Trend Spotting

Throughout, you’ll hear from Ben Ellis, Senior Research Consultant at Brandwatch. He helps organizations identify knowledge gaps and shows them how to use digital consumer intelligence solutions to fill them in order to answer big business problems. Essentially, he jokes, his role is to show the organization how to get going with the insights setup they need so they don’t need his help anymore.

Since joining Brandwatch, Ben has worked with some of our biggest clients, and he’s played a pivotal role in helping businesses spot unexpected consumer trends during the Covid-19 pandemic. He’s also done trend spotting as part of his work at Microsoft, BT, and We Are Social.

Drawing on Ben’s wealth of experience, here’s our advice for approaching trend spotting in 2020.

Approaching trend spotting

What’s the value?

The value of trend spotting may seem intuitively obvious – it helps you understand your consumers better and, ideally, discover opportunities before the competition. But the value can be hard to get to.

‘Trends’ disappear quickly, and even those that are most long-lived might not stick around for long. That makes it doubly important to check in on new and existing trends regularly – no easy feat for stretched teams.

The cadence of insight isn’t the only challenge in getting to the value. Going from insight to action is another hurdle.

“What is simple for a lot of companies is being able to spot the trends. What’s difficult is the ‘so what?’” says Ben. And it’s within this ‘so what?’ space that the success of your trend spotting endeavours is determined. The outcomes of the action you take as a result of trend spotting insights are the most important thing.

How often should teams do analysis like this?

To get to the value, you need the technical know-how, regular insights that reach the right people, and decisive, effective action. We’ll get to the technical know-how shortly – for now, let’s talk about how often trend spotting analysis must be done, and how it should be distributed.

“The stress and anxiety that comes with trend spotting is totally valid,” says Ben, who acknowledges the pressure on companies to be trend spotting 24/7 so as not to miss out on any opportunities.

That said, he does advocate for the importance of regular insights.

So, how does a team organize their trend spotting analysis and insight distribution? For Ben, it depends on the industry, team, and people who can action the insights. Looking back on a previous role, he says: “My cadence depended on the frequency of meetings where decisions can take place. It depends on how often the stakeholders meet, and by stakeholders I mean people who make decisions and make things happen.”

Don’t over-do trend spotting reports if nothing will come of them. Instead, schedule them to coincide with the ability of a team to take the insights and make real change as a result.

A four tiered approach

For general trend analysis, a wide net is needed to capture all potential opportunities.

Ben describes his four-tiered approach as so:

1. Industry or vertical
2. Brand or organization
3. Products or services
4. (All encompassing) consumers

“Being able to track trends for all four means that even if someone hasn’t even mentioned the brand, it’ll still be tracked as a new trend,” he says.

He gives the example of bleach and Covid-19, around which dangerous rumors spread. Many consumers and high-profile influencers have discussed bleach and how to use it to combat the virus, but they didn’t always mention a particular bleach brand. A company with a range of bleach products would miss out on the important trend without monitoring non-branded conversations relating to their sector. Especially for crisis moments like this, time is of the essence and being alerted early to new trends that could be dangerous is vital.

“If you can track all those four pillars, you can identify any trend that’s important to you. Those cover all of your blindspots. There’s virtually nothing outside of that that you’d be missing. That sounds very overconfident, but I’m happy to be proved wrong with that,” Ben says.

Approaching predictive trend spotting

“There’s always a risk with predictions,” says Ben. “The usual example I go to is life expectancy – even with all the smart people that work on it and the data science that goes in, there’s still a margin of error. But that doesn’t mean there is no value in life expectancy prediction.”

The nature of trends, particularly on social media, is fickle. This means that successful predictive trend spotting relies on identifying what’s going to be sticking around. Every trend spotted should be scrutinized – is there a lot of spam in the conversation? Who is it being driven by?

Ben recommends using a range of solutions to get the strongest insights. “There is more to a trend spotting machine than a single tool,” he says.

Using historical data can also help with this kind of trend spotting (we’ll look at this in our ‘in practice’ how-to guides below). For example, what patterns in one dataset preceded a significant milestone in another? Perhaps it’s a positive review from an influencer coinciding with a boost in sales for a particular color of eyeliner, or an increase in searches around a particular topic that preceded an influx of inquiries on your website. If you’re able to replicate these results and start to establish cause and effect, you can look to the future and know what signs to look out for.

In practice: How-to guides for common use cases

How to spot trends from the known, and those that come out of the blue

“There are two types of trends you can expect,” says Ben. “Trends that come from things you know about, and trends from places or topics or themes you’re not aware of.”

With Ben’s four-tiered approach outlined above, there’s no reason not to pick up on key trends, regardless of where they come from. Here, we’ll get a little more granular and show you how you can do it yourself with Brandwatch Consumer Research.

How to spot trends among known topics/themes

Spotting trends around your own brand is easy with the right set up.

You could examine a broad conversation with the topic cloud component in Brandwatch Consumer Research, which will surface key phrases, keywords, hashtags, emojis, people, or places that are trending.

You can even do some trend spotting with a simple line chart. Here’s an example, looking at mentions of ‘Twinkies’ across social media between January 1 and August 11 2020.

Our AI-assistant Iris is great at picking out short-term trends. Here, it has picked up several spikes in conversation and the distinct driving forces behind those spikes.

In case A, there are two tweets driving the conversation. They both relate to wearing masks due to the pandemic while buying Twinkies (seen here as a trivial thing to do).

This is the kind of boost in conversation that could easily be picked up in real time (not a month later) by a Custom Alert.

For example, you could set one up so that you and any other key stakeholders can be notified any time there’s a significant increase in conversation around your brand name.

You may also want to set one up to pick up individual mentions that relate to key topics. For example, every organization should be keeping tabs on how people are talking about them in relation to the pandemic. Setting up an Alert to notify you any time someone mentions your brand name near or alongside words relating to Covid-19 is simple – you just need to add relevant keywords, as shown below.

Custom Alerts can be made much more granular than this. For example, you could be alerted when there’s a mention from a verified account with more than 10,000 followers that mentions the taste of one of your products.

The key thing about Alerts is that you’re predefining what it is you want to be alerted about – mentions must fall into the criteria that you’ve set, which means you already know what you’re looking for.

How to spot trends coming from ‘the unknown’

Picking up on trends that relate to your own brand name can be pretty simple, as shown above. A little more complicated is spotting trends that come out of the blue. While these are harder to find, Ben says that both trends ‘from the known’ and trends ‘from the unknown’ are equally valuable.

“A good example I have for that is homeopathic or natural remedies people are coming up with to make their own hand sanitisers,” says Ben. “Maybe they’ve created a new solution with a new name that didn’t exist before that day. Unless you were monitoring conversations broadly, not just about your own or known competitor brands, you wouldn’t pick it up.”

So picking up unknown trends starts with a broad query that looks at your sector and your product types. Then you can do all the same things we mentioned above, like using topic clouds, line charts, and Alerts. To go one step further, you can use Signals to pick up on trends that you couldn’t have expected to pop up.

These, like Alerts, will email relevant stakeholders about key incidents within the data. The difference is, Custom Alerts search for things you might already be aware of while Signals pick up on changes in the data you might never have been able to plan for or predict. The technology analyzes your data in real time, sending an email alert to your specified recipients as soon as it detects significant or sudden changes in your data, such as trending stories and authors, or shifts in sentiment.

Setting up a Signal is really easy.

Step 1: To create a new Signal, you simply need a query (a search that picks up mentions on a topic, brand, or product). You can also add filters, but in this case you’d want to keep things broad – you’re looking for unknown trends after all!

Here we’ve used the example of a query that looks at general conversation around antibacterials and sanitizers, so any sudden change in the conversation can be picked up even if it’s nothing to do with a known-brand hand sanitizer.

Step 2: The other thing you’ll need is a list of recipients to send the Signal to.

How to spot trends from across the world

When thinking about known and unknown places for trends to originate, it’s important to think about physical spaces as well as social groups or online communities.

Brandwatch has access to over 105 million sources (Instagram and Twitter count as just two) across 85+ languages. This means we can track trends emerging all over the world. It might not sound important to do so if your business only operates in the US, but previous research has found that some regions are more influential than others in starting trends.

When searching for trends broadly, make sure to cast the net as wide as possible. In Brandwatch, that means turning off location filters and writing queries in multiple languages (something our validators are able to assist with, if needed).

How to spot trends among key consumer groups

Looking for trends among a general audience is one way to trend spot, but what if you want to look at trending topics among a particular consumer group?

Social Panels in Consumer Research make it easy.

For example, if your target audience is UK-based accountants, you might perform a search within Social Panels to filter authors by those that have ‘accountant’ or ‘accounting’ within their bio, and they must both be an individual account (as opposed to an organization) and be located in the UK.

Brandwatch immediately returns a list of over 10,000 individuals who fit the profile.

With this panel set up, we can begin to look at what’s trending within conversations UK accountants are having online. This is great for content strategy, but also to spot general trends in what your audience cares about and are getting fired up by. You can do this analysis over years, months, weeks, days, hours, or minutes, depending on what you’re looking for.

For example, here’s what UK accountants were posting about in a period of two weeks in August. Brandwatch Consumer Research’s many visualization options can help you find not only what people are talking about now, but also what’s starting to fade as a topic and what’s a more stable trend. Everything is clickable, so you can easily understand what’s behind each mini trend.

Once again, you can set up Alerts or Signals to pick up on important changes coming from the data (in this case, posts from UK accountants) and notify you even while you’re not logged into the platform.

How to use historical data for predictive trend spotting

Earlier we mentioned the importance of historical data in predictive trend spotting. With Brandwatch Consumer Research, you can look at up to ten years of historical data on any brand, product, or topic.

By pairing this with other datasets, you can establish patterns and potential cause and effect relationships. With these identified, you know what to look for when trying to predict outcomes in the future.

Here’s a really simple example. We looked at conversation around sun protection products alongside sun hours in the UK between January 2019 and July 2020, finding a pattern between the amount of sun people experienced and their interest in sun protection.

Armed with this information, sun protection brands can use weather forecasting to plan future social activations. They can also delve into the topics people are talking about within sun protection discussion to find out what those activations should focus on and even pass this information on to product teams for further development of suncare products. This year we found lots of consumers talking about cruelty-free products and hydrating minerals.