How should an agency triage accessibility alerts across a client portfolio?
One dashboard, twenty stores, hundreds of alerts. Triage across a client portfolio means sorting by client risk first, repeating template patterns second, and single low-impact issues last.
Portfolios do not fail one issue at a time
An agency watching accessibility across twenty client stores does not have an alert problem. It has a triage problem. A single theme update can fire the same alert across six stores at once, and a quiet Tuesday can surface a critical checkout barrier on the one client already holding a demand letter. Treating every alert the same means the urgent ones wait behind the trivial ones.
Triage across a portfolio works in three layers: client risk first, pattern second, individual issue last. Get the order right and a small team can keep twenty stores moving. Get it wrong and the dashboard becomes wallpaper.
Layer one: sort by client risk, not by alert severity
The same missing form label is a different problem on different clients. On a store that received a demand letter last month, it is evidence of non-compliance with a deadline attached. On a quiet store with no legal history, it is a routine fix. Severity scores from scanners do not know this. Your triage board should.
Tag every client with a risk tier: active legal matter, prior complaint, high revenue at risk, or standard. Alerts from tier-one clients jump the queue regardless of their technical severity. This feels unfair to the clean clients. It is the only fair way to run a portfolio, because legal exposure is measured per client, not per issue.
Layer two: hunt the repeating pattern
Agencies reuse themes, sections, and app stacks across clients. That reuse is an efficiency until one shared component ships a defect, and then it is a portfolio-wide incident. When the same alert appears on four stores in a week, stop fixing it four times. Find the shared source.
The triage move is to group alerts by probable cause before assigning any of them. Same theme version, same app, same custom section: one root cause, one fix, one rollout. Document the fix once and apply it everywhere, then verify per store. This turns twenty tickets into one, and it is the main reason portfolio monitoring beats per-store monitoring.
Layer three: rank what remains by user impact
After risk and patterns, what is left is the long tail of individual issues. Rank these by where they sit in the shopper journey, not by WCAG criterion number. Purchase path first: cart, checkout, product pages with add to cart. Then navigation and search. Then footer and policy pages. A contrast failure on a checkout button outranks a contrast failure on a careers page every time, even if the scanner scores them identically.
Within the same page, rank by barrier type. Anything that blocks task completion (keyboard traps, unlabeled required fields, broken focus order) beats anything that degrades the experience (minor contrast, redundant links, imperfect heading order). The question is always: can the shopper finish what they came to do.
Set the cadence per layer
Each layer gets its own rhythm. Client-risk alerts get same-day review. Pattern hunts happen weekly, because patterns need a few days of data to emerge. The long tail gets a monthly pass, batched into the retainer's regular maintenance window. Three cadences, three different meetings, no alert older than its layer allows.
Report to clients at the layer they care about. Tier-one clients get incident-style updates: what fired, what it means for their matter, what is fixed. Standard clients get the monthly summary: issues found, issues fixed, trend line. Nobody wants the raw alert feed. The agency's job is to turn the feed into a story per client.
Keep the board honest
Triage systems rot when alerts are closed without fixes. Require every closed alert to carry a resolution: fixed, accepted risk with a client sign-off, or false positive with evidence. "Closed" with no note is how the same alert reappears next month and how a client discovers, during litigation, that you saw their barrier and did nothing.
Review the triage rules quarterly. Client tiers change, themes get replaced, and the pattern you hunted in January may be gone by April. A triage board that never changes its rules is not stable. It is stale.