What should an agency hand off when an accessibility retainer ends?

A clean accessibility handoff is a living file, not a final report. When the retainer ends, the agency should hand the client's team four things: the full scan history, the open backlog ranked by user risk, the evidence archive in one place, and a single walkthrough that teaches the three habits that keep a store compliant. Anything less and the store starts decaying the week the agency leaves.

The final report is not the handoff

Most agencies close an accessibility engagement with a polished PDF: the final scan, the fixes shipped, the conformance statement. The client files it and nobody opens it again. A final report is a snapshot. A handoff is a transfer of capability, and it looks very different. The test is simple: six months after the agency leaves, can the client's team answer what broke, what still needs fixing, and how to check? If the handoff was only a PDF, the answer is no.

Build the handoff as a working folder the client's owner can actually use. A shared drive or repo with a clear top-level README, not a 40-page document. The README should name the current owner on the client side, link the scan history, the backlog, and the evidence archive, and list the three habits below. If the new owner can find everything in five minutes, the handoff worked.

Hand over the scan history, all of it

The value of an agency engagement is not just the final state of the site. It is the history: what was broken at onboarding, what got fixed, what regressed, and when. That history is how the next person learns what to watch. Export every scan the agency ran, dated, in a consistent format, and put them in one folder. If the scans live inside a proprietary tool the client cannot access after the contract ends, export them now.

Include the trend line, even the embarrassing parts. If the store regressed in March because of a theme update, that entry teaches the client's team more than a year of green scans. Regressions are normal; pretending they never happened teaches nothing.

Rank the open backlog by risk, not by effort

Every engagement ends with an open backlog. The mistake is handing it over sorted by fix effort or, worse, unsorted. The client's team has limited attention, so rank by user risk: barriers that block checkout or account access first, then barriers that block product discovery, then everything else. A keyboard trap on the checkout is worth more than fifty low-contrast labels on the about page, and the backlog should say so explicitly.

For each open item, keep the reproduction steps the agency used. A finding without reproduction steps is a rumor. With steps, it is a ticket the client's developer can pick up on day one without asking the agency anything.

Transfer the evidence archive in one place

Evidence is the unglamorous backbone of accessibility work: dated scans, fix records, test notes, the accessibility statement, VPATs if they exist. During the retainer the agency accumulates this across tools, emails, and ticket comments. The handoff is the moment to consolidate. One folder, one naming convention, everything dated. This archive is what the client shows if a demand letter ever arrives, and it is worthless if it is scattered across three inboxes.

While consolidating, note what is missing. If nobody ever ran a screen reader test on the checkout, write that down. An honest gap list is more useful than an archive that implies completeness it does not have.

Run one 30-minute walkthrough, then stop

The walkthrough is the highest-value hour of the whole handoff, and it should be short. Thirty minutes, one screen share, the client's named owner. Show them where the scans live, how to read the backlog, and how to run the two-minute keyboard check that catches the most common regressions. Then teach the three habits: check contrast on every new banner before it ships, re-scan after every theme or app update, and keep the evidence folder current. That is the whole curriculum.

Resist the urge to train them on everything. A overwhelmed owner adopts nothing. Three habits, one walkthrough, one README. The client's team can learn the rest when they need it, and the archive is there when they do.

Leave the monitoring decision in writing

The question every client asks at the end: do we still need monitoring? Answer it in writing, not in a call. If the store changes themes or installs apps frequently, say so plainly: unmonitored stores regress, and the evidence shows it. If the store is stable and the team is trained, say that too, and name the quarterly check that replaces continuous monitoring. Either way, the decision should be a paragraph in the handoff, signed off by the client owner, not a verbal aside that nobody remembers.

This is also the moment to be honest about the decay curve. Accessibility is not a state, it is a practice. Without any monitoring, most Shopify stores develop new issues within a quarter. Put that in the README so the decision not to monitor is an informed one.

Make the handoff a deliverable in the next contract

If this handoff felt improvised, fix it for next time. Add the handoff to the contract as a named deliverable with acceptance criteria: scan history exported, backlog ranked, evidence consolidated, walkthrough completed. What gets specified gets done. The agencies with the best client retention are not the ones that never leave; they are the ones that leave so cleanly the client comes back for the next project.