Improve Translation Performance
Review caching, batch size, and background processing when optimizing translation throughput.
Translation time depends on the amount of content extracted, provider response time, batch size, background-job scheduling, network conditions, and whether matching translations are already cached. The built-in cache may reduce repeat work, but it cannot guarantee a specific speed or provider cost.
Start with a measurable baseline
Choose a representative test post and target language.
Record roughly how many strings are translated and how long the workflow takes.
Review the output and note skipped or unchanged fields before changing performance settings.
Repeat with a comparable post or batch after making one controlled change, such as reducing the batch size.
Avoid changing several variables at once. Provider latency and rate limits can affect timing independently of the cache.
Use translation memory appropriately
TranslateSense checks for matching cached results for the requested target language. The persistent cache is shared by source-string identity and target language rather than by provider, model, or prompt. This can reduce repeat provider requests, but it may also reuse wording generated by an earlier provider or prompt. If you need to evaluate a new provider or prompt, use a controlled new test phrase and review the output.
Chrome Built-in AI is not stored in the persistent translation-results table by this version. Its processing behavior and availability depend on the browser API, language pair, and model readiness. For cache behavior, see Understand Translation Memory and Configure Database Caching.
Tune batch and background processing carefully
Begin with a small batch and increase the batch size gradually while watching response times, errors, and server resources.
Check provider rate limits and connection settings before assuming a slow response is a database problem.
Use Prepare a Bulk Translation Batch and Monitor Translation Progress to prepare and observe bulk jobs.
Review Understand Background Translation Processing and Check WordPress Cron for Translation Jobs if queued work remains pending.
Measure results instead of assuming
Compare similar content and the same target language before and after a change. A cache hit can avoid a provider call for a matching string, but total runtime still includes extraction, editor mapping, database access, provider calls for cache misses, and applying results. The plugin does not promise a fixed percentage improvement or a fixed reduction in API charges.
If translation is still slow
Confirm that the provider connection works and that the selected model is available.
Check whether the slow step is extraction, provider response, queue processing, or applying content.
Inspect XML mappings when content is unexpectedly skipped; mapping issues can look like incomplete processing rather than a speed issue.
Check WordPress cron and the bulk job status for background work.
Keep a note of the test content, provider, language pair, batch size, elapsed time, and errors so results are comparable.
Screenshot to add: [Translation performance test and batch progress]
