Synthesio’s Trend Detection & Insights Series, Part 4 (see parts 1, 2 & 3): What’s the math behind something going viral? Turns out it isn’t as simple as you think. Not all topic peaks are created equal. If your brand is like most, you probably have some way of monitoring online conversations that could be relevant to your brand. Most brands do this through a Social Media Intelligence program, or what used to be known as Social Listening.

To get an idea of the difference between Social Listening and Social Media Intelligence, check out this blog post!

So you’ve probably got a bunch of keywords and topics that you’re monitoring, and a platform that aggregates mentions that include those keywords. Most Social Media Intelligence platforms can show you how the volumes of mentions that include those keywords have changed over time in a visualization like a mentions timeline that looks something like this:

Just looking at the mentions timeline above, finding a peak in mentions may seem trivial. Why not just circle them and start an investigation into why mentions for that topic spiked?

Because, unless you’re a statistician, that probably won’t get you anywhere.

What makes a peak significant?

If you’re defining a peak of mentions as any time there is a spike in mentions, chances are, most of those spikes are well within the norm of conversation volume and aren’t indicative of a true social media trend. In other words, most of those peaks that you’ve been spending time understanding are probably not statistically significant and more often than not, aren’t indicative of something that is important to, or affecting, your brand. They’re not anywhere close to going viral. Those peaks aren’t picking up the traffic you want.

How can you tell if something is heading in that direction? What makes a peak significant? Simply speaking, peaks are either spikes (or dips) in conversation volume that differ greatly (significantly) from the normal conversation volume for the topic in question. Spikes and dips usually require additional investigation.

Identifying significant peaks requires conducting statistical analysis.

Again, like most marketers, you’re not a statistician. Luckily, you don’t have to be.

Today, thanks to advances in data science and machine learning technology, Social Media Intelligence programs can and should be able to detect numerous social media trends, including peaks and viral topics. That means your Social Media Intelligence platform can tell you which peaks are significant, in a statistical sense, using an algorithm that is tailored to the conversation volume for the topic in question.


On the previous chart, all bars exceeding the upper standard deviation line could be technically considered as peaks, as they represent higher values than expected. However, just a few of them are true peaks. Statistically speaking, a peak is the highest value reached when a line goes from upwards to downwards. So, when designing algorithms capable to spot such peaks, we must take additional computations into consideration – to avoid catching just any high value as a peak.

Your Social Media Intelligence platform should conduct this type of analysis on all of your data, across all of your topics, to identify social media trends that are truly significant and require your attention automatically — in real-time.


Identifying peaks is not enough.

Beyond alerting you about the existence of a social media trend such as a peak, your Social Media Intelligence provider should be able to explain why, where, and how the peak took place. What made something gain enough traffic to go viral? Just like the detection of the peak itself, the generation of an in-depth report for each social media trend should be generated automatically.


Why being aware of significant peaks is important for your brand.

Marketing professionals always need to stay on top of consumer perception of their brand, products, competitors, and industry at large.

Among other things, identifying a peak can inform you of a:


Logging into your Social Media Intelligence platform one day, you’re alerted to a significant peak of mentions surrounding a product your brand just launched. Digging into the report that your platform generated about the Peak, you find that most conversations are negative in sentiment, and many of them mention a defect in the product.

It’s not a situation any brand wants to be in, but it happens. Now you can know as soon as it does, and take the right steps to get ahead of the problem.

Campaign’s performance.

Planning a campaign takes a lot of time and effort and it’s usually not until the campaign has concluded that a brand can evaluate the ROI it brought in. That means that if a campaign is under-performing, you wouldn’t know until it’s over. In other words, until it’s too late to make adjustments.

Not anymore! Now, once you’ve launched a campaign you can look out for spikes in mentions surrounding it to figure out if your audience is engaged, and if not, adjust your strategy so your efforts don’t go to waste.

Competitor’s strategy.

You’ve been alerted to a peak in mentions surrounding your competitor. Why is their content going viral so quicky? Consulting the report generated for the peak, you see that the mentions are surrounding an event your competitor is attending or campaign they’ve launched. You’ve now got a leg up on the competition and can form a strategy before it’s too late.

The most important websites.

Your Social Media Intelligence platform has alerted you to 5 peaks over the last week. Where did most of them take place? Using automated peak detection can help you figure out which websites your audience is most active and engaged on. Now you can plan content strategies that will drive real results.


Are you ready to take your brand’s Social Media Intelligence to the next level with state-of-the-art trend detection? Learn more about peak detection and other signal types.