A flashing failure is a lap failure. Sealing it is not repairing it.

Behind every cladding there is a water-resistive barrier — house wrap, building paper, or a fluid-applied coating — which is the wall’s drainage plane. It is not decorative and it is not backup: cladding leaks by design at every lap, nail and joint, and the barrier is what carries that water back out. Everything that penetrates or terminates against a wall has to be integrated with it, and “integrated” means lapped in the right order.

The Building America Solution Center states the rule for this joint without ambiguity: the upturned leg of the step flashing is “behind the vertical drainage plane (house wrap) on the wall or sealed to it”, and, more generally, “the top layer of the weather-resistant barrier (WRB) or flashing laps over the bottom layer”. On the correct side, water running down the barrier reaches the outside face of the metal and is carried out onto the roof. On the wrong side, the barrier delivers that water to the back of the metal and down past the level of the deck, into the wall.

Why this one is so persistent

• It is invisible on a finished building. Both arrangements produce the same photograph from the ground and the same photograph from a window.
• It usually produces no interior symptom, because the water is inside a wall cavity rather than above a ceiling. The first person to find it is often a buyer’s inspector or a siding contractor a decade later.
• It cannot be corrected from the roof side. New flashing slid up under existing siding cannot be lapped with a barrier nobody can reach, so the repair is finished with sealant and the original geometry survives underneath.
• It sits on the boundary between two trades. The metal is a roofing component, the barrier is a wall component, and a proposal from one trade is frequently silent about the other.

The practical consequence is a single question, and it is the one worth asking before agreeing to any roof-to-wall repair: is the cladding coming off, and will the barrier be re-lapped over the upturned leg? If the answer is no, the work being proposed is a sealant maintenance visit. That can be a reasonable thing to buy — but it should be bought knowing what it is.

A kickout is the small angled piece at the bottom of a roof-to-wall run. Model code text asks for it by description rather than by name — R903.2.1: “a flashing shall be installed to divert the water away from where the eave of a sloped roof intersects a vertical sidewall”, quoted here from the 2021 Seattle Residential Code, where that sentence is adopted law. The RICOWI roofing guide describes the same component as “the bottom portion of a vertical flashing component that angles away from the building”, which “prevents capillary action behind the flashing”. For the geometry, and for how much water arrives at that corner, read the anatomy page; what follows is what happens once it is already missing.

Why it is diagnosed backwards

Almost every other flashing failure is reported by someone who saw a ceiling. This one is not, because the water goes into a wall. The Building America Solution Center lists what it reaches — “wall sheathing, framing, insulation, and mold inside the wall cavities” — and then explains why nobody is told: older painted wood siding announced water damage by peeling, but “new wall claddings like fiber cement, vinyl siding, and brick veneer can mask the evidence for years”.

So the reporting order is inverted. The evidence appears outside before it appears inside, if it appears inside at all, and it appears on the wall rather than on the roof. That has two consequences worth acting on: a proposal that inspects only the roof has not looked at the evidence, and the time between the first wetting and the first complaint is measured in years. EPA’s guidance is that materials dried within 24 to 48 hours will in most cases not grow mold; a closed cavity taking a concentrated stream several times a month is never in that window.

What a kickout repair has to include to be a repair

The piece itself is trivial to buy and quick to fit, which is why its line on a proposal tells you almost nothing. These are the parts that decide whether the work holds, in the order they are usually left out:

1. The barrier re-lapped around it. The diverter has to be integrated with the wall’s drainage plane the same way the step flashing above it is — which means the cladding at that corner comes off. A kickout screwed to the face of intact siding and sealed has moved the water a few inches further out and left the lap exactly as it was.
2. The gutter extended under the intersection. A diverter discharging past the end of a gutter has relocated the problem rather than ended it, and gutter work is frequently a different contractor on a different day. It belongs in one scope with one person answerable for it.
3. An assessment of what the omission already did. The sheathing and framing behind that corner have been taking the flow for as long as the detail has been missing. That is the line nobody quotes until the cladding is off, and it is the one that decides what this costs — so it belongs in the proposal as an allowance with a unit rate, not as a surprise.
4. Evidence that the cavity is dry before it is closed. Correcting the geometry over wet framing seals the consequence inside the wall and removes the only symptom anybody had.

Three of those four happen in the wall rather than on the roof, which is the practical reason a kickout retrofit priced as a roofing repair is usually pricing the metal and nothing else. Ask which of the four are inside the number in front of you. That question, put before anyone starts, is worth more than any observation available from the ground.

Flashing is fastened to things that do not move the way it does. Wood framing, brick and the metal itself all change length with temperature, at rates that differ by a factor of five. Over a short piece nobody notices. Over a long run of edge metal, a chimney counterflashing, or an apron across a headwall, the difference has to go somewhere: into a joint that can slip, or into the fastener holes, or into a buckle.

The Insurance Institute for Business & Home Safety publishes coefficients of thermal expansion for common building materials and the movement each shows over a 10-foot length for a 100 °F change. The table below applies simple arithmetic to those coefficients — movement (in) = coefficient × 120 in × 100 °F — and then subtracts the substrate’s movement from the metal’s to get what the fasteners and joints actually have to absorb.

A worked example: forty feet of aluminum drip edge

Take an ordinary eave: four 10-foot lengths of aluminum drip edge, lapped and nailed to a wood deck. Each length wants to move 0.155 in over a 100 °F swing while the deck under it moves 0.037 in — a differential of about 0.118 in per length, and roughly 0.47 in across the run if the pieces are locked to each other.

IBHS asks for a joint gap of “1/8–1/4 inches wide between sections to allow for thermal movement” and for fastener clearance holes to be “oversized or slotted to enable movement without causing fastener damage”. Now put that beside the model code, which requires drip edge to be “mechanically fastened to the roof deck at not more than 12 inches (305 mm) o.c.” The two requirements point in different directions, and they are both right: the metal has to be held down against uplift and allowed to move.

That tension is the honest explanation for a failure everybody sees and few name — edge-metal joints that open, nail holes that elongate into slots, fastener heads that stand proud after ten summers. It is not evidence that anything was installed contrary to code. It is evidence that a run of thin metal on a sunlit edge is a moving part, and that the repair for it is a lapped joint that can slip, not a bead of sealant that cannot.

The same arithmetic explains a specification choice: aluminum moves roughly 2.7 times as far relative to the wood it is nailed to as galvanized steel does. That is a real reason to ask what metal is going back on a long run, and it is a reason that has nothing to do with corrosion.

The other one: putting two wrong metals together

IBHS states the mechanism in a sentence — “when two dissimilar metals come in contact in an electrolyte solution, galvanic corrosion will occur” — and the risk rises the further apart the metals sit on the galvanic series. Their published order runs, from most protected to most corroded: gold, titanium, passive 304 stainless, copper, yellow brass, nickel, naval bronze, tin, lead, active 304 stainless, mild steel, aluminum, cadmium, zinc and galvanized steel, magnesium.

Two practical consequences for a repair, both of which get created accidentally:

• Aluminum below copper. Aluminum sits near the corroded end and copper near the protected end. New aluminum flashing installed downslope of an existing copper valley, or under a copper gutter, is being asked to corrode. IBHS’s remedy is to keep metals close together on the series or “separate them with an electric isolator such as a rubber membrane or bituminous paint”.
• Aluminum against treated wood. IBHS notes that “many pressure-treated wood solutions contain copper and should be treated as if it were copper”. A repair that installs fresh treated blocking behind a flashing has introduced a copper source against the metal, in exactly the place that stays damp.

The fastener counts too: IBHS specifies that fasteners “should be galvanically compatible with the materials with which they will be in contact”. This is not an exotic consideration. It is a line in a proposal, and the absence of that line is the finding.

Stage 1 — before anyone visits: record what you actually know

1. Where is the symptom, precisely? Room, wall or ceiling, and how far from the nearest outside wall. Photograph it with something for scale.
2. When does it appear? During rain, hours after rain, only in wind-driven rain, only during a thaw, or on cold clear nights with no rain at all. This single answer separates a flashing question from an ice question from a condensation question.
3. How long has it been happening, and has it been repaired before? Find the invoices. What the last repair consisted of is the most useful evidence you own.
4. What is directly above and uphill of it, on the outside? A wall intersection, a chimney, a valley, a vent stack, a skylight, an eave. Photograph each from the ground.
5. Is there a symptom with no interior stain? Walk the outside and look below every roof-to-wall run for discoloration, bubbling paint, or trim and siding that has changed shape.

Stage 2 — what a competent visit produces

1. A stated hypothesis about which detail failed, with the evidence for it. Not “the flashing” — which flashing, and why that one.
2. Photographs of the detail as found, taken from the roof, not from the driveway.
3. An answer to the sealant question: would this repair still work if the sealant failed tomorrow?
4. An answer to the opening question: what has to come off to reach the lap, and who reinstates it.
5. An answer about the damage already done, and who is pricing that. If the answer is “we won’t know until it’s open”, that is correct — and it should arrive as an allowance with a unit rate, in writing, before work starts.

Stage 3 — the four questions that classify the proposal

Ask these in order. The answers place the proposal into one of three categories, and the category is what you are actually deciding between.

1. Does the repair depend on sealant to work?
2. Is anything being opened — cladding, covering, masonry?
3. Is the water-resistive barrier being re-lapped, where the detail meets a wall?
4. Is anyone establishing that the assembly is dry before closing it?

• A repair. Something is opened, a lap is rebuilt, the barrier is re-lapped, the assembly is checked. Expect it to cost what opening the assembly costs, and expect the price to be dominated by the cladding or the masonry rather than by the metal.
• A maintenance visit. Sealant renewed, nothing opened. This is a legitimate purchase with a known interval — but it should be described as one, priced as one, and scheduled. A maintenance visit sold as a repair is the mechanism behind almost every “we’ve had this fixed three times” story.
• A postponement. Nothing opened, nothing renewed, and an explanation of why the leak is not serious. Sometimes true — a condensation pattern, for instance, genuinely is not a flashing problem — but it needs to come with the alternative diagnosis attached, not merely with reassurance.

Stage 4 — what goes in writing before work starts

The list in the questions section above doubles as the specification. The three lines most often missing, in order of how much they cost when absent: whether the cladding is being opened; whether the barrier is being re-lapped; and the allowance and unit rate for decayed sheathing or framing found behind it.

If you are comparing more than one proposal, they are only comparable once all three of those are stated in each. See how to compare roofing quotes for normalizing scope before price, and repair or replace if the covering is close enough to the end of its planning range that opening a wall now is buying the same work twice.