Get Rid of Irrelevant Social Media Monitoring Results
This article is getting shared all over the place and is a good start… Each tool has specific ways to limit junk data, and helps the analyst measure only important information.
I’d like to throw my hat in and push this a little further…
Analysts should recognize that spam is a signal and shouldn’t be ignored. If you are seeing spam, and it continues to increase there is a signal that you should work off of. The importance of the brand is actually growing… Oddly enough spam means the brand is experiencing success.
If you are running a scale report do a 2% random sample… yes you may end up reading 2,000 tweets 1,000 blog post starts and 500 Facebook posts but you will have a statistical sample to work with. After a few weeks it becomes a quarterly exercise. Use original users to run the 2% random sample and count the number of spam posts a user is posting. Now you have a reduction equation, after a few weeks you actually have another data point and trend line. If SEO is on your horizon, this is actually a win.
Another way to limit spam- Identify the worst offenders and exclude them from your data… if someone is posting 1,000s of spam posts cut them out (the boolean -“…” is accepted on almost all tools.) Source filters are another powerful tool that should be used by the analyst to remove problem source. Source filters can also limit scope of a report to only include paid/ gifted/ PR targets in reports.
This works for paid bloggers as well, however, if your organization is using paid bloggers there should be a list kept somewhere that can be focused very easily into a source filter.
Finally use the power of the word cloud… Almost every monitoring tool out there gives the analysts a word cloud tool. Do some research, is there a keyword that should not be in the cloud? If so you know you need to do some refining on the data baseline.
A solid topic profile is a living breathing thing, the analyst should be making small refinements everyday. The profile nearly breaths, expanding and contracting little by little. If this is done correctly and carefully, the data will retain consistency and validity. That said, I think every analyst should be fighting to get a longer view of the data. A multi-year view always adds unique prospective and information. Its important to note everyday the internet changes, so, historical data needs its own lens (I have now crossed to a different blog all together)
