Attribution Modeling for Contractors: Credit the Right Channel When Leads Take 8 Days to Close
Marketing attribution is the practice of assigning conversion credit to the channels and touchpoints that influenced a booked job. For contractors with multi day sales cycles, last click attribution misreads the picture: it credits whichever channel was closest to the phone call, not the search, the review, or the ad that actually convinced the homeowner to buy.
A roof replacement is not an impulse purchase. A homeowner notices a stain on the ceiling, searches, reads reviews, asks an AI assistant for local options, gets two quotes, waits for payday, then calls back. That sequence can stretch across eight days and five touchpoints. If your reporting says "branded search" because that was the last click before the form, you are about to defund the organic content and the display ad that did the persuading.
What is marketing attribution, and why does it break for service businesses?
Attribution modeling is a way of measuring the impact of marketing activities across the customer journey by analyzing interactions with various touchpoints and working out how much each contributed to the conversion, as AppsFlyer defines it. The concept is straightforward. The application to contracting is not.
Three things make service businesses harder to attribute than ecommerce:
- The conversion is offline. The revenue event is a signed estimate or a completed job, not a checkout page. Your ad platform never sees it.
- The gap is days, not minutes. Cookie windows expire, people switch from phone to laptop, and the session that produced the lead is not the session that produced the decision.
- Job values differ wildly. Twelve drain cleanings and one sewer line replacement are not the same outcome, even though both show as "1 conversion."
TruLata builds marketing data and attribution tooling for local service businesses across the United States: contractors, home service companies, and trades operations that need lead counts from their own forms rather than from the ad platform's estimate. That distinction matters more than any model choice, because attribution built on inflated platform conversion counts is precise arithmetic on the wrong numbers.
Why platform reported conversions inflate your best channel
Every ad platform is incentivized to claim credit. Google Ads counts view through and cross device conversions inside its own attribution window. Meta does the same. Run both, and the sum of platform reported leads routinely exceeds the number of forms that actually hit your inbox. When you start with lead tracking counted from your own forms, you have a fixed denominator that every model has to divide, which is the only way two channels can be compared honestly.
First click vs last click attribution: which one should a contractor use?
Neither, exclusively. First touch attribution assigns 100% of credit to the very first interaction a customer had with your brand, while last touch does the mirror image. As Kissmetrics puts it, first touch tells you what generates awareness and last touch tells you what drives conversion, and relying exclusively on either one will lead you to systematically under invest in channels that are quietly driving enormous value.
For contractors, the failure modes are specific and predictable.
What last click gets wrong
Last click overcredits branded search and retargeting. Someone who already decided to hire you types your company name into Google, clicks the branded ad, and fills out the form. Last click hands that job to paid search. But the homeowner decided three days earlier, reading a service page they found through an unbranded query. Cut the content budget on last click logic and your branded search volume quietly drops two months later, with no obvious cause in the dashboard.
What first click gets wrong
First click overcredits top of funnel discovery and ignores the closing work. A blog post that introduced the homeowner to your business gets full credit, while the review profile that beat your competitor and the follow up email that recovered a cold quote get zero. That is a fast way to underfund the parts of the funnel where jobs are actually won, which is usually why your best leads go cold after a quote.
What are the main marketing attribution models, and which fits an 8 day cycle?
Attribution models provide a set of rules or algorithms for determining which touchpoints along the buyer's journey are responsible for conversion, as Amplitude frames it. They fall into two categories: single touch (first or last click) and multi touch attribution, which splits credit across the path.
Linear
Every touchpoint gets equal credit. Five touches, 20% each. Useful as a baseline sanity check because it is the least opinionated model, and useful for spotting channels that appear constantly in winning paths but never as the first or last click. Weak point: it treats an accidental homepage visit the same as a 12 minute pricing page session.
Time decay
Touchpoints closer to the conversion get more credit, with earlier ones decaying. This is the most defensible default for contractors with cycles measured in days rather than months, because urgency is real in the trades and the touches near the call genuinely matter more. Set the half life to roughly match your observed cycle: if your average lead closes in eight days, a short half life distorts the picture.
Position based (U shaped)
Typically 40% to the first touch, 40% to the last, 20% split among the middle. This works well for emergency and replacement services where discovery and the final decision are the two hinge moments, and the middle is mostly comparison shopping.
The practical recommendation
Run two models side by side, not one. Compare time decay against last click. Where they agree, act with confidence. Where they disagree by more than a small margin, you have found a channel that is either being robbed of credit or stealing it. As HockeyStack notes, the best attribution model is the one that aligns with your business goals, customer journey, and resources, and it is worth experimenting and refining over time rather than locking in one framework. Seeing both at once is exactly what the Command Center dashboard is built to show.
How do you actually build multi touch attribution for a contracting business?
You do not need a data science team. You need four things wired together correctly and refreshed often enough to trust.
Step 1: Capture the source on the form, every time
Add hidden fields to every lead form that persist UTM parameters, referrer, landing page, and a first touch timestamp stored in a long lived cookie or local storage. Persist first touch for at least 30 days. If you only capture the last session, you have built a last click system by accident, whatever model you claim to run.
Step 2: Track phone calls with the same rigor as forms
In the trades, a large share of high value jobs arrive by phone. Use dynamic number insertion so the number a visitor sees is tied to their traffic source, and route those calls into the same record as your form leads. Untracked phone numbers on a service page are the single largest attribution blind spot most contractors have.
Step 3: Close the loop from lead to signed job
Push the source data into your CRM and write the outcome back: quoted, won, lost, and job value. Attribution that stops at "lead" ranks channels by volume, which is how contractors end up scaling a campaign that produces plenty of tire kickers and few signed contracts. If your systems do not talk to each other, that is a case for CRM automation that stops losing leads in your pipeline.
Step 4: Measure your actual lag, then set your reporting window
Calculate the median days from first touch to signed job, by service line. Emergency plumbing might be under a day. Roof replacement might be several weeks. Then stop judging any campaign before that window has fully elapsed. Reviewing a roofing campaign after seven days when your median cycle is 21 days means you are grading a test where two thirds of the answers have not been submitted.
Why does data freshness decide whether attribution is usable?
An attribution model recalculated once a month is a history lesson. If a campaign started misfiring on the 3rd and you see it on the 30th, the model was mathematically correct and commercially useless. This is why live marketing data, refreshed when you open it is not a cosmetic feature: the value of attribution is entirely a function of how fast it changes your decisions.
Equally important is honest labeling of lag. Some sources update in near real time, others settle over 24 to 72 hours. A dashboard that shows every number with the same confidence teaches you to distrust all of them. A dashboard that tells you which numbers are still settling teaches you when to act and when to wait.
What should contractors do with attribution findings?
Attribution is only worth building if it changes budget. Three moves that follow directly from a multi touch view:
- Protect assisting channels. Any channel that appears in 30% or more of winning paths but rarely as the last click is an assist engine. Do not cut it on last click math.
- Segment by service line, not just channel. Paid search may dominate emergency calls while organic and AI answer engines dominate planned replacements. One blended number hides both truths.
- Weight by job value, not lead count. A channel producing fewer, larger jobs can beat a channel producing triple the leads. Attribution without revenue weighting is a popularity contest, which is the core argument in calculating marketing ROI when every job value is different.
Do not forget the touchpoints that leave no click
An increasing share of discovery now happens inside AI assistants, where a homeowner asks for a recommendation and gets a named answer with no referral click. These touches influence conversions and are invisible to every click based model. Tracking whether your business is named in those answers gives you a proxy signal for a channel your attribution model will otherwise score at zero.
Credit the channel that convinced, not the one that answered last
Attribution modeling is not an accounting exercise. It is the difference between scaling the campaign that wins jobs and scaling the one that happens to sit closest to the phone. Start with lead counts from your own forms, capture first touch and persist it, close the loop to job value, and compare at least two models before reallocating a dollar.
If you want to see multi touch attribution built on your real form submissions rather than platform estimates, book the live demo and walk through your own channel paths with the data refreshed in front of you.
