AI Quoting Tool for Landscaping Companies: Pressure-Testing the Numbers Behind the Quote

Landscaping quotes drift when the pricing lives in people's heads. That was the situation at one landscaping and hardscape company: quotes were inconsistent, margins were unknown, and the logic behind a price had never been written down in one place.

The company already had a custom GPT for sales quoting. We spent a call pressure-testing it with the owner, a sales lead and an office and marketing manager. If you run one of the many home service businesses that quote by feel, the steps below show what a real test of an AI quoting tool looks like.

The situation

The call covered a filming plan and a shot list, but the center of it was the quoting workflow. The team had been building quotes with a custom GPT, and the goal was to tighten margins and make quoting faster and more consistent.

The test was designed to show how much each person's prompts changed the price. Three people with three sets of instructions can produce three slightly different prices for the same kind of job.

What was missing

A quoting tool can only be as accurate as the inputs it is given. Here the tool did not yet have the company's real labor costs or its typical subcontractor margins to work from, and the team was asked to flag where the numbers felt off.

The knowledge was also stuck in conversation. The owner knew how labor should be priced, but that reasoning had not been captured in the tool.

What we did

We ran the tool the way a skeptical customer would, with real jobs and real people.

• Same job, different hands: each team member was asked to run an independent test quote in the sales-quoting GPT, then at least one more, and note where the numbers felt off, so we could compare how each person's prompts changed the price.
• A margin check on a real quote: one real landscape quote's full GPT thread was emailed to us so we could run a margin-health analysis on it.
• Cost inputs: the owner was to pull a rough hourly cost, including overhead and insurance, for six to eight field staff, plus typical subcontractor margins, for us to wire into the GPT.
• A recorded Q&A: a 20-minute labor-pricing question-and-answer session was scheduled to be recorded and added to the GPT's knowledge base, so the owner's reasoning sits inside the tool.
• Quoting from the truck: the estimator was asked to confirm that the shared GPT opens on a phone, so a quote can be built on site.

What changed

No margin or speed result was reported. This was a testing call, and the review of the test results was set for later that week. What changed is where the pricing logic lives.

Before the call it was private judgment. By the end of it, the logic was moving into written inputs: labor cost with overhead and insurance, subcontractor margins, and a recorded explanation from the owner. The tool could be tested by three people against the same jobs, so the differences between their results would become visible instead of hidden.

How to pressure-test your own quoting tool

This section is advice for your own business. Give the same job to two people and compare the prices. If they differ, the gap tells you which input is missing.

Then take one real, finished quote and check the margin against your actual costs. Finally, confirm that the people who quote can open the tool on a phone, because a quoting tool nobody can reach on site will not get used.

Frequently asked questions

How do you test an AI quoting tool?

Give the same job to more than one person and compare the prices. Here each team member was asked to run an independent test quote, then another, and report where the numbers felt wrong. Differences between results point to missing inputs or loose prompts.

What data does an AI quoting tool need?

At minimum, the real hourly cost of labor including overhead and insurance, and typical subcontractor margins. We asked for both here, and planned a recorded 20-minute pricing Q&A so the owner's reasoning could sit in the knowledge base.

Can crews use an AI quoting tool in the field?

That was a goal here. The estimator was asked to confirm the shared GPT opened on a phone so quotes could be built from the truck. Check phone access and sign-in before you roll a tool out to the crew.

Why check the margin on a real quote?

A real quote shows whether the price leaves a healthy margin once actual costs are applied. We asked for one full quoting thread to run a margin-health analysis on. A single quote will not prove the tool, but it shows you where to look first.

If your quotes depend on who happens to write them, book a 30-minute AI follow-up and reporting audit. We look at how you price today and show you where an AI tool can help and where it needs better inputs first.