The Twelve Months After You Leave Site
The only phase of a contract with cost and no production. The team has demobilised, the subcontractors have been paid, nobody remembers why the slab was poured that way — and every repair from here comes straight out of margin.
Practical completion feels like the end. The client moves in, the site agent moves to the next job, the containers come off the plot, the final valuation goes in. What has actually started is the strangest phase of the whole contract: six, twelve or twenty-four months in which you carry an obligation to return, with no production, no valuation to recover against, and no team within a hundred kilometres of the building.
Every shilling spent in the defects liability period is pure margin. There is no measured work to price it against and no interim certificate to put it on. A contract that closed at a respectable margin can be brought back to nothing by fourteen months of return visits, and — this is the part that stings — most of those visits are for things the contractor was never liable for in the first place.
The distinction that pays for itself
On site everything reported after handover is called a snag, and the working assumption is that the contractor fixes it. Commercially, four quite different things are being lumped together, and only one of them is yours:
The work does not match the specification or the workmanship failed
A defect — you pay
Fix it, record it, and record it in a way that lets you find out whether the same trade is producing the same failure across three sites. This is the only category that is genuinely your cost.
The client changed something, or damage came from use, occupation or a third party
A variation or a chargeable repair — they pay
Price it and issue it as new work. It is not a defect and treating it as one is a gift. The instruction discipline is exactly the same as for any variation nobody priced.
Normal deterioration in the period — sealant, minor settlement cracking, finishes in heavy traffic
Fair wear and tear — nobody pays under the contract
Explain it once, in writing, with reference to what the specification actually promised. Doing this well in month two prevents a pattern of free visits for the next ten months.
A failure traceable to the design, or one that could not have been seen at handover
Latent or design — it depends who designed it
Establish the origin before mobilising anyone. On a design-and-build contract this may still be yours; on a traditional contract with a client-appointed consultant team, it may not be. Either way, cost incurred before that question is answered is cost you cannot recover.
The reason this matters more here than anywhere else in the contract is the absence of leverage. During construction, an argument about who pays for something is had between two parties who both need each other next week. After handover, you need the client to release retention and the client needs nothing from you. Every unchallenged non-defect is therefore a free repair, and the pattern establishes itself in the first month.
Speak the same language as the contract
What it is called on site What it means commercially
Snag An item on a completion list
Identified at handover and part of achieving completion. Snags are expected and were priced. They are not the same thing as a defect reported nine months later, and mixing the two lists destroys your ability to argue about either.
Defect A failure to comply with the contract
The work does not match specification, or the workmanship has failed. Yours to make good, at your cost, within the period.
Making good The obligation, not the goodwill
Returning to correct a defect. Worth using the contract's word deliberately, because "making good" is bounded by the contract while "sorting it out" is bounded by whoever asks loudest.
Defects liability period A clock with a start and an end
Usually six or twelve months from practical completion. Its end date is what triggers the final inspection and, ordinarily, the second half of retention. It has an expiry and somebody must be watching it — most often nobody is.
Latent defect Not visible at handover
Sits outside the defects period and inside the general limitation position, which is a legal question rather than a site one. Do not let it be handled as an ordinary snag call.
Final certificate The commercial end of the contract
Issued after the period closes and defects are made good. Until it exists, the contract is open however finished the building looks.
What the period actually costs
Contractors rarely know this number, because the costs arrive one at a time, months apart, after the job has been mentally closed and often after the project has been marked complete in the system:
A KES 96,000,000 contract, twelve-month defects period
Nearly a quarter of the reported margin, and by the arithmetic of the last row roughly two-fifths of the spend was avoidable — four visits for items that were wear and tear or client-caused, plus a subcontractor paid twice because their retention went out before ours came back. The transport line is the one that surprises people: the cost of going barely depends on the size of the repair, which is why nine separate visits cost multiples of one planned visit fixing nine items.
That last observation is the single most useful operational lever in the period. Unless something is unsafe or is preventing the client using the building, batch defect calls into scheduled return visits — monthly, or at a defined threshold. It converts nine mobilisations into three, and it gives you a written position on each item before somebody drives to the site to fix it.
Where the tracking and the cost stop meeting
Now the honest part about running this in software, and it is the same structural problem that shows up whenever an operational record and a financial record have to describe the same event.
There is no work order in this product. A defect call has two plausible homes: a task under the site's project, where it would sit alongside the rest of the project structure, or a helpdesk ticket. The ticket is the better fit for the lifecycle by a wide margin — it has a reference, a requester who can be the client, a priority, an assignee, a resolution clock that pauses while you are waiting on the requester, a reopened counter, and email and portal intake so the client's report lands as a record rather than a phone call. For a defects period that is genuinely most of what you need.
What a ticket cannot do is cost. A ticket carries a customer and a department; it does not carry a project. An expense, a stock issue and a timesheet all carry a project and none of them carries a department. So the defect call and the defect cost live in two structures with no dimension in common — you can report how many defects each site generated, or what the defects cost in total, and joining the two is a manual exercise.
The two weeks before you demobilise — while it is still cheap
- Record the defects period start and end dates somewhere that will be read: the project description and a recurring task dated three weeks before expiry, so the final inspection is arranged rather than remembered.
- Close the snag list formally and separately. Snags at handover and defects afterwards are different arguments, and one list cannot hold both.
- Photograph the completed work — comprehensively, dated, room by room. The cheapest defence against a claim for damage caused during occupation is evidence of the condition on the day you left.
- Confirm that no subcontractor retention has been released ahead of your own, for any trade likely to be recalled. Paying a tiler twice is entirely avoidable and entirely common.
- Get the operating and maintenance information to the client in writing, and keep proof of the handover. A large share of "defects" are maintenance items on equipment nobody was told how to maintain.
- Name one person as the defects owner, with the authority to say "that is not a defect" — and if calls will come through the helpdesk, make sure they are routed to that person rather than to whoever is on duty. If nobody holds that authority, the answer defaults to yes.
- Leave the project open in the system for the whole period, so defect labour, materials and expenses land on the job that produced them and the final margin is the true one.
- Agree with the client in writing how defects are to be reported — one channel, in writing. A defects period run through WhatsApp messages to a site agent who has moved on is how obligations are missed and how invented ones are accepted.
What AWRA OpsHub does today
- Helpdesk tickets with a real lifecycle — open, in progress, waiting requester, resolved, closed, cancelled, with priorities, a category, an assigned handler, a first-response and resolution clock, and a reopened counter that is genuinely useful here: a defect fixed twice is a workmanship pattern, not an incident.
- The resolution clock pauses while the ticket is waiting on the requester, so time spent waiting for the client to grant access does not count against you.
- Email and portal intake, so a client's report becomes a dated record with a reference instead of a phone call to somebody who has left the site.
- A standing project per site that keeps collecting cost — labour, purchase orders, expenses and stock issued from your own store — so defect cost lands on the job that caused it if you leave the project open.
- Recurring tasks that reliably spawn on a schedule: the working mechanism for periodic inspections and for the notice before the period expires.
- Custom fields on tickets and on tasks, which can be made mandatory — the only place a "defect / wear and tear / client-caused / latent" classification can live.
What it does not do
- No work order entity. No scheduled visit, no assigned crew, no parts list against a job, no per-visit cost roll-up. A defect is a ticket or a task, and neither is a work order.
- A ticket carries no project. It holds a customer and a department; expenses, stock issues and timesheets hold a project and no department. So defect counts and defect costs cannot be joined by any shared dimension — the report you want, cost per site per defect, is a manual exercise.
- No defects liability period, retention or final-certificate field anywhere. The period is a date in a description and a recurring task; retention receivable and payable are not records, as the subcontractor payment article also has to state.
- Nothing prevents spend landing on a closed project. Coding an expense checks only that the project exists — a completed or cancelled job accepts costs silently. Convenient during a defects period, and a genuine weakness for final-account integrity afterwards.
- No batching or scheduling of visits. Grouping calls into a monthly return trip — the biggest saving in the period — is a habit, not a feature.
- No warranty or guarantee register. Manufacturer warranties on installed equipment, and the subcontractor guarantees that should flow through to the client, have no home and no expiry watching. See the certificate that expired while the work carried on for why dates without a watcher are the recurring theme.
The honest position: the lifecycle of a defect is well served by the helpdesk, better than most contractors manage on paper, and the cost of a defect is well captured by the project. The two do not meet, and no amount of configuration makes them meet, because the ticket and the expense share no dimension. If your requirement is a costed work-order system for a maintenance business, we are not it and a dedicated field-service package will serve you better. If your requirement is to stop losing margin to unchallenged calls and forgotten expiry dates, the ticket queue plus an open project plus one recurring task covers most of the money — provided you accept that tying a specific repair to a specific cost is done by a person.
The verdict
The defects period is not an administrative tail; it is an uncosted, unrecoverable phase that quietly reprices the whole contract. Run it with three deliberate acts: classify every call before anyone mobilises, batch the visits that are not urgent, and leave the project open so the cost lands where it was caused. Then the number you report at practical completion and the number you actually earned are within sight of each other — and you find out which trade keeps failing while there is still a subcontractor with retention outstanding.
Frequently asked questions
How long should a defects liability period be?
Six or twelve months from practical completion is standard, twelve being the common default on Kenyan building contracts because it takes the building through a full cycle of both rainy seasons — which is when roofing, drainage and waterproofing defects reveal themselves. Longer periods appear on mechanical and electrical installations. What matters more than the length is that somebody knows the end date, since it is what triggers the final inspection and the release of the remaining retention.
What is the difference between a snag and a defect?
Timing and status. Snags are identified at or before handover, form part of achieving practical completion, and were priced into the job. Defects appear during the liability period afterwards. Keeping them on one list is a common mistake with a real cost, because it lets items that should have been closed at handover reappear as fresh obligations, and it muddies any attempt to argue about liability months later.
Can I charge the client for repairs during the defects period?
For anything that is not a defect, yes — and you should. Damage from occupation or misuse, a change the client has requested, and normal wear and tear are all outside the making-good obligation. The practical difficulty is not entitlement but habit: once two or three non-defect calls have been attended free, the expectation is set for the rest of the period. Establish the position on the first one, in writing.
Should the project stay open in the system after handover?
Yes, for the whole period. It is the only way defect labour, materials and expenses land on the job that produced them, which is what makes the final margin true rather than optimistic. Worth knowing that nothing enforces this either way — coding an expense to a project checks only that the project exists, so a completed or cancelled job will silently accept costs. That is convenient here and a control weakness generally.
Is a helpdesk ticket really the right home for a defect?
For the lifecycle, yes: reference, requester, priority, assignee, a resolution clock that pauses while you wait for site access, a reopened counter, and email or portal intake so the client's report is a dated record. The limitation to plan around is that a ticket carries a customer and a department but no project, while costs carry a project and no department — so counts and costs cannot be joined automatically. Put the site name in a required custom field on the ticket and you can at least align the two exports by hand.
Why does the same defect keep coming back?
Usually because the first visit treated the symptom. The reopened counter on a ticket is the cheapest diagnostic you have: a defect fixed twice is a workmanship or design pattern rather than an incident, and the same pattern across three sites points at a trade rather than a building. Track it deliberately, because the second and third visits cost as much to mobilise as the first and earn nothing.
How do I make sure subcontractors come back?
Do not release their retention before your own is released, and make the release trigger — practical completion, period expiry, a certificate — explicit in the subcontract rather than implied. A subcontractor already paid in full has no commercial reason to return, and you end up paying twice for the same work. This mirroring of retention on both sides is dealt with in more depth in [paying for work that was actually done](/blog/construction-subcontractor-payments-kenya); the point to carry into the defects period is that the release *date* is the control, not the percentage.
