Measure the same capture and reporting job twice.
This is our published method for comparing a team’s existing process with SiteWarden. No measured benchmark result is published yet. Use the worksheet to record what happens with the same records, evidence and review requirements in both workflows.
Product and record-keeping guidance by SiteWarden. Read the linked source in its stated jurisdiction and project context, and use the project’s authorised people for contract, technical or compliance decisions. How these guides are reviewed.
Download the benchmark worksheet.
Editable trial log for the published method. Enter observed phase times and add them for the total; no benchmark result is prefilled. The six blank observations cover three trials for each workflow. Record the test conditions and output checks alongside the timings.
Download the capture-to-report benchmark worksheet (CSV)Prepare five fixed illustrative items
Prepare the same source notes and redacted photos for both workflows. Freeze each item’s identifier, location, element, observation, trade and priority so the final output can be checked against the source.
- B01: Bedroom 1 door edge, chipped finish, joinery, medium priority.
- B02: Kitchen cabinet door, alignment needs review, joinery, medium priority.
- B03: Bathroom junction, unfinished sealant detail, tiling, medium priority.
- B04: Living room wall, paint touch-up, painting, low priority.
- B05: Hall door, missing doorstop, carpentry, low priority.
Write down the baseline and SiteWarden tasks
Document the team’s current capture, correction and report process, then freeze it for the trials. Use the same item order and required PDF content. This measures record preparation from fixed inputs; it does not measure inspection, repair quality or rework costs.
In SiteWarden, record each item with the prepared media and notes. If using AI-assisted voice or video, review the suggested fields and correct them against the source. Keep the original evidence. Select all five records, including outstanding items, generate the PDF and check its output. Approval remains with the authorised project team.
Time capture, review and reporting separately
Capture starts immediately before preparing the first record and ends when the fifth record is saved. Review starts when checking the first saved record against the source and ends when the last required correction is saved. Report time starts when selecting or assembling the output and ends when the PDF passes the agreed checks.
Log interruptions, retries and failed output checks. Keep them visible rather than silently removing slow runs. Total minutes are capture plus review plus report minutes. Compare each phase as well as the total; a fast capture that leaves more correction work should remain visible.
Repeat the runs and check equivalent output
Complete an untimed familiarisation run, then three timed trials for each workflow. Alternate which workflow goes first: baseline then SiteWarden, SiteWarden then baseline, baseline then SiteWarden. Reverse that starting order for another participant. Reset to a fresh list each time and retain the source set and exported PDFs.
Check all five identifiers, observations, locations, trades, priorities, current statuses and evidence references. Images must match their items and be readable. Apply the same completeness checks to both PDFs. Mark a failed check and its correction time; do not compare an incomplete output with a complete one.
Record conditions before interpreting results
Record the date, plan, app version, device, connection, installed AI model if used, prior workflow familiarity and project scope. Use an anonymous participant label and controlled or redacted evidence. Keep installation and setup time separate, and document the tested capture mode and any online requirements.
Publish each trial, median phase times, completed-output count and the actual baseline, with participant counts, conditions and limitations. These instructions establish neither hours saved nor rework-cost reductions. The calculator’s adjustable 30% input remains an estimated planning target.
Has SiteWarden published measured time savings?
No measured capture-to-report benchmark result is published yet. This page provides the method and worksheet for collecting comparable observations. Any future result needs its baseline, trial count, conditions and output checks.
Does this test measure a 30% reduction in rework costs?
No. It times record capture, review and PDF preparation from fixed illustrative inputs. Rework-cost savings require separate project cost evidence. The adjustable 30% calculator input is an estimated target, not a measured SiteWarden result.
Which SiteWarden features should be included in the test?
Test current capture, reviewed AI assistance and PDF output in the selected plan. Mobile capture works offline and syncs online; desktop and online features need connectivity. API access is planned and excluded from this test. Record the plan and conditions.
Bring your list.
We’ll walk you through it.
Unfinished work, defects and the details in between. See how the current app captures, organises and reports on them.