Estimate Analysis

What Estimate Analysis is

Estimate Analysis checks one estimate's lines against your own past estimates for the same line item, and flags the ones that look unlike how you have billed the same work before. It is an error-catching lens on your own consistency: the dehumidifier line priced at twice what you usually charge per day, the air-mover count scoped at three times your norm for the same kind of loss, and the line you include on almost every job of this type that is missing from this one. It never compares you to a peer cohort, and it never prescribes a target price. The question it answers is not "what should this cost," it is "is this estimate consistent with your own book," which is exactly the question to ask before an estimate goes out the door with a typo in it.

Two things make it different from every other analyst app. It is deterministic: the outliers are computed statistically from your own history, no AI model runs, so a result is a fact about your book, not a judgment call. And because no model runs, it costs no Intelligence Units and shows no consent step; you pick an estimate, press the button, and the result lands.

Where to find it

Open Jobs from the sidebar. In the Apps row, Estimate Analysis is the teal tile reading "Lines priced unlike your own past estimates," sitting beside Job Performance. Clicking it opens the app as a card over the page. There is also a per-line version of the same check: on a job's Line Items view, individual lines carry their own outlier read, covered in job line items. This app is the whole-estimate view, one keepable finding across every line at once.

Running it

The card opens on a list under the heading Pick an estimate to check: your jobs and bid-time estimate documents that have priced line items, each showing how many priced lines it carries and the date of its latest estimate. Pick one and press Analyze this estimate. The analysis runs in one pass and the result opens in the card, titled "Estimate review:" followed by the job or document name, saved like any other app result: pinnable, listed for re-reading, exportable as a PDF.

Reading the result

The finding sorts the estimate's lines into a few honest buckets:

• Flagged lines lead, most extreme first. Each one names the line, whether the price or the quantity is what stands out, your value on this estimate against the middle of your own history for that line, and how many of your own comparable lines sit behind that baseline. A flag is a fact plus an invitation to look; it is not a verdict that the line is wrong. Sometimes the answer is a typo, sometimes it is a genuinely bigger loss, and you are the one who knows which.
• Lines that check out are counted, so a clean estimate reads as a clean result rather than an empty one.
• Lines too new to your book to judge are counted separately. A line item you have rarely or never billed before has no personal baseline, and the honest treatment is to say so rather than to measure it against nothing.
• Possible omissions close the finding: lines you usually include on this kind of work that do not appear on this estimate. Leaving scope off an estimate is the quietest way to lose money on a job, and it is invisible without exactly this check.

Where an estimate's peril is governed by an IICRC standard, the result also names the governing standard as a reference pointer, so you know which rulebook applies without the app reproducing its text.

Before you have a baseline

Estimate Analysis is built entirely from your own history, so its cold states name what fills them, and it stays quiet rather than borrowing anyone else's numbers.

• No estimates to analyze yet. The card reads "No estimates with priced line items yet," and tells you exactly what to do: "Forward an estimate (the Xactimate .esx or the PDF) and it will appear here to analyze."
• Estimates in, but no history behind their lines. The result says it plainly: "This is early history for these lines, so there is no personal baseline to compare against yet. It builds as you send more estimates carrying these lines. Forward a few past estimates (the Xactimate .esx or the PDF) to seed it faster."

Both of these are gates you can close alone, and both name the artifact that closes them: estimates, from your estimating tool, as the .esx export or the PDF. The more of your past estimating history flows in, the more lines earn a baseline and the sharper every future check gets.

Best-practice example

An estimator finishes a water loss estimate on a Friday afternoon and runs Estimate Analysis on it before it goes out. The result flags two lines: the LGR dehumidifier priced at roughly twice the usual daily rate, which turns out to be a unit typo, and equipment days scoped at three times the norm for a loss this size, which turns out to be real, this one dried slowly, and is worth a note in the file for when the adjuster asks. One omission surfaces: the antimicrobial application that appears on nearly every comparable job is missing. Three minutes of reading catches one error, pre-answers one question, and recovers one line of scope, on an estimate that looked fine.

Data sources