Choosing What to Fix First: A Prioritisation Method for Small Stores

Prioritise by three factors, scored roughly and honestly: how many buyers the problem plausibly costs (traffic reaching the step times the size of the drop), how sure you are the problem is real (evidence strength, from a guess up to recordings of people failing), and what the fix costs in time and risk. Work the top of that list, one or two items a month, and re-score monthly as the funnel changes. The method matters less than the discipline: a written order, defended by numbers, beats a month of fixing whichever page annoyed you most recently.

A problem's cost is bounded by the traffic that meets it. A confusing returns policy read by 30 people a month cannot cost more than 30 decisions; a slow product page seen by 3,000 can cost hundreds. Estimate the leak as: sessions reaching the step, times the shortfall you believe exists. You will not know the shortfall precisely - a step rate visibly below its neighbours, or recordings of repeated failure, are enough to rank. Precision is not the point; order of magnitude is. Big leaks first is the whole of factor one.

Rank evidence in this order: people observed failing (recordings, usability sessions, support complaints quoting the page) beats step-rate anomalies, which beat heatmap impressions, which beat a competitor doing it differently, which beats a blog post saying everyone should. Weak-evidence items are not banned - they are queued behind strong-evidence ones, or converted into research tasks: watch ten recordings of the page before rebuilding it. This factor is where small stores save the most time, because most fix lists are half folklore.

Cost is your hours plus the chance of breaking something that works. A text change on a product template is an afternoon and reversible. A checkout overhaul on a platform you half-control is a week and touches the page that takes money. All else equal, cheap and reversible goes first - not because it is most valuable, but because a small store's scarcest resource is the owner's attention, and quick certain wins fund the patience for bigger ones. Note platform limits honestly: some checkout fixes are simply not available on some plans, and a fix you cannot ship scores zero however big the leak.

Illustration: a store's candidate list holds (a) product pages load slowly on mobile, step rate down, fix is image compression - big leak, strong evidence, cheap; (b) add reviews to product pages, currently none, fix is an app and moderation time - medium leak, medium evidence, medium cost; (c) redesign the homepage hero - unknown leak, weak evidence, medium cost; (d) rewrite checkout reassurance copy - small leak, weak evidence, cheap. The method orders them a, b, d, c. Without the method, (c) usually wins because it is the most visible - which is exactly the failure this page exists to prevent.

Fix the list monthly, at the same sitting as your baseline review. New evidence arrives (a step rate moved, a complaint pattern formed), finished fixes leave, and the order reshuffles legitimately. What you must not do is re-prioritise mid-month because a podcast mentioned a tactic. The list is a queue, not a mood board. Stores that hold a queue for six months learn their funnel deeply; stores that chase weekly tactics learn nothing twice.

Every prioritised list creates a second, shadow list: the ideas explicitly deferred. Keep that list written down with a one-line reason - too costly this quarter, evidence too weak, platform cannot do it yet. Deferred ideas have a habit of returning as urgent suggestions in meetings and podcasts; a written deferral with a reason answers them without a fresh argument each time. Review the deferred list at the quarterly re-score: some items graduate when traffic grows or a platform adds the feature, and a few get deleted as the store's shape changes. Prioritisation is as much the discipline of recorded refusal as of selection, and the recorded refusals are what keep a one-person business from spending the year on other people's priorities.