May 17, 2023

5 Things to know if you're getting started in social intelligence

Date & Time (GMT):
May 17, 2023 12:47 PM
Date & Time (EST):
May 17, 2023 12:47 PM

Since starting out in social intelligence a few years ago, I’ve fallen in love with social data. Where else can you access people talking spontaneously, in real time, about the things that really matter to them?

However, there are a few things I’ve learnt along the way that are worth keeping in mind if you’re thinking of getting started in social intelligence. Read on for a run-down of the different data sources to think about (it’s not just social media!), the importance of writing a good query, and why you can’t trust machine analysis.

1. It's not just social media

Starting out, it’s easy to assume that social listening is just that: listening to social media conversations. And yes – obviously, Twitter and other social platforms play a big part.

But actually, there’s a wealth of other online data that you can gather manually or by using various digital listening tools. My team pull in insights from online forums, Reddit, product reviews, blogs or search engine data just as often as Twitter data.

This is important because it opens up a whole new host of topics, and audiences, that you can explore. Whilst people might not be talking about dentures, pensions or paediatric allergies on their social media accounts, there are hundreds of forum conversations about how people navigate these topics, and a lot you can learn from search data or product reviews about the functional questions people consider. In fact, forum conversations are often home to some of the rawest emotional insights.

Don’t limit your insights by focusing only on the big social platforms.

2. It's still evolving

Social intelligence is changing all the time. In the last few years:

- Clubhouse has launched
- TikTok has exploded
- Platforms have released new features (e.g. Instagram Reels and IGTV)
- Regulatory changes have impacted how data can be used (e.g. Mumsnet and Instagram have both reduced access to their data)
- Changes to the visibility of ‘likes’ have affected how we report on metrics.

To ensure you’re making the best use of social data, you need to stay on top of how the landscape is changing. For example, recently I’ve been thinking about new analysis frameworks for video and audio data, and looking for ways to gather and use data from more sources such as Pinterest or TikTok.

Make the effort to stay up to date with industry news, keep an eye out for new tech solutions, and explore different ways of gathering and analysing new forms of data.

3. Your data is only as good as your query

If you do nothing else, it’s worth learning how, and taking the time, to set up your social listening queries properly.

At a basic level, a social listening query will consist of keywords and other requirements (e.g. language, location and data source) that your listening tool will use to find relevant online conversations.

A poor query can mean the data you get back will be full of irrelevant posts or might be missing key parts of the conversation.

When setting up a new query, prepare: use search data, desk research, take a look at Twitter or even open a thesaurus to identify the most relevant language to be using. Do a test run of your query, and exclude terms you notice that are muddying your data.

And if conducting research in a local market, work with someone who knows the language and market context, to ensure that your search terms are localised and relevant.

How to create a boolean search query

4. Take automated analysis with a pinch of salt

Many social listening platforms have built-in tools that will analyse posts and tell you whether they are broadly positive, neutral or negative in sentiment. However, the accuracy of these tools is still no match for human analysis.

Here are two social media posts from last summer, around the time of the Black Lives Matter protests when there were calls to defund the police:

Tweet shows how sentiment can be mis-analysed by social listening tools
Tweet showing how automated analysis can cause problems

Pretty negative, right? But the machine sentiment analysis of a leading social listening tool classified both of these posts as “Positive” in sentiment.

That level of inaccuracy across the whole of your dataset can give you a really misleading view of how people feel about a topic - whether that’s a social issue, a campaign, a product or brand.

To really understand what’s going on in your data, I’m a firm believer in qualitative, human analysis. A human lens allows you to navigate and understand context, complexity and ambiguity in a way that machines can’t.

Through qualitative conversation analysis you can get an accurate, in-depth sense of how people feel and what really matters to them. And going a step further and coding up your data can give you a much more accurate quantitative read of key themes or sentiment than you’ll get from machine text analysis.

5. Think about the ethics

Working in social intelligence brought home to me how public much of the data we create online actually is.

Your 3am existential Google searches, angry Twitter posts to customer services, or private moments with friends can all end up as a data point to be analysed.

But most people posting on social media or online forums are not expecting these posts to end up in a research report, so we need to be sensitive to this context.

As a general rule of thumb I would:

- Avoid reporting usernames alongside posts (unless it’s a public figure or brand)
- Remove all personally identifiable information from posts
- Blur out faces if using an image shared online

Anything that has the opportunity to embarrass or cause harm to someone if traced back is a big no. And there is also a need to ensure that reporting is in line with the regulations of the individual platforms and data sources.

As new contexts emerge and the field grows, ethics is an area that’s worth reconsidering regularly.

Social intelligence offers unrivalled opportunities for understanding people – particularly when you take the time to go beyond the ‘obvious’: the big social platforms, and automated dashboards.

Hannah is a Research Manager at Listen + Learn Research. If you’re just starting out in social intelligence, then she’d love to hear about your journey!

Or view the interview on LinkedIn

This interview was recorded via LinkedIn Live, if you prefer to view on LinkedIn, click the button below.

View Interview

See related content

Webinar & Panel
Writing a Winning Nomination for the SI Insider 50
Interview
Social Intelligence Insider 50 Launch
Webinar & Panel
SI Insider 50 Reveal Party