phase | Weak |
| Missing or fake phone number, spam patterns | Low quality | Negative |
| Job type you do not offer (e.g., commercial when you are residential) | Poor fit | Negative |
These weights are starting assumptions, not facts about your market. Your own booked-job data decides the real numbers, which is the whole point of the monthly review below.
Calls usually carry more intent than forms
A caller has spent effort and is usually mid-problem. We break down the trade-offs in calls versus lead forms for home service jobs. For scoring, the takeaway is to track the source and the outcome of both, then let the data set the weights instead of assuming. Google also offers call history in Google Business Profile, which keeps calls from Search and Maps in one place so missed calls can be answered. A missed call is a lead that decays fast.
Timing and recency
Intent decays. A lead from three minutes ago and a lead from three days ago should not share a score. Build recency into the model: the same behavior counts for more today than last week. The Salesforce engineering post and the research below both treat recent engagement as more informative than static profile data.
Build a simple scoring model first
You do not need machine learning to start. A points model in a CRM is better than gut feel, and it teaches you which signals matter.
- List your booking paths. Calls, forms, Local Services Ads messages, chat, text, referrals.
- Tag every lead with source, service requested, ZIP code and timestamp. If you cannot see this per lead, fix that first. Our guide to GCLID to CRM setup covers how to stop losing the ad click that created the lead.
- Assign fit points. Service you offer, inside the service area, job type you want. Make out-of-area and wrong-service leads a hard stop, not a few lost points.
- Assign intent points. Call, urgent wording, service-specific page, repeat visit. Subtract points for spam signals.
- Set three tiers. Hot, warm, cold. Three tiers are easier for a dispatcher to act on than a 0 to 100 scale that nobody remembers.
- Attach a response rule to each tier. Hot gets a call immediately. Warm gets a call or text within the hour. Cold goes into automated follow-up.
- Compare tiers to outcomes after 60 to 90 days. If your cold tier books as often as your hot tier, the weights are wrong.
Step 6 is where most scoring projects fail. A score that sits in a CRM field and changes nothing is decoration. If you need the follow-up piece, see our practical guide to automated lead follow-up.
A worked example
Say you run a plumbing company. Three leads arrive within ten minutes:
- Lead A: Calls from a Google Ads call extension, in your service area, talks for four minutes about a leaking water heater. Fit high, intent high. Hot.
- Lead B: Fills out a form on your "tankless water heater installation" page, in area, no urgency noted. Fit high, intent medium. Warm. A call within the hour, and the form should still get a prompt reply.
- Lead C: Fills out a form from a blog post about drain cleaning tips, ZIP code 90 miles away. Fit low, intent low. Cold, or disqualified.
Without scoring, a busy office might return them in the order they came in, or whichever blinks loudest in the inbox. With scoring, Lead A goes to your best call handler first. That is the entire value of the exercise: the right person, the right order, no wasted minutes.
When rules stop being enough: predictive models
Points models are transparent, and you can explain them to a dispatcher in a minute. Their weakness is that you set the numbers by hand and have to keep adjusting them. HubSpot notes that predictive scoring removes much of that manual tweaking by using machine learning to find conversion patterns across many data points, including ones a person would miss.
There is academic support for this. A 2025 paper on B2B lead prioritization, published in PMC, tested machine-learning models, including gradient boosting classifiers, against traditional approaches. It found they identified high-quality leads more accurately by weighing features such as lead source and lead status from actual conversion patterns. That study looked at B2B companies rather than home service businesses, so treat it as evidence for the method, not a promise about your results.
A predictive model has prerequisites:
- Enough outcomes. It learns from leads that booked and leads that did not. A company with a few dozen closed jobs does not have enough to train on.
- Clean data. If your CRM marks every lead "new" forever, the model has nothing to learn from.
- Closed-loop outcomes. Booked, completed and paid status must flow back to the lead record.
If you are below that threshold, run the points model and collect clean outcomes. You are building the dataset a predictive model will need later.
What Google is doing with intent scores
Intent scoring is now showing up inside the ad platforms themselves. Google describes a lead intent score feature that categorizes submissions as high, medium or low intent using Google's own signals, so you can prioritize outreach without manual vetting. This is useful as one input. It does not know your service area rules, your job margins or which technicians are free, so it works best alongside your own scoring and not instead of it.
Feed the score back into Google Ads
Scoring does double duty when it flows back to the ad account. If Google only learns that a form was submitted, it treats a spam lead and a $14,000 install inquiry the same. If it learns which leads turned into real jobs, it can bid toward more of those.
Google's documentation explains that its qualified lead and converted lead goals replaced the older imported-leads category so advertisers can measure deeper lead-to-sale events. Its offline conversion help page recommends choosing either qualified lead or converted lead as the goal when setting up enhanced conversions for leads. Enhanced conversions for leads supplements imported data with hashed first-party details like email or phone number to improve accuracy. The developer docs, updated this month, describe the same approach as an upgrade to offline conversion import.
In practice:
- Send a qualified lead event when a lead reaches your hot or warm tier and a human confirms it is a real job opportunity.
- Send a converted lead event when the job is booked or paid.
- Do not send every form fill as a conversion, or your score has no influence on bidding.
The measurement plumbing is covered in our post on GA4 lead tracking for home services. Trade-specific playbooks are here for HVAC, roofing and turf installers. If Local Services Ads are part of your mix, see how we approach LSA management, since those leads arrive through a different flow.
Keep the score honest
A scoring model is a hypothesis about what a good lead looks like. Gartner Peer Community contributors note that a lead score does not need to be set in stone and works best with a regular two-way feedback cadence: sales reports on real lead quality, marketing adjusts targeting and follow-up. For a home service company, "sales" is your call center or dispatcher, and "marketing" is whoever manages your ads.
A monthly review takes about an hour:
- Pull leads by tier for the last 30 to 90 days.
- Compare booking rate and average job value per tier.
- Ask the people on the phones which hot leads were junk and which cold ones were gold.
- Adjust weights, then note what you changed so you can see its effect next month.
Watch for two failure modes. The first is a score that rewards the wrong behavior, like heavy clicking from comparison shoppers. The second is a team that stops trusting the score because of a few visible misses, and goes back to calling in order of arrival. Showing the tier-to-outcome numbers resolves most of that.
Common mistakes
- Scoring on fit only. You end up chasing a perfect-looking homeowner who is not shopping.
- Scoring on activity only. You over-reward browsers.
- No disqualifiers. Out-of-area and unserved-job leads should drop out, not just sink a few points.
- No speed rule. A hot lead that waits 40 minutes is a lead the next company may have already booked.
- No outcome data. Without booked and paid status in the CRM, nothing can be validated.
Our perspective
We see the same pattern with home service owners who come to us after spending heavily on ads: the ads are generating leads, but nobody can say which ones are worth a fast call. The fix is rarely more traffic. It is a system that tells your office, in the first seconds, which lead deserves attention and then gets it there.
That is why we build lead scoring and dispatch around home service workflows, not generic sales funnels. A homeowner with no heat in January behaves differently from a software buyer downloading a guide, and the model has to know the difference. We engineer our AI in-house and run it on on-premise servers, so your lead and customer data stays proprietary and is not passed through a third-party AI tool. You also work with senior strategists directly, because tuning a score to your service area, your trades and your capacity is judgment work, not a template.
Our view on tooling is simple: start with the cleanest tracking you can, use a points model until you have the outcomes to justify prediction, and never let a score run without a monthly look at what it is doing to booked jobs.
— Everson Gorski
Where Leapify Media fits
If you would rather not build this yourself, our On-Premise AI Dispatch service handles intent scoring and lead dispatch for home service companies. It is $1,000 per month, with a one-off AI Dispatch Setup fee of $7,500. CRM Integration, which connects scoring to your booking and follow-up data, starts from $3,000 one-off. Our packages, Foundation from $2,500 per month, Growth from $7,500 per month, Scale from $15,000 per month and Enterprise from $25,000 per month, bundle ads, SEO, content and automation. You can see everything on our pricing page, or browse our services.
We work with home service businesses doing at least $500K in annual revenue, where the volume of leads makes prioritization worth the effort.
FAQ
How is a lead score calculated?
You add points for fit signals (right service, right area, right job type) and intent signals (a call, an urgent request, a visit to a quote page, recent activity), and subtract points for negative signals like spam patterns. The total places the lead into a tier. Predictive models skip the hand-set points and learn the weights from past leads that booked and leads that did not.
What is the difference between lead scoring and intent scoring?
Lead scoring is the broader practice of ranking prospects by likelihood to buy, usually blending fit and behavior. Intent scoring focuses on the behavioral side: what the person did and how recently. The strongest models use both, because high engagement without fit wastes time.
How many leads do I need before using a predictive model?
There is no universal number, but a model needs a meaningful history of both booked and unbooked leads with clean outcomes attached. If your CRM does not record booked and paid status reliably, start with a points model and fix your data first. A model trained on messy outcomes will reproduce the mess.
How often should I update my scoring model?
Review it monthly against booked jobs and job value by tier. Change weights when the data shows a tier underperforming, and record each change. A lead score is a working assumption, and it needs regular feedback from the people who answer the phones.
Sources
- Lead Scoring Explained, HubSpot
- Machine Learning Driven Sales and Marketing with Einstein Behavior Scoring, Salesforce Engineering
- Empower Sales Teams with Data-Driven Strategies, Salesforce Trailhead
- The relevance of lead prioritization: a B2B lead scoring model based on machine learning, PMC
- Lead Intent Scores, Accelerate with Google
- About qualified leads and converted leads, Google Ads Help
- Offline conversion imports FAQs, Google Ads Help
- Manage offline conversions, Google Ads API
- Call history in Google Business Profile
- Sales Development Representatives and Pipeline Generation, Gartner Peer Community