Repair Order: Fixing Leaks in Sequence and Verifying the Fix

Repair leaks in the order the prioritisation method ranks them - size of unintended loss, strength of evidence, cost of repair - one stage at a time. Verify each repair on its own stage's step rate (same segment, whole weeks, promotions noted), then confirm the chain's end moved too, so the gain is real rather than displaced to the next stage. Declare a leak closed only when the stage rate has held its improvement across two review cycles; a single good month after a repair is weather until proven climate. The quarter's leak work, logged this way, compounds into the funnel map the audits bench uses.

A realistic quarter for a small store holds two to four stage repairs, not twelve. Sequence them so each repair's read window completes before the next overlapping change ships where stages interact (product page and add-to-cart interact; arrival and post-sale do not). Log the intended order with its reasoning; when the quarter is interrupted - and it will be - the log lets the sequence resume instead of restarting from the loudest new opinion. Resist the mid-quarter insertion of an unrelated redesign; park it in the deferred list with its reason, exactly as the prioritisation guide prescribes.

Three checks. The stage rate moved as predicted, in the segment the repair targeted. The end of the chain (orders, qualified conversations) moved consistently - or its failure to move is itself the finding that the leak was misdiagnosed or borrowed. And the qualitative stream agrees: recordings at the stage show the old behaviour reduced, tickets naming the cause decline. Two of three is a provisional close; all three across two cycles is a close. Anything less keeps the leak on the watch list, where honesty costs nothing and premature closure costs the quarter.

Stages trade losses. A stricter arrival (better targeting) raises every downstream rate while total orders hold - an improvement in efficiency, reported honestly as such. A product page that pushes undecided visitors into the cart raises adds and lowers checkout completion; the chain's end is the referee. Report repairs against both the stage and the chain, in the same log entry, so displacement is visible as displacement rather than celebrated as two half-victories by two different pages' owners - who, in a small business, are both you, arguing with yourself in the notes.

A flat read after an honest repair is information, and the log should say which kind: the cause was wrong (re-diagnose with the recordings guide), the repair was weaker than the cause (a copy patch against a structural problem), or the stage's traffic changed underneath the read (segment discipline catches this). Roll back only when the repair made things worse or costs maintenance without benefit; a neutral repair addressing a real, evidenced friction may stay while the next hypothesis is prepared. The failure to avoid is the unrecorded shrug - the repair neither verified nor questioned, quietly becoming next year's mystery.

End the quarter by redrawing the funnel beside the quarter's start: which stages tightened, which moved adversely, which were untouched. The comparison is the leak programme's report card and the next quarter's priority input. Stores that do this four times own something genuinely rare: a measured, causal history of their own funnel, against which every future claim - from tools, agencies or this site - can be checked. That history, more than any single repair, is what this bench exists to build.

Verification needs its mirror: the rule for stopping. Leave a stage when its remaining loss is mostly intended (the evidence says the leavers were never buying), when the next repair's cost exceeds the recoverable orders at your margins, or when two honest repairs have failed to move it and the diagnosis needs the quarter's distance before a third attempt. Record the stop and its reason; a leak consciously left is a decision, while a leak silently abandoned returns as next year's surprise audit finding. The funnel will never be tight everywhere, and the mature operator's skill is knowing which looseness is the business model and which is money on the floor.