Skip to content
Sima.Journal
← Back to the journal

Use WordPress 7.2 RC1 to rehearse your client upgrades

Sima editorial · 20 November 2026 · Analysis · Release candidate announced 17 November 2026
Three miniature houses beside a checklist representing client website upgrade rehearsals.
AI-generated editorial illustration

An agency maintaining twenty WordPress sites has twenty different answers to “does the update work?” One client needs to publish vacancies. Another takes deposits. A third relies on a custom dashboard that only the office manager sees. A useful rehearsal starts with those jobs and leaves enough evidence for someone else to understand what was checked.

WordPress 7.2 RC1 arrived on 17 November, with the final release scheduled for 8 December 2026. It remains under development. WordPress advises testing it on a test server and site, away from production and mission-critical websites. Treat the coming weeks as time to find compatibility problems and prepare the eventual upgrade decision.

Choose a sample you can explain

Consider Fieldwork, a fictional agency maintaining eighteen sites: ten brochure sites sharing a block theme, five shops with two different payment integrations, and three membership sites with custom administration screens. This is an illustrative planning example; we have not run these tests or observed results.

Fieldwork initially selects five copies: one brochure site, one shop for each payment integration, and two membership sites covering the distinct custom dashboards. The brochure example includes the oldest supported theme variant and the busiest enquiry form. That gives the team a reason for its selection, although the remaining nine brochure sites may still have individual differences.

Before testing, list those differences: theme overrides, forms, multilingual plugins, hosting configuration and client-specific code. Add another representative wherever a difference introduces a workflow the sample does not cover. Keep an explicit list of sites awaiting their own checks. Five selected sites tell you about those five configurations; they cannot establish compatibility across eighteen installations.

Create controlled test copies with personal data removed or replaced. Before allowing those copies to run, isolate live payments, outgoing customer messages and production webhooks, and substitute sandbox credentials. Check scheduled jobs too. Route test email to a mail sink that captures messages without delivering them to customers. Use named test accounts matching the client’s actual roles. A successful administrator session says little about what a shop manager or editor can do.

Record the starting WordPress, PHP, theme and plugin versions, plus custom-code revisions and relevant hosting settings. Run the chosen journeys before changing core. Then install RC1 on those test copies and repeat them with the same inputs. Keep unrelated plugin updates out of the first comparison so a new failure has a smaller set of possible causes.

The November developer roundup identifies two useful prompts for bespoke admin screens: SVG menu icons and changes to admin asset loading. Check icon styling that targets ::before, and screens whose code assumes scripts load in the page head. The roundup also identifies React 19’s default activation as a Gutenberg plugin change, rather than a WordPress 7.2 core change. Record the plugin version and experiment settings where it is installed; those results belong to that configuration, and default activation does not establish which React version a particular site actually uses.

Keep a record at workflow level

Give each tested installation a configuration ID linked to its exact version inventory. The following matrix is Fieldwork’s proposed test sheet. Every observed result is deliberately pending. Replace those cells only after someone runs the check.

Configuration and versionsWorkflow and expected resultObserved result and evidenceOwner
B1: baseline → 7.2-RC1; attach PHP, theme, plugin and code inventoryEditor replaces a service-page image, publishes the change on the test copy and reopens it; the selected image and alt text persist in the rendered pageNot run; capture saved content and matching desktop/mobile viewsPriya, content QA
B1: same inventoryVisitor submits a valid enquiry; one record is stored and one message reaches the test mail sink; invalid input retains an actionable errorNot run; retain record ID and captured messageLeo, developer
S1 and S2: separate inventories, each baseline → 7.2-RC1Customer completes sandbox checkout; the expected order status, totals and stock change are recorded onceNot run on either configuration; retain sandbox transaction and order evidence separatelyAmira, commerce QA
M1 and M2: separate inventories, each baseline → 7.2-RC1Restricted staff user opens the custom dashboard and edits an allowed record; saved values survive reload; an attempted prohibited edit is rejected by the server and leaves the record unchangedNot run on either configuration; capture role, screen, request outcome and saved recordSam, developer

Add the execution date, browser and viewport to each result. A screenshot can establish a visual difference; a saved record or captured message answers a different question. Keep both where the workflow needs both. A message in a mail sink proves capture, not delivery to a real inbox. If mail cannot be exercised in the chosen environment, mark the check untested and assign a follow-up.

When a check fails, write the shortest reproducible sequence, expected behaviour, actual behaviour and configuration ID. Compare it with the baseline, then assign the investigation to a person. Avoid turning an unexplained error into a confident claim that core or a particular plugin is responsible.

For client communication, report the scope plainly: “We are rehearsing your upgrade on a test copy. Checkout checks are pending; we have not scheduled your live update.” Once the final release is available, recheck its notes and relevant vendor guidance, refresh the configuration inventory and rerun critical journeys against that build. Production scheduling should follow those results, with a restorable backup, rollback procedure and named person checking the live workflows afterwards.