Every missed commitment on a jobsite has a story behind it, and the story you hear first is almost never the real one. "Drywall didn't get hung because the material wasn't there." Okay. But the material not being there is a symptom, not a cause. If you close out that reason as "material delay," slap it on the whiteboard, and move on, you've learned nothing. You'll be writing the same reason on the same board six weeks from now.
The Last Planner System gives you a weekly measurement — Percent Plan Complete, the share of the commitments you made that you actually finished — and, more importantly, a reason code for every one you missed. That variance data is gold, but only if you dig into it. Root cause analysis is the digging. This is how you do it without turning it into a blame session or a paperwork exercise nobody reads.
Surface reasons lie to you
Start by being honest about how thin most variance reasons are. When PPC comes in at 68% on a Friday and you're standing at the board with eight subs, the reasons come out fast and shallow: weather, no material, prior trade wasn't done, short-handed. Those are real, but they're the outermost layer. Compare these two:
- Surface reason: "Framing inspection wasn't signed, so insulation couldn't start."
- Root cause: "We called for inspection Thursday afternoon knowing the AHJ books 48 hours out, so it was never going to land before Monday — and nobody flagged that when we made the commitment."
Same missed activity. But one of those you can fix. The other is just a fact you record and repeat. The whole point of root cause analysis is to keep asking questions until you hit something your team actually controls — a process, a habit, a hand-off, a decision — because that's the only kind of cause you can change.
What your variance data is actually telling you
Before you analyze anything, your reason codes have to be clean. If half your misses are logged as "other" or a vague "coordination," you've got nothing to work with. Discipline the capture first. A workable set of categories for most commercial and multi-family jobs:
- Prerequisite work — the trade ahead of you wasn't complete or wasn't inspected.
- Materials — not on site, wrong material, or damaged.
- Information — open RFI, missing submittal, no approved shop drawing, unresolved design conflict.
- Labor — undermanned, crew pulled to another job, no-show.
- Equipment — lift down, no power, hoist conflict.
- Directive change — owner or PM changed scope or priority mid-week.
- Weather / conditions — genuinely uncontrollable stuff.
- Overcommitment / planning — you committed to more than the crew could physically do, or the durations were fantasy.
Log the category, but also log the context: which trade, which area, which phase, what day. When that data lives in one place — a whiteboard photo won't cut it, which is where a look-ahead scheduling tool like LookAheadWall earns its keep, tying the reason to the specific activity, location, and week — you can start slicing it. And slicing is where the patterns fall out.
The Five Whys, done for real
Everybody's heard of the Five Whys. Most people do it badly, because they stop at the first answer that lets them off the hook. Here's a real chain from a mid-rise job, run properly:
- Why didn't the plumbers finish rough-in on Level 3 west? Because they ran out of a specific fitting.
- Why did they run out? The delivery came up short.
- Why did it come up short? The order was placed off an outdated takeoff that didn't reflect the ASI that added two more restrooms.
- Why was the order off an old takeoff? The ASI got distributed to the field but never made it back to whoever does material orders.
- Why didn't it make it back? There's no step in our process where a change directive triggers a re-check of pending material orders.
Notice the first "why" answer — "ran out of a fitting" — is exactly the surface reason you'd have written on the board. Five questions later you're looking at a missing feedback loop between change management and procurement. That you can fix, and fixing it prevents a whole class of future misses, not just this one fitting. The number five isn't sacred, by the way. Sometimes you hit bedrock in three whys, sometimes it takes seven. You're done when the next "why" would be something outside your control ("why does the county take 48 hours to inspect?") or when you've reached a process you own.
Two traps to avoid. First, don't let the chain drift into a single person — "because Dave forgot." Dave forgetting is not a root cause; a process that depends on Dave remembering is. Second, watch for branching. Real misses often have two or three parallel causes. If the honest answer to a "why" is "these two things both had to go wrong," follow both branches. That's your cue to switch to a fishbone.
Fishbone when there's more than one culprit
The fishbone (Ishikawa) diagram is for the messy ones — the miss that wasn't one clean chain but a pile-up. Draw the problem as the fish's head ("Level 4 didn't reach drywall-ready on schedule") and run bones for the usual construction buckets: prior trades, materials, manpower, equipment, information, and site conditions. Then get the people who were actually there to fill in causes under each bone.
What the fishbone buys you is honesty about complexity. On that Level 4 example you might find the ceiling grid was 80% done (prior trades), the fire caulk submittal was still in review (information), and one crew got pulled to chase a punch list on another floor (manpower) — all in the same week. No single Five Whys captures that. Once it's all on the board, you circle the one or two causes that, if fixed, would have saved the week. Usually it's not all six bones. It's two.
Pareto: fix the vital few, ignore the trivial many
You cannot run a formal analysis on every miss, and you shouldn't try. This is where Pareto keeps you sane. Take a month or a quarter of variance data and sort the reason categories by frequency — or better, by the days of schedule they cost you. You'll almost always find that a small handful of categories drive the bulk of your misses. Classic 80/20: maybe "information / open RFIs" and "prerequisite work not inspected" together account for most of your lost commitments.
That tells you exactly where to spend your root cause energy. Doing a deep-dive on the one week the crane was down for a freak hydraulic failure is a waste — it's not going to recur. Doing a deep-dive on why RFIs keep landing on your critical path is where the real PPC gains live. Chase the tall bars on the chart, leave the short ones alone.
Running an actual root cause session
The weekly PPC review is not a root cause session — there's no time, and the subs want to get to work. Root cause work happens off to the side, on the patterns worth the effort. Keep the session tight:
- Pick one pattern. Not "let's fix our PPC." Something specific the data surfaced: "prerequisite-work misses on the podium levels."
- Pull the receipts first. Walk in with the actual variance history — dates, areas, trades. Arguing from memory turns it into opinions.
- Get the right five people in the room. The foremen who lived it, the super, whoever owns the process you suspect. Not fifteen people, and nobody who'll make it political.
- Run the Five Whys or fishbone out loud, on a wall everyone sees. Land on one or two root causes, not ten.
- Write countermeasures with an owner and a date. "Add a lead-time check to the six-week look-ahead review — Maria — starting next Tuesday." A countermeasure with no name and no date is a wish.
- Keep it under 45 minutes. If it runs long, your problem was too big; break it down.
Make it blame-free and mean it. The first time someone gets hammered in a root cause session, honest reasons dry up forever and every future miss becomes "weather." You want foremen comfortable saying "I committed to more than I could do" — because overcommitment is one of the most fixable causes there is, and it only surfaces in a room where the truth is safe.
The root causes you'll keep finding
After you've run enough of these, the same underlying causes show up again and again across different jobs. Know them going in:
- Constraints spotted too late. A six-week look-ahead exists precisely so long-lead materials, permits, and inspections get flagged with time to react. When the miss traces back to "we found out too late," the fix is almost always making the look-ahead review actually screen for that constraint type — not extending the horizon for its own sake.
- Commitments made under pressure. A foreman commits to Friday because saying "Wednesday of next week" in front of the owner's rep feels like weakness. Then it misses, and it was never real. The countermeasure is cultural: reward the honest commitment, not the optimistic one.
- The same resource promised twice. One crew or one lift committed to two activities in the same window. This one hides until you look at commitments across trades in one place — a shared weekly work plan makes the double-booking visible before Monday instead of after.
- Information latency. RFIs and submittals sitting in someone's inbox. If your Pareto chart is topped by "information," the root cause is rarely the field — it's turnaround time upstream, and the fix lives with the PM and the design team.
- Hand-off gaps. The change that never reached procurement. The inspection that got called but never confirmed. These are missing steps in a process, and they're the most satisfying to fix because one small process change kills a recurring miss for good.
Close the loop or it was theater
Here's the step most teams skip, and skipping it is why root cause analysis gets a reputation as busywork. After you implement a countermeasure, you have to verify it worked. Go back to the variance data and watch that specific category. If "prerequisite-work misses on the podium" was running four a week and it's down to one after your process change, the countermeasure took. If it didn't move, your analysis was wrong — you fixed a symptom, not the cause — and you go back to the wall.
Tie it straight back to the number. PPC should climb as you retire root causes, and if it isn't climbing, either you're not finding the real causes or the countermeasures aren't sticking. A rising PPC over a couple of months is the honest scoreboard for whether any of this analysis is doing anything. Everything else is talk.
Push the learning past the project
The last mile is making sure a lesson learned on this job doesn't die when the job closes out. The hand-off gap between change directives and procurement isn't unique to your building — it'll bite the next project too. When a root cause turns out to be systemic, it belongs in your standard process, your preconstruction checklist, or your onboarding for new supers, not just in one job's meeting notes. That's how a company actually gets better instead of relearning the same painful lesson on every job.
None of this requires fancy tools. It requires clean variance data, the discipline to keep asking why past the comfortable first answer, and a culture where a foreman can tell you the truth about why he came up short. Get those three things right and your look-ahead scheduling stops being a wall of reasons you're doomed to repeat, and starts being the feedback loop that quietly makes every crew on the job more reliable. Find the real reason. Fix it once. Then go find the next one.