Hi,

we've recently had an issue were we created 2 projects from a single source file to two different target languages; both projects had a different TM attached.

After the initial analysis, the project showed a discrepancy in the analysis statistic. The 'Total' field in the Analysis had a 12 word difference. However, the 'Confirmation Statistics' showed both projects to have the exact same Word, Character and Segment count. The Reports tab also indicated the discrepancy:
the 'translation count' page looked identical for the two project in word count, the 'analyze files' page showed a 12 word difference. The one project in itself showed a 12 word discrepancy between the Translation Count page and the Analysis Page?

But, and this is the real issue: after running the batch task - Analyze files again , the analysis discrepancy disappeared? Both projects now have the exact confirmation and analysis counts, as you would expect.
This indicates the analysis was faulty the first time around. We use this analysis to calculate project costs.
Can we trust the analysis count, as it appears to bug out sometimes? Or could this be related to the translation memory used?

Any feedback on this would be appreciated.

Kind regards,
Geert