Understanding your customers is a crucial tool in improving your products, as well as improving your marketing and customer service efforts. Big Data is an increasingly valuable currency in the world of business, with predictive analytics services being able to achieve some frighteningly powerful results – such as knowing your customer’s future behaviour better than they do themselves.
In an attempt to know your customers better, acquiring good quality data is just as important as the quantity of it. Here are some practical ways to implement your findings
Understanding customer preferences
Understanding your target audience through market research is vital before starting up a business, but don’t let it stop once you have got the ball rolling. Businesses should explore the data that understand the demographics of their audiences and which purchases they make. If there are no demographic-product correlations, there will be some categorisation that you can do somewhere.
Grouping customers can mean the messages you do send them are more targeted. This can increase the conversion and click-through rates, providing greater returns on marketing investments. In the instance of a beauty salon – it’s not enough to write down their email address and reach out to them with promotions. If you know their favourite services to buy, you can group them and send 5 different emails, so everyone receives an advert for their favourite service.
Finally, data can be collected in the form of customer feedback and is a brilliant qualitative way to understand them better – and better understand the ways you can improve as a business (and what you’re doing right). This feedback can be collected in real-time too, such as a one-question survey presented immediately after purchase.
Furthermore, all of the conversations with customer support can be analysed using NLP techniques to categorise the types of problems that customers are having which can highlight areas of improvement. Furthermore, you can track which types of problems successfully got resolved (thanks to the customer feedback following the exchange) and which did not. This could potentially point out the limitations of your chatbot too, better understanding which problems require being put through to a human and which do not.
Rewarding your best customers
Pareto’s principle would indicate that 80% of your income comes from 20% of your customers. Whilst this may be a gross generalisation dependent on the industry, it’s true that repeat customers are at the core of most businesses – from cafes and hairdressers to social media websites and hardware stores.
Putting more focus on data can help us identify who these people are. Perhaps a club card could track their spending or the web traffic more carefully. Once you can identify them, perhaps having them, you can target them with messages and promotions – rewarding their loyalty. Making this 20% (aka minority) of customers feel like VIPs is cost-effective, because they spend so much money with you, and ensure their business isn’t taken elsewhere.
If loyal customers do drift elsewhere, they can be targeted even more aggressively to win them back. This is common with emails to recently inactive users which are often titled “We’ve missed you” in the subject followed by a promotion.
Sentiment analysis is the quantification of language and emotion in order to measure sentiment. So, having a broad range of lexicon with a positivity or negativity rating attributed to it, which can then help determine if each Tweet or customer review is positive or negative (there are also more complex machine learning approaches to this).
The benefit here is that you can detect on a large scale what your customers and target demographic think. If you run a tennis sneaker brand, you may decide to analyse the tweets of people who follow Babolat, as these should surely be interested in tennis, and filter just the tweets that are mentioning sneakers. Now, you can assess the opinions that a select group of tennis fans has on sneakers. Perhaps most tweets containing any colour other than white (the traditional colour for tennis footwear) is negative.
You can do this with reviews on your own product, or a competitors product. It’s easy to scrape Amazon reviews and such and see what features of the product receive what kind of aggregated sentiment.
It’s one thing to understand customer preferences and sentiment, as already mentioned, but you can take this a step further with predictive analytics services. We can begin to predict future behaviour through Big Data and help this shape our marketing habits.
For example, if thousands of people on your website are getting halfway through the checkout process and then bouncing off the site, we can assess why – perhaps the checkout process is too long? So now we know this will also be the future behaviour of customers, we can make changes and see if we have caused a difference.
Perhaps a pop-up message can appear during the checkout process say “only 2 more clicks until your order is processed”. Perhaps we can advertise a “30-second checkout process” in email newsletters. Understanding how the customer is behaving can help identify flaws in our website, products, and marketing. Knowing when a customer is about to bounce off the site can be crucial information for some interventions.