Until it gets implemented, here's a workaround that outputs the feature counts of the FeatureWriters to the FME log and also makes it possible to use the feature counts in the shutdown script etc. using the standard mechanisms.

The solution basically consists of re-creating the same number of dummy features and do a fanout to the NULL writer. Since the dummy features contains no geometry and hardly any attributes, it shouldn't be too penalizing performance wise:

'

From the FME log after execution:

'

I had to replace a reader by FeatureReader and need the statistics for the FeatureReader

'

Yes, the FME server report writer count should include both the counts from featurewritter and conventional writer.

Whilst we're at it here, why not consider changing the wording of the message entirely? Instead of "Total Features Written", how about "Total Features sent to be Written"... It would make explaining why nothing has actually been written due to rejected features much simpler!

Could really use some traction on this one after 5 years!

This does happen @timboberoosky, if you look back through the logs it produces a summary of features written for any FeatureWriter transformers in the workspace at the point at which they write their features.  Sometimes this can make it seem as though the process has completed, however there may still be other features to be written.  Often you end up with a few of these summaries in the log if you have a number of FeatureWriters in there. 

See below, with my FeatureWriter_3 doing some early writing of features in a long process:

Grouping into an overall summary at the end - perhaps indicating which FeatureWriter or writer wrote however many features to whichever feature type - is probably the thing that is missing here.

Maybe it’s just me but I can’t see the screenshots - could you post these again? I feel like this might really help but I can’t see it! Thanks!