Splitting In-Progress Activities
Last reviewed 9 September 20262,304 words10 min read
Preparing a revised baseline — what to do with the activity that is half done on the new data date
🟩 In one line: When a revised baseline is built, every activity that is in progress at its data date is either kept whole (actual start retained, remaining duration re-planned) or split into a completed part and a remaining part — and the choice is made by a rule, not by mood, with the mapping recorded so that the old baseline still compares to the new one.
👤 Who this is for: M/S, with J reading for the P6 steps. Prerequisites: baseline-approval-and-control, revised-baseline-vs-recovery-schedule, data-date, forecasting-honestly.
First, let's be honest about why this page exists
A revised baseline is built from the last accepted update, at that update's data date. Everything before the data date is as-built; everything after is plan. The problem is the 60 to 150 activities that straddle the line — started, not finished. Leave them alone and the new baseline carries a "planned" activity that started three months ago with a percent complete nobody can baseline against. Split them badly and the quantities, hours and logic end up in two places with the totals wrong.
There is no clause that says "split". There is a clean way to do it and a messy way, and the messy way shows up six months later as BEI values that make no sense and a Schedule Comparison that cannot match activities between BL0 and BL1.
🟨 The standard — what "good" looks like
| Source | What it asks for (paraphrased) | Consequence for the split |
|---|---|---|
| FIDIC 1999 Cl 8.3 / 2017 Cl 8.3 | Revised programme consistent with actual progress | As-built to the data date is not optional |
| NEC3/NEC4 Cl 31.2, 32.1 | Programme shows actual progress and its effect on remaining work | Remaining work is re-planned, not carried forward untouched |
| SCL Protocol 2nd ed., Guidance Part B (updated programme) | Updated programme = as-built progress to the date, planned thereafter | The revised baseline is an updated programme frozen as plan of record |
| AACE RP 38R-06 | Basis documented; revised with the baseline | The SBM Rev 1 records the split rule and the mapping table |
| DCMA 14-Point, Metric 9 (Invalid Dates) | No actual dates after the data date; no forecast dates before it | A whole activity kept with actual start is fine; one with a planned start before the DD is not |
| GAO Schedule Assessment Guide, BP 10 | Baseline revisions documented with rationale; traceability maintained | Old ID → new ID mapping kept for the life of the job |
| PMI Practice Standard for Scheduling, 2nd ed. | Baseline changes controlled and traceable | Same |
🟢 Rule: as-built to the data date, re-planned after it; split only where the remaining work is long enough or different enough to justify it; and never lose the link between the old activity ID and the new ones.
How it actually works
1. The decision rule
| Situation at the revised-baseline data date | Action | Why |
|---|---|---|
| Remaining duration ≤ 5 working days, or physical % ≥ 80 | Keep whole — retain actual start, set remaining duration from achieved rate | Splitting a 3-day tail creates two activities for no gain |
| Remaining ≥ 10 working days and logic, method or resourcing of the remaining work changes | Split | The remaining part needs its own logic and its own basis |
| Remaining ≥ 10 working days, logic and method unchanged | Keep whole — but re-forecast the RD and add a "Forecast Basis" note | Same activity, honest remaining duration |
| Between 5 and 10 days remaining | Planner's call; default keep whole | Document either way |
| Activity has been re-sequenced by zone (e.g. L5 blockwork now done in two halves) | Split by zone into as many parts as the new method needs | Logic follows the zones |
| LOE or WBS Summary | Never split — they take their dates from their children | — |
On a typical mid-rise with 120 in-progress activities at the revised data date, 20–30 get split. If you are splitting 100 of them, the rule is wrong.
2. The split, step by step (P6)
| Step | Action | Detail |
|---|---|---|
| 1 | Work in the revised-baseline project copy, never the live update | Copy of the accepted update, renamed "BL1-Draft-DD 01Dec25" |
| 2 | Filter: Activity Status = In Progress, Remaining Duration ≥ 10 | Add columns: Actual Start, Remaining Duration, Physical %, Budgeted Units, Actual Units |
| 3 | For each activity to split, copy it (Ctrl+C / Ctrl+V in the same WBS) | Copy carries logic, codes, resources — you will edit these |
| 4 | Original becomes the completed part: ID suffix -A; Actual Finish = last IR/pour card date, or the data date if work continued to it; Remaining Duration 0; Physical % 100 | Never set the actual finish after the data date |
| 5 | Copy becomes the remaining part: ID suffix -B; clear Actual Start; Original Duration = remaining from achieved rate; Status = Not Started; Physical % 0 | Notebook "Forecast Basis": quantity remaining, achieved rate, calc |
| 6 | Logic: -A keeps all predecessors; -B keeps all successors; delete successors from -A and predecessors from -B; add -A → -B FS 0 | Any SS/FF pairs re-pointed to the correct half |
| 7 | Resources: -A Budgeted Units = Actual Units to date; -B Budgeted Units = original budget − actual | Same for cost. The two sum to the original budget unless a scope change is separately recorded |
| 8 | Steps (if used): completed steps on -A, remaining steps on -B; weights re-normalised so each activity sums to 100 | Rules-of-credit method unchanged |
| 9 | Activity code "Split From" = original ID on both halves | Filter later gives you every split in one view |
| 10 | Quantity sheet: one row per half; installed-to-date locked against -A; remaining against -B | Sheet stays the source |
Alternative for zone splits: create -B, -C, -D as needed, each with its own quantity, and let -A hold everything completed regardless of zone.
3. Keeping the comparison alive
Schedule Comparison and BEI both work on Activity ID. Once L05-BLK-020 becomes L05-BLK-020-A and -B, the old baseline no longer matches. Three things prevent this becoming a problem:
- Mapping table in SBM Rev 1: original ID, new IDs, split date, method (time / zone), budget split.
- Activity code "Split From" on every child, so a Global Change or a filter can re-associate.
- BEI for the transition month is calculated on the mapping: the -A halves count as finished against the original baseline-finish where the original activity's BL finish ≤ data date.
After the first month under BL1, everything measures against BL1 and the problem goes away. It is only the transition that needs the table.
4. Whole-kept activities — what still changes
Even where an activity is kept whole, three things are done in the revised baseline: remaining duration re-forecast from achieved rate; any out-of-sequence relationship repaired so that Retained Logic and Progress Override give the same dates; and the Notebook "Forecast Basis" written. The revised baseline is the one moment where every remaining duration is defended from scratch, because the SBM is being rewritten anyway.
5. What the revised baseline data date must look like afterwards
| Check | Pass |
|---|---|
| Actual dates after the data date | 0 |
| Planned starts before the data date on not-started activities | 0 |
| In-progress activities with RD = original − elapsed | 0 (every RD has a basis) |
| Activities with both actual finish and remaining duration > 0 | 0 |
| Budget totals per WBS = pre-split totals ± recorded scope change | Yes |
| Every -B has exactly one -A predecessor | Yes |
| Split mapping table complete and in SBM Rev 1 | Yes |
🟥 Where people go wrong
- Splitting everything. Two hundred activities become four hundred; the network doubles in size, the comparison breaks everywhere and nothing is gained. The rule above gives 20–30 splits on a typical job.
- Setting the -A actual finish on the data date when the last work was three weeks earlier. That invents three weeks of as-built progress. The actual finish is the last evidenced work date; the gap before the data date is just a gap.
- Leaving both halves with the full budget. Hours and cost double in the roll-up and the cash flow curve jumps. -A gets actuals to date, -B gets the remainder, and the WBS total is checked before issue.
- Successors left on -A. The following trade now starts when the completed half "finishes" — which was weeks ago — and the schedule shows finishes starting before the plaster they depend on. Every successor moves to -B.
- Splitting in the live update instead of the baseline draft. The next monthly update then compares to itself with different IDs. Split in the draft, accept the draft, then the live update inherits the split IDs from the accepted BL1.
- No mapping table. Six months later a reviewer asks how BL0 activity L05-BLK-020 became two activities with different totals, and nobody can answer. One table, kept in the SBM, ten minutes to build.
⚖️ When you're challenged
"Why not just keep the actual start and change the remaining duration? Much simpler." For most of them I do — around eighty percent are kept whole. I split the ones where the remaining work has different logic or a different method, because a single activity can't have half its successors driven by work that's finished and half by work that hasn't started. The split rule and the count are in the SBM.
"Your new baseline has more activities than the old one. Have you added scope?" Thirty-two of the extra activities are splits of in-progress work — the mapping table is Appendix D of SBM Rev 1, and the budget on each pair sums to the original. The remaining six are VO-014, already instructed and reported. Scope hasn't changed outside those six.
"The BEI dropped to 0.78 this month. What happened?" Transition month. The comparison is running against BL0 IDs that no longer exist in the split activities. Against the mapping table it's 0.91, consistent with last month. From next month it measures against BL1 and the question goes away — that's noted in the narrative.
📄 Related pages
- Baseline Approval and Control — where BL1 gets its name and its acceptance letter
- Revised Baseline vs Recovery Schedule — why this only happens for a baseline, never a recovery
- Forecasting honestly — the achieved-rate RD on every -B
- Out-of-sequence progress — repaired at the same time
- Quantity tracking and physical progress — the sheet that carries the quantity split
- CPLI and BEI — the transition-month calculation
- Schedule Basis Memorandum — where the mapping table lives
- P6 Constraint Types — the one constraint the example allows
✏️ Worked example
A mid-rise residential block in Ajman, revised baseline BL1 at data date 1 December following a 25-day EOT. Activity L05-BLK-020 "Level 5 blockwork" — original 24 days, 640 m², 1,920 manhours, started 6 November, 260 m² laid and IR'd by 28 November, remaining work re-sequenced into two zones because Zone B is now held by a revised MEP riser drawing.
| Field | Original (in Upd 10) | L05-BLK-020-A (completed) | L05-BLK-020-B (Zone A remaining) | L05-BLK-020-C (Zone B remaining) |
|---|---|---|---|---|
| Status | In progress | Completed | Not started | Not started |
| Actual Start | 6 Nov | 6 Nov | — | — |
| Actual Finish | — | 28 Nov | — | — |
| Quantity (m²) | 640 | 260 | 200 | 180 |
| Budgeted manhours | 1,920 | 780 (actual) | 600 | 540 |
| Original / Remaining duration | 24 / — | 20 (as-built) / 0 | 8 (200 ÷ 27 m²/day achieved, rounded up) | 7 (180 ÷ 27, rounded up) |
| Predecessors | L05-STR-050 FS | L05-STR-050 FS | L05-BLK-020-A FS | L05-BLK-020-A FS; MEP-RISER-DWG-R2 FS (Start On or After 12 Dec, document ref DRN-0442) |
| Successors | L05-PLA-010 FS | L05-BLK-020-B; -C | L05-PLA-010-ZA FS | L05-PLA-010-ZB FS |
| Code: Split From | — | L05-BLK-020 | L05-BLK-020 | L05-BLK-020 |
| Notebook | — | "Split 01Dec25: completed to IR-1187" | "Forecast Basis: 200 m² ÷ 27 m²/d achieved Nov" | "Forecast Basis: 180 m² ÷ 27 m²/d; held by DRN-0442" |
Check: 260 + 200 + 180 = 640 m². 780 + 600 + 540 = 1,920 mh. Plastering was also split by zone to match. The constraint on -C is in the constraint register with its document reference — the only reason it is allowed.
📖 References
- FIDIC Conditions of Contract for Construction, 1999, Cl 8.3; 2017, Cl 8.3 (check the edition in your contract)
- NEC3 / NEC4 Engineering and Construction Contract, Cl 31.2, 32.1 (check the edition in your contract)
- SCL Delay and Disruption Protocol, 2nd ed. (2017), Guidance Part B
- AACE International RP 38R-06, Documenting the Schedule Basis
- DCMA 14-Point Schedule Assessment, Metric 9
- GAO Schedule Assessment Guide (GAO-16-89G), Best Practice 10
- PMI Practice Standard for Scheduling, 2nd ed., baseline maintenance
- Oracle Primavera P6 Professional User Guide — Copy/Paste Activities; Activity Details (Status, Resources, Relationships, Notebook); Activity Codes
From the field
Experience from working planners. Unreviewed — read it as experience, not guidance.
Add what you know about splitting in-progress activities. What worked, what the consultant pushed back on, what you would do differently next time. A paragraph is plenty.
Contributors get their name and one link on the site — your own templates, course or consultancy. We take nothing and hold nothing.
Add your experience