Categorising buzz by topic is difficult

People talk about all sorts of things online in all manner of ways. The discussion around your brand will be no exception. It will be unstructured, even chaotic. To make sense of it you will need to organise the buzz into topics.

Organizing social media buzz into categories by topic is a very tricky thing to do. It is also something automated tools are notoriously bad at. See the section on the challenge of categorizing buzz by topic.

WaveMetrix uses custom classification schemes to analyse buzz by topic

At WaveMetrix all discussion is read, analysed and classified by human analysts. This is important because it ensures far greater accuracy than can be achieved with automated software. It also enables much more complex categorisations (of brand values, for instance), as well as consistency across languages.

Every WaveMetrix client has a bespoke classification scheme. This ensures that online buzz is analysed in the format which makes most sense to you. Locating data in the right categorisation scheme makes it more actionable and increases its value to our customers. We work with every client to develop coding schemes which work for them and which we can continue to improve over time.

All WaveMetrix client classification schemes share the following properties:

• 1. Customised. We work with each client to analyse project information in the best possible way for them. We can be very flexible in adjusting categorisation schemes, for example by adding new categories on the fly (something we call 'issue-based coding')
• 2. Logical. Although flexible, categorisation schemes are always logical. This means our categories have clearly defined limits and are mutually exclusive. It also means categories are rational and reflect the key drivers of the underlying business we are analysing
• 3. Hierarchical. WaveMetrix categories are always hierarchically arranged. If you have a category called “design”, then the sub-categories “size” and “colour” will fit underneath it. You will never see “size” classified under “value for money” or “price”
• 4. Multi-dimensional. Our categorisation schemes can always include multiple layers. For example, we typically code product features in one dimension (for instance, “design”, “quality”, “value for money” and so on). We also code other relevant information in other dimensions. These typically include brand values, purchase drivers, use cases and recommendations. Multi-dimensional coding maximises the insight gained from online discussion

Custom classification schemes power deep insight

Detailed, logical and multi-dimensional categorisation schemes transform the insight you can draw from online buzz.

By using human analysts to code posts in far more sophisticated ways than can be done automatically, WaveMetrix classification schemes extract truly actionable insights - such as who is recommending your brand and why.

Graphical illustration of multidimensional coding

Comparison of WaveMetrix vs. automated categorisation