You Can't Automate a Report Nobody Agrees On
Ask five PMs for percent complete and you get five defensible numbers. The fix is not a dashboard, and it is definitely not AI. It is a one-page definitions sheet, written before you automate anything.
Blog
Practical writing on construction operations, project controls, and AI that earns its keep. No hype, no robot stock photos.
Ask five PMs for percent complete and you get five defensible numbers. The fix is not a dashboard, and it is definitely not AI. It is a one-page definitions sheet, written before you automate anything.
A WIP report tells you whether the jobs make money. A 13-week cash forecast tells you whether payroll clears on a particular Friday. Here's how to build one from real draw lags, holdback dates and a few minutes of PM input each week.
Your finished jobs already know what a metre of pipe or a cubic metre of formwork really costs your crews. Here's how to get that number out at closeout and into the next bid, in a spreadsheet, without buying a database.
Most wrong-site mornings aren't a whiteboard problem. They come from a plan with no cutoff, no single owner and late changes posted to a group chat. Here's a daily dispatch routine that holds up past the third active site.
A subcontractor's progress claim lands on the PM's desk three days before the owner billing goes out. Here is a line-by-line review that catches front-loading, missing change order paperwork and holdback errors without turning it into a week of back and forth.
Owned equipment usually sits in overhead, which means it belongs to no job and shapes no bid. Here's how to build an internal rate from your own ledger, capture the hours without buying a platform, and reconcile the two once a year.
The estimate handover doesn't fail because a document is missing. It fails because a number built to win the job never gets translated into codes, rates and dates a crew can work from. Here's how to do the translation in one sitting.
Every AI tool sold to plumbing, HVAC and electrical contractors sits in one of three tiers, sorted by what it needs from you before it can do anything. Knowing which tier you're looking at tells you whether it will work in your shop next month or get cancelled after the trial.
Payroll hours get checked by the people being paid. Job cost hours get checked by nobody until month end. Here's how to capture field time once, daily, by cost code, so both reports come from the same record and can't disagree.
Sign-out sheets and office spreadsheets fail because tools move between sites without passing through the yard. Here's how to number, tag, and track custody so tools stop walking, and when you actually need to buy software for it.
Long-lead materials arrive late because the order date was set by when someone got around to it. Run the procurement log backwards from the crew's need-by date: install, on-site, release, submittal due.
Most contractors track labour hours carefully and installed quantities never. That's half a productivity ratio. Here's how to collect the other half weekly, from the field, so the cost report can forecast instead of just recording.
Most RFI logs track the date sent and the contract turnaround, which is why the answer lands a week after the crew needed it. The log that works carries a need-by date pulled from the look-ahead, one question per RFI, and a proposed answer the consultant can simply approve.
A 200-item punch list at final walk isn't a discovery problem, it's a timing problem: everything got checked once, at the moment it's most expensive to fix. Tying deficiency checks to the trade milestones you already have keeps the list small and the holdback clock from becoming a bet.
The schedule of values gets treated as contract paperwork, but it's the document you bill against every month for the life of the job. How to structure one with measurable lines, honest mobilization, and a clean map to your cost codes, so progress claims stop being arguments.
Most job costing fails at the timesheet, not the software. How to build a cost code list sized for the foreman filling it in, matched to your estimate, and short enough to be coded right.
Most three-week look-aheads are a filtered export of the master schedule, which is why crews stop reading them by Wednesday. The version that works screens every activity for what has to be true before it can start, and commits to nothing else.
Change order margin leaks through three gaps: never priced, never approved in writing, or approved and never billed. A working log, a named owner, and a fifteen-minute review before every progress claim close all three.
Costs to date are history. Cost to complete is the one input in a WIP schedule that is a forecast, and it carries the entire report. How to build a number that holds, and why asking a project manager for percent complete gets you the wrong answer.
Contractors keep buying AI estimating tools and seeing the same numbers a quarter later. The reason is usually that the tool got bolted onto a bid process nobody redesigned. Here is how to take one workflow apart properly, and what to measure once you have.
Every article about outgrowing QuickBooks is written by someone selling the replacement. Here's how to tell whether your problem is the software or the process, and what the realistic options actually cost, with no pitch attached.
BC's Construction Prompt Payment Act starts fixed payment clocks, formal notices, and 30-day adjudication. The firms that struggle won't fail on the law. They'll fail on their own month-end billing process. What to fix during the runway.
Crews write down progress, delays, and crew counts every single day, and then the office rebuilds the same information from memory. How to turn the daily report from a compliance ritual into the most useful data you collect.
Most small contractors freeze on AI because deciding whether it's safe feels like an IT-department question. It isn't. Three plain questions about any tool, and a setup a twenty-person firm can actually run.
By six weeks in, the schedule and the cost report almost never agree, and it is rarely one bad mistake. Here is where the drift actually comes from and how to close it without a six-figure platform rollout.
Most consulting engagements start by picking a tool. Ours starts with a stopwatch. Here's why the order matters more than the tool you land on.