If a newsroom wants to take impact measurement seriously, it must commit resources to it. In addition to full-time Impact Editor roles, newsrooms must take care to catalog their own content in a system of subject matter tags that make sense. An analyst can only compare, for example, the success of the latest energy multimedia package against past feature projects if those articles are properly tagged in the system.
Because newsrooms often publish too much content to do this cataloging and tag standardization post-facto, editors must employ discipline in tagging stories at the time of publishing. Without such standards, analysts waste a vast amount of time and the newsroom loses vital internal information about its own operation. Apart from the workflow aspect, a newsroom that doesn’t view tagging as a key part of its daily operation inevitably lacks insight into the long-term operation of its own coverage.
One feature we weren’t able to include for this version of NewsLynx was a “Train Your Lynx” section that would let users train a classifier to potentially auto-label articles for them. While some of this tagging problem might be alleviated through better technology like this, it can’t solve the whole puzzle. In other words, even if a computer could find good groupings among published articles, institutionalizing those computer-determined buckets is the tail wagging the dog. Similar to setting goals, understanding content buckets should be a human-made decision, not one outsourced to a black box.