Why most social media monitoring tools are inaccurate
Social media monitoring tools can provide you with a large quantity of information, but they are still missing the human touch.
Here in Arekibo Digital we do social monitoring, for ourselves, our clients, our projects, and our industry. Solid and reliable information gives any company an edge in their market, and just as the news media is using social networks to source content and opinion we use the social web to gain a better understanding of our users’ interaction with our work.
But we don’t rely on all-in-one social media monitoring solutions, we use a selection of tools, the data from which we analyse with information from other sources; such as web analytics, click analytics, news media, blogs, and most importantly our in-house search engine optimisation team, copywriters, and social experts.
We didn’t develop this system lightly, we have used the big players in the all-in-one social media monitoring field and sadly enough, we haven’t come across any solution that actually works as accurately or reliably as we and our clients require.
The reasons for this are simple, it’s all about quality and context rather than quantity. When it comes to social monitoring the type of deduction needed simply cannot be automated without an advanced form of artificial intelligence, just as social media monitoring, much like social media, cannot be automated without sacrificing quality.
Social Media Monitoring
If you are using or are considering using social media monitoring applications then there are three things that you need to do before you can properly understand your data. You need to be able to assess the quality of the content you are analysing, understand how sentiment analysis is an inexact science, and realise how influence stands for very little online.
Content Quality: In April this year Google updated its search algorithm; this isn’t remarkable, Google does this all the time, but what was remarkable was why Google issued this update. The search company wanted to remove what it saw as low quality content from its index such as content farms and duplicate (plagiarised) pages.
But for argument’s sake let’s say that Google’s change has resulted in some quality content about your brand being pushed lower in the index. By result this content will be harder to find by the user and therefore have less of an impact on your brand perception (be it positive or negative). However, from our experience, all-in-one social media monitoring tools have not yet been able to bring this Google level of quality to their search results. We have found searching on all-in-one social media monitoring tools returns all results their own search engines have been able to index, regardless of the quality of the content or how likely or unlikely it would be to find on the world’s largest search engine.
We cannot see how a quality judgment of your brand’s placement online can be made if signal cannot be sufficiently separated from noise.
This debate over quality and quantity is not new, indeed we’ve discussed it before on this blog, but accessing quality results is vital to streamlining your social media monitoring activities.
Sentiment Analysis: If you are not aware of sentiment analysis (or opinion mining) then you will be soon. The theory goes that it is possible to analyse how people feel about a product or service (like, dislike, neutral) from the content that they post online. Sentiment Analysis has become a growing concern for businesses but it is far from an exact science, last year researchers in the United States made some noise when they said that Twitter was 87% accurate in predicting the movements of the stock market. But this data has, as one would expect, been controversial.
It seems to us that the way social monitoring / analytics suites extract this information however is incorrect. Systems like this often have a simple formula, “Negative/Positive keyword + brand = Alert”. So for example the following tweets would provide an alert to your social monitoring suite.
- “I love this new product of Arekibo”
- “I’m not very happy with this new product of Arekibo”
- “This new product of Arekibo is kick ass! Try it out”
Many of the all-in-one social monitoring suites will interpret the first two tweets as having a positive tone (Love, Very Happy) and the last one as a negative tweet (kick ass!). However there is only one negative tweet.
We are not saying that sentiment analysis accounts for nothing. It can be valuable but without human involvement the value of sentiment analysis is questionable.
Influence: One of the main barriers to for companies entering the social web is ROI. How can you measure the worth of your social media activities against the time and financial investment required to maintain such accounts. Effectively the questions comes down to, what is the value of a like, what is the worth of a retweet, do more followers mean greater returns?
The difficulty comes with applying quantitative measurement methods to qualitative processes, as it has done with advertising for decades.
Again this debate comes down to the argument over quality vs. quantity. On Twitter, for example, having a greater number of followers can look impressive to all-in-one systems but there are various means of inflating Twitter and Facebook audience counts.
It is only by reviewing the accounts that you can see how influential your, or another, profile is.
All-in-one social media monitoring applications have their place, and will improve with time, but regardless of the means that you use to monitor social data it must be remembered that numbers and statistics without qualitative analysis will never tell you the full story. Above all remember that social media monitoring is a qualitative process which requires human involvement to be of worth to you and your company.
Related posts:
This entry was posted in Social Media and tagged analytics, content, copywriting, data analysis, facebook, google, google panda, influence, quality, Search Engine Optimisation, sentiment, SEO, social media monitoring, twitter. Bookmark the permalink.
very nice thnak you.
_____________________________
http://www.biltekakademi.com
Antonie
We think alike!
I love this debate on quality vs quantity. For social media, the answer is probably in between. If you outreach to a lot of people without doing the proper due diligence, they'll kick your ass ;-).
Where I work, we measure influence by "enough of the right people that trust you". i.e: enough for quality and 'right people that trust you' for quantity. For example we'll breakdown the social web in network of tribes which are networks of influencers and we'll evaluate the influence on a tribe per tribe basis; i.e: the people who are experts in personal finance decide who are influencers in personal finance.
As for sentiment analysis, too many parameters are loose for it to be automated.
Indeed the whole social media stuff can't be automated. Someone needs to be on top of it so that the output make sense and can lead to some actions.
Laurent
Good article. Sentiment analysis is really the key feature that can drive productive decision making. I agree with sdolukhanov below:
Social data + business intelligence = IMPACT
Monitoring alone isn't really worth anything. It's about results, nothing else.
brettrelander
The end justifies the means; great call Brett.
sdolukhanov Thanks for the comment ! It's much appreciated.
Antonie
The problem with social media monitoring software is that it doesn't show how social data actually impacts key business performance indicators. It's great for hearing mentions of your brand, but when it comes to actually quantifying your companies investment in to social media activities, trouble ensues.
Social data + business intelligence = IMPACT
Social media business intelligence is emerging as the glue to finally connect businesses with their social investments.
Antonie, thanks for the post! It's nice to hear others are skeptical about the practicality of automated monitoring software, and is taking a stance to combat the irrelevance of social monitoring software for enterprise users.
- Sergei
sdolukhanov
thanks.
www.bilgiegitim.com
- spam
- offensive
- disagree
- off topic
Like