The week our scheduling broke, the trigger was the smallest possible thing: an electrician booked twice across two sites. We caught it on Monday morning. By Wednesday afternoon we'd lost three days of crane time, two pours had been pushed, and the curtain-wall subcontractor was sending us emails with red-bordered headers. Nothing about the underlying schedule was wrong — except that the schedule itself was wrong, because nobody had ever been looking at the same one.
This is the story of how we replaced a thirty-tab Excel workbook with a single shared timeline. It is not the story of a software rollout. The software took two weeks. The cultural shift took eight months, and we are still inside it.
The thirty-tab problem
Our pre-construction team had been running every project the same way since 2008: a master Excel workbook with one tab per package, hand-edited by the project director, distributed weekly as an attachment. It worked when projects were 18-month single-site builds. It stopped working the day we crossed our first multi-phase, multi-borough programme — which happened in 2016, though we kept pretending otherwise until 2024.
The Excel approach has three failure modes, and over a long enough time horizon every project hits all of them:
- The single source of truth is whoever opened the file last. One missed merge and you have two truths. We had projects with three.
- Subcontractors don't read attachments. They read what their PM tells them on a phone call, and the phone call is only as good as the PM's recall of yesterday's emailed update.
- Operations and field can't see the same view. The site super needs the next three days. The CFO needs the next three months. Excel asks both to scroll past each other's columns.
By the time we admitted we needed to change tools, we'd had four near-misses on coordination — including the doubled electrician — and one actual claim. The claim came from a curtain-wall sub who'd been told by email that his pour-cure window was a week longer than the schedule actually allowed. He'd shipped, we'd received, and the deck wasn't ready. We paid for the storage. He paid for the embarrassment.
What we picked, and why
We evaluated four products over six weeks. The selection criteria, in order of weight:
- Real-time collaborative editing for crews who can't be trained on bespoke software.
- A view that field staff can read on a phone in direct sunlight.
- API access — we knew we'd want to pipe in cost data, weather, and tender returns later.
- Pricing that scaled by user count, not by per-project licence. Construction is full-time-equivalent-light.
The tool itself matters less than people think it does. What matters is that you pick one and commit. We picked the one that scored highest on (2) — readability on a phone in sun — because we'd watched site supers squint at every other option and knew the rollout would die on the first scaffold deck.
"The best schedule is the one the crew on the deck actually opens. Everything else is a Gantt chart you printed for yourself."
The eight-month rollout
The software was live in two weeks. Onboarding the first project took six. Onboarding every active project took eight months — and we still discover Excel workbooks in shared drives that someone has quietly kept maintaining. Here is what we'd do differently.
One project at a time, ruthlessly
We started with three pilot projects in parallel — a mid-rise residential, a healthcare fit-out, and a civic centre. By month two it was clear that splitting attention across three project teams meant none of them got the change-management focus they needed. We collapsed back to one project at a time, with the project director spending two days a week embedded with that team. After that, projects landed inside two weeks each.
Train the PM, not the crew
Site crews don't need to be trained on the tool. They need to be told what view to open, where, and when. We made laminated A6 cards with a single QR code per project — scan, see your three-day look-ahead, done. The PM is the one who needs the four-hour training, because the PM is the one keeping the schedule alive.
Kill the workbook on day one
The hardest lesson, learned painfully on project two: if the old Excel workbook is still sitting in the shared drive, somebody will edit it. We started archiving the old workbooks the day the new schedule went live, and forwarding any "did you see the latest tab?" emails straight back to the team with the new URL. The first week was uncomfortable. The second week, the muscle memory rewired.
What it actually changed
Eight months in, three measured improvements and one we didn't expect.
- Coordination near-misses, down 71%. We log every catch a PM makes. The weekly count fell from a rolling-average 12 to 3.5 by month six.
- Schedule re-baseline frequency, down from monthly to quarterly. Because the timeline is live, drift is visible early. Drift early is drift you can negotiate. Drift late is drift you absorb.
- Subcontractor RFIs about timing, down ~40%. Subs can see their own dates without asking. The RFIs we still get are about substance, not about what week they're meant to be on site.
- The unexpected one: client confidence. When a client asks "where are we?" and the project director shares a live link instead of an attachment, the conversation changes. Three of our last six wins were referred by clients who specifically named the shared timeline as the reason.
What we'd watch for if we were starting now
Two things we'd flag for any GC about to make this transition:
First, don't underestimate the cost-data tie-in. The single biggest unlock from a live schedule is connecting it to live tender returns and committed-cost data. We're six months into that integration and it's where most of the future value sits. If we'd started building the data pipeline earlier, we'd be a year ahead.
Second, the client-facing view is not the internal view. Clients want milestones, risks, and a single confidence indicator. Internal teams want every dependency visible. Build two views from day one. We tried to share the internal view with clients for the first three months and watched a lot of polite confusion.
Closing thought
Tooling is rarely the bottleneck on a construction programme. People are. But people work better with tools that show them the same picture, and the cost of a missed coordination is high enough that a $40-a-seat-a-month upgrade is the cheapest insurance you'll ever buy. We should have done this in 2018. The next time you find yourself opening a thirty-tab workbook to figure out whether the electrician is double-booked, treat it as the warning it is.
