Resource
Repair or Replace a Commercial Roof: How to Decide
Two buildings can look identical from the parking lot and deserve opposite answers. What separates them is evidence, and gathering that evidence is neither slow nor expensive.
Updated
The membrane is the part you can see, not the part that decides
Owners tend to judge a roof by its surface, because the surface is what a walk shows. A weathered sheet looks finished. A sheet that still looks presentable feels like it has time left. Neither impression is reliable, and both of them regularly point the wrong way.
What settles the question sits underneath. Insulation board that has taken on water does not dry once it is sealed under a covering. It loses most of its thermal value, it softens so the surface above it dents and holds water, it stays chemically aggressive against a steel deck, and the wet area spreads sideways through the board a little further each season. The framework below therefore has nothing to do with repair technique. It asks how much of the assembly is still doing its job, how fast that share is shrinking, and what the money going in every year is actually buying.
Cores tell you what percentage of the roof is still working
A core cutter takes a plug through the full assembly, and the plug is read as a stack from the surface down. Out of it comes the number of coverings present, the board sitting under each of them, the thickness of that board, whether moisture is trapped inside it, and what material the structural deck below turns out to be. Ten minutes of work answers questions that a full day of walking cannot.
One plug describes one location. A useful survey takes plugs in every quadrant, plugs in every area with a reported leak, plugs in the low spots, and at least one plug somewhere the roof appears healthy so there is a baseline to compare against. On a large field, a nighttime infrared scan narrows where to cut, since areas holding moisture release heat more slowly after sunset. Infrared flags candidates. It does not confirm anything, and every flagged area still gets a plug before it appears in a scope.
What comes out of that is a rough percentage of the field that is wet, and that percentage is the single most useful number in the entire decision. Confined saturation around identifiable failures points toward repair: open the wet areas, rebuild the details that let water in, and the rest of the assembly is intact. Moisture that shows up in plugs with no obvious source, or in plugs spread across unrelated parts of the roof, means water is entering at points nobody has located yet, and cutting out the wet board will not stop it.
- Localized wet areas tied to specific failed details: repair, replace the wet board, and keep inspecting.
- Moisture in a modest share of the field, sources identified, deck sound: repair now and plan replacement.
- Moisture in plugs with no traceable source: water is getting in somewhere the survey has not yet found.
- Widespread saturation across quadrants: repairs chase a problem that keeps moving.
Annual spend against remaining service life
Every roof has a number attached to it that most owners have never totalled: what was spent on it last year, and the year before, across every invoice from every trade. Emergency calls, patches, drain work, interior ceiling repair, damaged inventory, and staff hours spent moving buckets all belong in that figure. Pulled together it is frequently larger than anyone expected, and it is often growing on a curve rather than a line.
Set that against remaining service life, which comes out of the condition assessment rather than out of the install date. An assembly with meaningful life left and a flat annual spend is being maintained correctly, and replacing it early throws away years you already paid for. An assembly with little life left and a spend that climbs every year is consuming capital in small increments while the replacement gets more expensive.
The comparison people miss is against the cost of not deciding. Repeated repair on a roof that is genuinely finished does not simply cost what the repairs cost. It costs the interior damage, the tenant credits, the lost inventory and the disruption, none of which arrive on a date anybody picked.
The replacement still happens either way. Deciding while there is time left buys the season, the competitive bid and the budget year. Deciding after the fact buys none of those.
Deck condition can end the discussion on its own
Below the insulation is the structural deck, and its condition can override every other input. A steel deck loses metal from its upper face first, following whatever path the water took, and by the time a hole appears the panel around it has usually given up a good deal of its section. That is a structural matter before it is a roofing matter.
A deteriorated deck cannot be evaluated or repaired through an existing roof, which removes any build-over option immediately and moves the conversation to full removal by default. It also introduces a quantity nobody can size in advance, which is why deck replacement belongs in a scope as a unit price against a measured quantity rather than as a lump sum guess.
Concrete and gypsum decks fail differently and just as decisively. A sound gypsum or cementitious wood fiber deck is fastened with anchors made specifically for that material, but long-term moisture degrades the deck until it will not hold even those, and a pullout test on the exposed deck is what establishes whether anything can be fastened to it at all. Where plugs come back showing a soft or delaminated deck, the honest recommendation is removal and a proper look, and any proposal that offers to build over that condition is worth reading very carefully.
Can the assembly physically take another covering?
Building over an existing roof is attractive when it is available, and availability is narrower than most owners assume. It needs a single existing covering, since codes generally cap a building at two. It needs dry board underneath, because sealing moisture in permanently is the one thing this approach cannot survive. It needs a surface flat and sound enough to build on, which rules out a badly ridged or blistered assembly. And it needs structural capacity for the added dead load, which is an engineering question rather than a roofing one.
There is also an attachment question that gets overlooked. Fasteners for the new assembly have to reach the structural deck and hold in it, passing through everything already in place, so a thick existing build-up means long fasteners and a verified pullout value. Where the deck itself is questionable, mechanical attachment may not be viable and the new work has to be adhered, which changes both cost and schedule. When every one of these conditions lines up, building over is the right call and saves real money. When they do not, everything up there still has to come off later, with one more layer added to the pile.
Budget cycles, and phasing as a legitimate third answer
The decision rarely gets made purely on condition, because capital does not arrive on the roof schedule. A useful survey therefore answers a slightly different question than the one usually asked: not only what the roof needs, but how long it can be held safely with targeted repair, and what that holding period costs per year.
That produces a real third option. A roof can be replaced in sections across two or three budget years, with each phase ending at a natural boundary: a parapet, a change in roof level, or a joint the structure already moves at. Areas in the worst condition go first, areas with life left stay under maintenance, and the whole property is not carried as a single capital request. Phasing costs somewhat more in total, since the crew mobilizes more than once and tie-in details have to be built where new work meets old. What it buys is a defensible plan and a spend that fits a cycle.
For a building being sold or handed back inside a defined window, the calculus changes again. Restoring a sound assembly, or repairing and documenting a roof properly, may be exactly the right spend for a short hold. The condition report matters more than usual in that situation, because a documented roof with a known remaining life is a materially easier asset to transact than an undocumented one.
When repair is genuinely the right answer
Not every roof that gets quoted for replacement needs replacing, and the pattern is worth naming clearly. If the plugs come back dry, the deck is sound, the failures trace to identifiable details, and the assembly has years of expected life remaining, then repair is the correct and cheapest intervention. It should come with an inspection interval, so the next failure is found while it is still small.
The same holds for a roof whose surface has weathered but whose structure is intact. A membrane that has lost its reflectivity, or an asphalt cap that has shed granules across the walking routes, is describing a surface problem. Surface problems are addressed at the surface, either by targeted repair or by a properly prepared fluid-applied system, both of which cost a fraction of removal.
Repair is the wrong answer in a specific and recognizable set of circumstances: moisture spread across the field, a deck that has begun to go, two coverings already stacked up, or a maintenance history where the same areas keep reopening season after season.
All four of those are checkable. Ask which of them apply to your building, and ask what evidence supports each answer: which plugs, which photographs, which locations marked on the roof plan. A recommendation with none of that behind it has not been tested against the roof.
Summary
Key takeaways
- The surface is not the evidence. What the board underneath is holding decides most of this.
- Plugs in every quadrant, in every leak area, and in one healthy area give you a percentage to work from.
- Total your real annual roof spend, including interior damage and staff time. It is usually higher than expected.
- A deteriorated deck removes the build-over option and forces removal, whatever the membrane looks like.
- Phasing across budget years is a legitimate third answer, not a compromise.
- Dry plugs, a sound deck, and failures traced to specific details mean repair is the correct spend.
Next step
Get the evidence before the argument
This decision improves enormously once plugs have come out and somebody has written down what they showed. Tell us about the building and we will bring a core cutter, walk the whole field, and give you the percentage, the deck condition and a straight recommendation, including the recommendation to leave it alone.
Questions about the decision
How much wet insulation is too much?
There is no universal threshold, and anyone quoting one precisely is oversimplifying. What matters is the pattern as much as the percentage. A modest wet share concentrated around two failed curbs is a repair. The same percentage scattered across four quadrants with no traceable source means water is entering in places nobody has found, and cutting out those areas will not stop it.
Can we just keep patching until the budget arrives?
Often yes, and that is a reasonable strategy when it is chosen deliberately rather than by default. It works when the deck is sound and the failures are identifiable. It needs a defined holding period, a scheduled inspection interval, and a running total of what the holding is costing, so you can see the point where it stops being cheaper than the alternative.
Does an infrared scan replace cutting plugs?
No. A scan narrows the search by showing where the assembly is releasing heat more slowly after sunset, which correlates with retained moisture. It also flags things that are not moisture, including debris piles, patched areas and differing board types. Every flagged area still gets a plug before it goes into a scope, and the plug is what anyone should be pricing from.
Our roof only leaks in one place. Why survey the whole thing?
Because the leak you know about is evidence of one failure, and the survey is looking for the ones producing no interior symptom yet. Water can travel a long way through insulation and along a deck flute before it appears inside, so a single stain frequently sits well away from the entry point. The areas with no complaints are exactly where an unpleasant surprise hides.
How long should a written condition report stay valid?
Treat roughly a year as the practical shelf life for a report used to make a spending decision, and less than that if the building takes a significant hail or wind event in the meantime. Conditions change, moisture spreads, and pricing moves. A report supported by dated photographs is still valuable years later as history, which is a different use than pricing from it.