← Blog
Opinion·27 May 2026·7 min read

Five ways a summary schedule lies — and how to keep it honest

A one-page summary schedule is the most-read artefact in the room and the least-scrutinised. Everyone trusts the rolled-up bars because they're clean. That's exactly the problem: cleanliness hides things. Here are five specific ways a summary quietly misleads a sponsor — and, for each, the tell that gives it away and the fix that keeps it honest.

I like a good summary schedule. A director shouldn't have to wade through four thousand activities to know whether the project lands on time. But every act of summarising is an act of hiding — you throw away detail on purpose, and detail is where the truth lives. Done well, a roll-up shows the shape of the plan. Done carelessly, it launders a bad schedule into a reassuring picture, and no one notices until the picture is wrong in front of the board.

These aren't hypotheticals. Each of the five below is something I've watched slide through a monthly report, get presented with confidence, and then quietly explode. The good news: every one has a tell you can spot in seconds, and a fix that costs almost nothing once you know to look.

Five ways a summary schedule lies 1 · It buries the critical path which phase actually drives the end date? 2 · Milestones that trace to nothing no real activity ID behind the diamond 3 · Roll-ups over dirty inputs wrong calendars, progress not brought to date 4 · The moving baseline variance shown against a re-approved plan 5 · Bars with no logic behind them a picture that can't feel a slip
Fig 1. The five failure modes, in order of how often I meet them. Each is a way a clean summary can be wrong; the rest of this piece is the tell and the fix for each.

The five lies

1 · It buries the critical path

This is the big one. A summary rolls up ten phases into ten tidy bars, all the same colour, and the eye reads them as equally important. They aren't. One of those phases drives the completion date and the other nine have float. A roll-up that treats them alike hides the single most decision-relevant fact on the page: which phase, if it slips a week, moves the end. Directors then obsess over the longest bar or the busiest-looking phase, when the driver might be a short, quiet one buried in the middle.

The tell: every summary bar is the same colour and weight, and nobody in the room can point to the critical phase without opening the detailed schedule. If the answer to "what drives the finish?" is a shrug, the summary is lying by omission.

The fix: mark the driving phase. Colour it, weight it, annotate it — whatever it takes so the critical path survives the roll-up instead of dissolving into it. In Sketchedule you keep the critical phase highlighted on the summary: drive a bar's colour from a status or "critical" column so the driver stays red while the rest sit grey, and it holds through collapse/expand and into every export. Be honest about the division of labour, though — Sketchedule presents the critical path; your P6 or MS Project network still computes it. The engine finds the driver; the summary's job is to not hide it.

Buried critical path vs critical phase highlighted — built in Sketchedule ✗ Buried — all bars alike Q1Q2Q3 Design Procure Civils M&E Fit-out Commission Handover Which one drives handover? You can't tell. ✓ Highlighted — driver stands out Q1Q2Q3 Design Procure critical Civils M&E Fit-out Commission Handover Procure → Civils → Fit-out → Commission drives it.
Fig 2. Same six phases, same handover milestone. On the left the roll-up is uniform grey and the critical path is invisible. On the right the driving chain is held red through the summary, so the one thing a sponsor needs — what moves the end — reads at a glance. The network still lives in P6/MSP; the summary just refuses to hide its answer.

2 · Milestones that don't trace to a real activity ID

Someone drags a diamond onto the summary labelled "Structure complete — 12 Aug" because that's the date the client wants to hear. It looks authoritative. But there's no activity ID behind it; it's a hand-placed date that agrees with nothing in the detailed schedule. When the network moves, the diamond doesn't — it just sits there, decorative and wrong, radiating false confidence right up until the day it's missed.

The tell: ask "what activity ID is this milestone?" and the honest answer is "it's… just a milestone we added." A date with no ID behind it is a wish, not a forecast.

The fix: every milestone on the summary must map to a real finish/start of a real activity in the source schedule, so when that activity moves, the diamond moves with it. In Sketchedule you bring milestones in from the P6/MSP import and keep them tied to their activity ID — and with conditional milestone symbology the diamond can flip colour or shape when its own date breaches a threshold, computed locally. No hand-placed dates that quietly detach from the network.

3 · Roll-ups over dirty inputs

A summary bar is only as honest as the child activities it spans. Feed the roll-up a phase where half the tasks are on the wrong calendar, or where progress was never brought to the data date, and the summary bar's span is simply wrong — it starts or ends on a date the real work will never hit. The roll-up doesn't validate its inputs; it faithfully summarises rubbish into a smooth, believable bar. Garbage in, tidy garbage out.

The tell: the summary bar's finish doesn't match what the site actually expects, and when you drill in you find a task on a 7-day calendar that should be 5-day, or a % complete that implies work done after the data date. The bar looks fine; the arithmetic underneath is broken.

The fix: fix the inputs, then re-summarise — never the other way round. Confirm calendars, and bring every activity's progress honestly to the current data date before you roll up. In Sketchedule the summary bars are automatic — they span their children — so the moment you refresh to the current data date the roll-up re-spans over clean inputs rather than stale ones. That refresh is only meaningful, of course, if the durations and calendars in the source P6/MSP file are right in the first place; Sketchedule spans what it's given, it doesn't re-calendarise the network for you.

A dirty child drags the summary's span Wk 2Wk 5Wk 8Wk 11 Civils (summary) wrong finish Piling Pile caps Ground beams on 7-day cal (should be 5) Fix the calendar → the automatic summary re-spans back to Wk 8, not Wk 11.
Fig 3. The summary bar (grey) faithfully spans its children — including one on the wrong calendar that overruns by three weeks, dragging the roll-up's finish to a date the work will never hit. Clean the input and the automatic span snaps back to the truth. The roll-up never lied on purpose; it just believed a dirty child.

4 · The moving baseline

Here's the quietest lie of all. A phase slips. Rather than report the slip, the baseline gets "re-approved" to the new dates — and next month's summary shows the variance against that, so everything reads on plan again. Do this a few times and you've walked the end date out by months while every monthly report showed green. The baseline was supposed to be the fixed thing you measure against; if it moves whenever reality does, it measures nothing.

The tell: variance is suspiciously small month after month despite obvious drift, and the baseline dates in this report don't match the ones from two reports ago. A baseline that keeps agreeing with the current schedule isn't a baseline — it's a mirror.

The fix: hold one approved baseline and show variance against it, with any re-baseline a formal, visible, dated event — not a silent edit. In Sketchedule you compare or redline the current summary against the real, frozen baseline, so accumulated drift stays visible instead of being absorbed. Keep the approved baseline in the source of truth, snapshot it once, and let the redline show exactly how far the plan has walked.

The re-baseline test. Before you accept a re-approved baseline, ask: "would this survive being shown next to the original approved plan?" If the honest answer is "no, because the drift would be embarrassing," you're not re-baselining — you're hiding a slip. A legitimate re-baseline is announced, dated, and sits alongside the original, never quietly in its place.

5 · Bars with no logic behind them

The prettiest summaries are often the emptiest: a row of hand-drawn bars with no links between them. It photographs beautifully and it's completely inert. Because nothing is tied to anything, the picture cannot feel a slip — push one bar out and its successors sit blissfully still, open-ended, promising a finish that no logic supports. A schedule with open ends isn't forecasting; it's drawing.

The tell: move one bar in your head and ask what should happen downstream — if the answer is "nothing moves," there's no logic there. Dangling activities with no predecessor or successor, and a total float that's implausibly generous everywhere, are the fingerprints.

The fix: the logic has to live in the network — and this is the honest limit of any presentation layer. Sketchedule is a presentation front-end with a light scheduler; the driving logic, the critical path and the float belong in your P6 or MS Project model, and that's where open ends must be closed. What Sketchedule does is refuse to launder the problem: refresh from the source and the summary reflects the real linked network, so a summary built over a properly logic-linked schedule stays honest, and one built over dangles has nowhere to hide the fiction.

The one-question audit. For any summary schedule, ask the five questions in order: (1) which phase drives the end? (2) what activity ID is each milestone? (3) are the inputs on the right calendars and brought to the data date? (4) is this the original approved baseline? (5) if I move one bar, does anything downstream move? A summary that answers all five cleanly is honest. Any shrug is a lie you haven't found yet.

Keeping it honest, in practice

Notice the through-line: every one of these lies is a place where a clean picture drifted away from the messy network underneath it. The fix is never "make a prettier summary" — it's "keep the summary tethered to the truth." A presentation layer earns its keep precisely by not breaking that tether: highlight the driver instead of flattening it, trace milestones to IDs, span automatically over clean inputs, redline against the frozen baseline, and refresh from the real linked network rather than redrawing it by hand.

  1. Highlight the driver. Colour the critical phase on the summary so the roll-up can't hide which phase moves the end — and keep it through collapse and export.
  2. Trace every milestone. Import milestones tied to their activity ID so a diamond moves when its activity moves, never a hand-placed date.
  3. Clean before you roll up. Fix calendars and bring progress to the data date, then let automatic summary bars re-span over honest inputs.
  4. Redline against the real baseline. Compare the current summary to the original approved baseline so accumulated drift stays visible.
  5. Refresh from the linked network. Relink from the source P6/MSP model — where the logic and critical path live — so the picture reflects a schedule that can feel a slip.
The lieThe tellThe fix
Buries the critical pathAll bars alike; no one can point to the driverHighlight the driving phase; compute it in P6/MSP
Milestones tracing to nothing"What activity ID is this?" → a shrugMap every milestone to a real activity ID
Roll-ups over dirty inputsSummary finish doesn't match site realityFix calendars + data date, then re-span
The moving baselineTiny variance despite obvious driftFreeze one baseline; redline against it
Bars with no logicMove one bar → nothing downstream movesClose open ends in the network; refresh from it

Key takeaways

Keep your next summary honest

Open Sketchedule in a browser — free, no install, nothing uploaded. Highlight the critical phase, tie milestones to IDs, and redline against the real baseline.

← BlogAll articles

Primavera and P6 are trademarks of Oracle Corporation; Microsoft Project is a trademark of Microsoft Corporation. Sketchedule is an independent product and is not affiliated with, endorsed by or sponsored by Oracle or Microsoft. Figures are illustrative, drawn in Sketchedule; Fig 2 is a faithful redraw of a buried-vs-highlighted critical-path summary built in the app.