Mastering AutoClassification: Pro Tips (Final Instalment)

These final 5 Pro Tips focus on anticipating adverse scenarios so you can ensure a successful implementation from the start.:

#21. Anticipate Opposition
Expect some resistance. AutoClassification is literally the task of understanding enterprise content and marking, or tagging, it for easier management, analysis and retrieval down the line. This notion can scare some employees who prefer the anonymity and invisibility of uncategorized data, particularly if they are engaging in questionable behaviors. Our suggestion to counter this kind of resistance is to give employees some advanced notice that classification efforts are about to begin and allow them ~ 2 weeks to “clean up” any content not under legal hold. By giving them both a voice and an opportunity to self-correct, you will save great headache once the efforts are underway.


#22. Be Prepared for “Unsavory” Content
Yes, we mean pornography. Sadly, many employees view corporate data stores as viable places to hold their personal information, perhaps unaware that it might be discovered. (See Tip 21 above.) For many years enterprise systems were far more sophisticated than personal ones at home, so enterprising staff found and kept a lot of questionable material at work. Since most enterprise data has never been categorized, let alone assessed for deletion, most of that unsavory content is still resident. Yes, you should expect to find pornography in your data stores, along with pirated media (both video and audio), as well as potentially documented indications of questionable, if not outright illegal, behavior. The trick is to be prepared for this inevitability and have a plan in place to remediate and act upon it, assuming you will uncover it at some point.


#23. Expect Exceptions
Handling the “edge cases”. As strong as AutoClassification technology is, there are always some files that will not yield meaningful results. This happens when the files themselves have issues, such as bad scanned image quality, or corrupted or password-protected contents. Sometimes the files literally have no content at all, or have only handwritten contents, which prove tricky indeed for text-based analysis. In general, expect < 1% of enterprise files to have such obstacles, and prepare a suitable method of exception handling.


#24. Count on Foreign Language Content
Global organizations use multiple languages. While much of business content is in English, there are many other world languages that are commonly present in global operations. Be prepared for German, Spanish, French, Japanese, Chinese, and Arabic, just to name a few common ones. Fortunately AutoTranslation is an easy fix for supplementing the native language with English – helpful for both analytics processing and search & retrieval.


#25. Foretell the Follow-on
AutoClassification success breeds more IG opportunities. There’s something about AutoClassification that captures everyone’s attention. Maybe it’s watching software “magically” create rich metadata tags. Maybe it’s the resulting catharsis from having the data house finally in order. Either way, people notice and want it for themselves. A key final task for any successful implementation is to ensure others see it and know about it. Schedule showcases within your organization. Help create case studies and lessons learned so that other may benefit from your wisdom and experience. You’ll quickly find yourself the reigning guru!


Back to Pro Tips

Back to Solution Spotlights