Packages bought late in the job: what the last awards do to the savings pool
By the time a GMP job is a year old, the structure, the envelope and the major systems have been bought, and the buyout report shows a saving. The projected pool goes into the monthly report. Everyone relaxes. Then the last packages are bought: flooring, ceilings, casework, landscaping, signage, the things that are always last. They are bought into a market twelve to eighteen months after the estimate, often in a hurry, and the saving that was projected shrinks or turns over.
Late buyout is where a projected pool most often disappears, and it disappears in small pieces that never make the report as a single event.
Why the last packages behave differently
The early packages were estimated close to the time they were bought, and were bought with time to bid. The late packages were estimated at the same time and bought a year later, after material prices have moved, after the drawings have been revised a dozen times, and often after the schedule has slipped so that the trade is being bought for a start date six weeks away. Fewer bidders, less time, more revisions between the estimate and the award.
On a fictional $29 million student housing building, the first twenty packages might come in a net $410,000 under their lines. The last eight, worth $3.6 million together, might come in $280,000 over. The buyout report at month nine showed a saving of $410,000. The one at month sixteen shows $130,000, and the report never had a month in which a single event explained the change.
Reading the late awards
Each late award should be read with three questions. Did the scope grow between the estimate and the award, through drawing revisions, and if so are those revisions in the change register as design gaps or owner changes? A flooring package that is $60,000 over its line because the drawings added a floor finish to the lobby is not a buyout miss; it is a change that has not been raised. Did the market move, and by how much, on material the contractor could have bought earlier? Was the package bid competitively, or awarded to the one subcontractor available on the date?
The answers decide which fund carries the overrun. A scope growth is a change order with a cause. A market movement on priced scope is a contractor risk against contingency. A package awarded high because the schedule slipped is a contractor risk if the slip was the contractor's, and an owner change if the slip was the owner's. The buyout tracker should carry the answer on each late award, not just the variance.
The pool the report has been projecting
The projected savings pool at month nine was built on twenty packages and an assumption about the last eight. The assumption is what should have been in the report: eight packages unbought, $3.6 million of lines, projected at their lines or at some stated adjustment. A report that projected the pool from the bought packages alone, and was silent about the unbought ones, was showing a number that the last eight awards were always going to change.
Timing the last buys
The owner cannot buy the packages, but the owner can ask when they will be bought, and can see in the tracker which lines are still open at each month. A package still unbought at month fourteen for a trade that starts at month sixteen is a package that will be bought in a hurry, and the owner should expect the award to reflect that. Asking the question at month eleven, when there is still time to bid it properly, is the only intervention available.
What the tracker should show
For each package: the GMP line, the award if made, the award date, the variance, the scope carried out, and for late awards the reason for the variance in a word: scope, market, schedule, competition. The projected pool is then the sum of awarded variances plus the unbought lines at a stated assumption, with the assumption printed. When the last package is bought, the projection and the actual meet, and the report can say by how much the assumption was wrong, which is a useful thing to know about the next project.
What to do this month
- List every unbought package with its GMP line and its planned award date, and put the list in the report.
- Project the pool from awarded variances plus unbought lines at a stated assumption, with the assumption printed.
- On each late award, record the reason for the variance: scope, market, schedule or competition.
- Ask in month eleven which packages will still be open at month fourteen, and why.
Questions on this
Why do late packages come in over?
They were estimated with the rest and bought a year later, into a moved market, from revised drawings, often on a short bid period. Any one of those adds cost; the last packages usually have all three.
Is a late overrun a buyout miss or a change?
Read the award. Scope that grew through drawing revisions is a change that should be in the register with a cause. Market movement on priced scope is a contractor risk. Schedule pressure depends on whose slip it was.
How should the pool be projected before buyout is complete?
From awarded variances plus the unbought lines at a stated assumption, with the assumption in the report. A pool projected from the bought packages alone is a number the last awards will change.
In the product
Buyout tracker, Shared savings, Change order register. Free tool: Change Order Exposure.
Keep reading
Earlier: Open change orders in the forecast: weighting what is not yet signed. Later: Unit prices and alternates inside a GMP: what the owner should record before they are used. All articles on buyout and shared savings.
See the baseline on a contract like yours.
Thirty minutes on a call. We read a fictional GMP amendment into the baseline and show what the basis flags reveal.