Image credit: Microsoft Clip Art
Social media measurement can be difficult, even if you follow our handy Six Steps of Measurement System for Social Media. In my decades of experience with it, I’ve run across some typical problems. Here are five common pitfalls that you can easily avoid:
1. The bad search string: be careful what you search for
On Twitter no one speaks in full sentences, or for that matter in words. Which makes for some interesting search results if you’re a public company and searching for your stock ticker symbol.
When we were working for a pharma company, we found that Aztra Zenica’s AZN is also short for “Asian,” AbbVie’s ABV is short for “Already Been Vaped,” and Bristol Meyers Squibb’s BMS is short for “By Myself.” It took a month of cleaning and tweaking the system to get accurate data.
Your company’s name itself may be the problem. In fact, SAS Software has written the book on filtering out bad mentions of its brand. Among the ineligible abbreviations: Sexual Assault Support, Serial Attached SCSI, Society for Applied Spectroscopy, and Students Against Sweatshops.
Every company has similar issues. Whether it’s your chief spokesperson getting married, or your chief salesperson volunteering at the local soup kitchen, there are plenty of irrelevant mentions out there. The average monitoring tool will collect them, thanks to powerful collection technology.
>>The Fix: Test, test, and retest your search strings, at least once a month. You never know what the world will abbreviate next.
2. Bots
Bots—computer driven postings—are more prolific than ever. Worse still, it’s getting easier to create them. Just Google “make your own bot” and within minutes you too can create annoying fake Twitter handles and Facebook pages.
What this means for your measurement is that as much as 40% of what you’re pulling into your reports is essentially spam—or, at best, not representative of what influences your target audience.
>>The Fix: Take a few minutes each week to review the content that comes into your measurement dashboard and make sure it’s relevant and real. Compile a list of the social media accounts that are most influential within your market and focus on those, ignore the rest.
3. Bad numbers: However you count them, chances are your ‘impression numbers” are wrong
Unfortunately, one of the evil legacies social media has inherited from traditional media thinking is the continued focus on impressions. Sadly, many of the social media listening firms provide them as well.
The problem is that none of the measurement platforms seem to calculate impressions the same way. Some, correctly, add up the total number of first line followers/likes that the author of a post has on the day that it was posted. Others use multipliers and/or second and third-line followers.
Adding to this confusion is that we all know that not every follower is reading every Tweet or Facebook page. In fact, Facebook estimates that less than 3% of what is posted on the site is actually read.
>>The Fix: Avoid impressions all together and focus on engagement. If you have to show impressions, then always use the smallest number.
4. Irrelevant platforms
Too frequently, clients simply say they want “social” included in their measurement programs. For most vendors this means turning on a full stream of social data from every platform they can get their spiders onto.
But if your target audience isn’t on Pinterest or Twitter, why do you want to include that data in your report? Sure, it may inflate your numbers, but it’s not relevant to your business. Most likely it’s just cluttering up your results with more irrelevant data.
>>The Fix: Find out where your target audience is getting information and focus your measurement on those platforms. For example, in the B2B world, it’s a lot more likely to be LinkedIn than Pinterest. So talk to your media-buying folks, if you have them. If not, try to get some readership data from your agency. And if you don’t have an agency, talk to your sales department or do your own research.
5. Comparing apples to athletes
We get that there’s a lot of metrics out there and almost as many platforms to track them. So we totally understand why you may want to combine a bunch of data from five or six different reports (typically by five or six different agencies) into a single comprehensive report.
But data doesn’t work that way. If you have multiple agencies using multiple platforms, they’re probably all calculating engagement in different ways, so you can’t compare them. We recently were asked by a client to report on improvements in social media conversations from last year. Rather than reporting on “improvements” I discovered a 60% decline in the volume of posts. As it turned out, the client admitted that they had switched vendors and the prior vendor didn’t have any spam or porn filters in place. So the majority of the 2015 data was probably irrelevant. The solution was to only compare Q1 to Q2 using only the new vendor’s data. (No one likes a report with a footnote that says:”All last year’s data was discovered to be bogus.”)
>>The Fix: Get all your agencies, departments, and businesses in a room. Establish a standard and force them to stick to it if they want to continue to do business with you. Or only compare data over time from the same vendor. ∞
