Why do some roof repairs fail to stop a leak?

This is the most common failure. Water can travel several metres along underlay, a rafter or a cable before it drops, so the ceiling stain marks the exit. A repair centred on it fixes a part of the roof that was never letting water in.

The corrective step is to work upwards from the highest wet point in the roof space, rather than from the room below.

Mastic, tape and mortar fillets are frequently applied over a failing junction. They may hold for a season, but they cannot accommodate movement and they trap water behind them, so the leak returns and often gets worse.

Where the answer is dressed leadwork, a properly formed valley or a renewed upstand, nothing short of that will last.

Do not simply repeat it. Ask what was found, what was done, and what evidence connected the two. If nobody can answer that, the diagnosis has not been done and it needs doing before more work is paid for.

At this stage a full investigation — roof space, external survey, and controlled water testing if necessary — is almost always cheaper than a third or fourth attempt.

Ask specifically what was inspected the first time: was the roof space checked, was the area upslope of the stain examined, and was any water testing carried out. If the answer to all three is no, the original work was likely based on an assumption rather than a confirmed route, which explains why it did not last.

Also ask whether the same contractor is proposing to repeat a similar approach, or whether they have identified something different this time. Simply repeating a failed method with slightly more sealant is not a meaningfully different repair.

Two failed attempts at the same leak is a reasonable point to bring in a different contractor for a fresh diagnostic look, particularly if the previous work was never documented with photographs or a written explanation of the cause. This is not about assigning blame; it is about getting a diagnosis that has not been anchored to the earlier assumption.

A second opinion is most useful when it starts from the roof space and the weather pattern, rather than from what has already been tried, since beginning from the previous repair tends to repeat its assumptions rather than test them properly.