Why the reviewer sees older text than the editor's preview

Two people can open what appears to be the same draft and see different wording. Before blaming a cache or a language route, establish whether both screens are reading saved material. An editor's browser may display local changes that have not reached the shared system.

This fictional investigation follows draft D-62 through that boundary. The CMS in the example supports explicit saving and preview of a saved checkpoint. Its behaviour is inspected rather than assumed to apply to every editor or platform.

Record the two views

The editor changes the introductory sentence of D-62 from “The report covers the pilot period” to “The report covers the pilot period and the follow-up review.” The local editing screen and its embedded preview show the new sentence.

The editor then copies the existing review link to a colleague. The colleague sees the earlier sentence. Both people are using the intended publication and language edition; neither has opened another report.

The evidence distinguishes the views:

The editor's screen carries an unsaved-change indicator, but the current handover instructions do not explain that its embedded preview can use local state. The review link does not include those local changes.

Locate the missing operation

Inspection of the supported workflow confirms that generating a shared review link uses a saved checkpoint. It does not transfer the editor's current browser state to the reviewer. The page has not lost a successful save; the editor has not performed the save required for the intended review.

That distinction changes the correction. Clearing the colleague's browser cache would not make the unsaved sentence appear. Recreating the entire publication could introduce a duplicate without addressing the same missing step.

The team keeps the existing draft identity and uses the platform's explicit save action. The save returns checkpoint C-63. The editor verifies that C-63 contains the new sentence before generating the review link for that checkpoint.

Check the shared material from another account

The colleague opens the new link through the permitted reviewer account. The view identifies C-63 and shows the sentence including the follow-up review. The editor's saved preview shows the same text. A separate publication D-63 remains on its previous checkpoint, with no changes to its content.

The editor then types another local change in D-62 but does not save it. The existing C-63 review link continues to show the saved sentence. The local interface identifies the new work as unsaved. This control establishes that shared review is deliberately tied to a saved state rather than silently following every keystroke.

If the platform's review links always follow the latest saved state instead of a fixed checkpoint, the instructions would need to say so and use that supported behaviour. The fictional example has verified checkpoint links; a real project must inspect its own capability.

Improve the procedure without inventing a new workflow

The revised instructions place three actions together: save the intended material, verify the saved preview and create the review link. They explain what the local preview represents and where the save confirmation appears.

The team does not add automatic saving merely to remove the need for that explanation. Autosave would introduce its own decisions about incomplete text, shared visibility and review timing. It may be useful, but it is a separate product choice, not a necessary consequence of this investigation.

The reviewer also receives a specific task and the checkpoint being reviewed. The review request no longer relies on “the version I am looking at” as an identity that another person cannot inspect.

Handle a failed save honestly

The process includes a stop condition. If the save does not return the expected confirmation or the saved preview still shows older text, the editor does not send a link claiming the change is ready. They retain their local work according to the platform's supported method and seek help with the save failure.

This prevents a procedural correction from concealing a different technical problem. The demonstrated case is an omitted save; a later storage or connection failure needs its own evidence.

For your own workflow, record the local state, saved state and reviewer state before choosing a fix. The editorial-workspace service can help make those boundaries understandable, while the development process includes an ordinary editor task in acceptance.