It’s a dog!

If you believe everything that's been written about AI text analytics recently, you'd think everything had been sorted and we were finally in the land of milk and honey.

Just tip your unstructured data into the AI machine and hey presto, out comes perfect categorisation and sentiment scoring. All those years of struggling to get machines to interpret customer comments to an acceptable level - all fixed in the last few months!

But then, if you already use text analytics on your unstructured voice of customer feedback, I guess you've realised that it's not that simple - and not really always that accurate.

Recognising that a dog is not necessarily a dog is not that simple. And language is full of colloquialisms and other lingustic traps, constantly tripping up machine analysis.

Every text analysis system varies in its methodology and in its ability to categorise accurately. The key thing is to make sure that it is continuously trained on your particular data so that the quality gradually improves.

Then you have to figure out who is going to do that continuous improvement - is it you, or is it a managed service provided by the vendor? Not many vendors provide that service, and not many users have the resources to methodically work through the improvements. Unless someone does it, you'll fall short of the full promise of the text analysis insight.

#VOC #CX #customerexperience #textanalysis #textanalytics #voiceofcustomer #voc #customerinsight