Simon McEvoy, Planning Director at Tangent Snowball, explains how to ensure you’re capturing the right data in the right way
A few months ago, we at Tangent Snowball were commissioned to solve a difficult business challenge by a leading online retailer. It had a large purchasing data set and had built some interesting but quite traditional segmentation models using RFM scoring (recency, frequency, monetary).
However, the retailer had no idea how this matched up with its behavioural data set, an enormous set of files detailing every email open, click, web visit and page view which everyone in its contact database had made over a period of years. The brief to us was this: ‘Is there an overlap between our most active fans and advocates and our most valuable customers?’
This is not an uncommon question and given the proliferation of online data at marketers’ disposal, many businesses are trying to establish the same thing. In part, retailers want to know if their marketing budget is being spent wisely. Do all those clicks, views and shares result in a higher-spending customer? But also, if there are customers who are very active but low in spend, can we leverage additional value from them in ‘amplified loyalty’ – spreading the word through their social networks and friendship groups?
This insight could just as easily come from external data sources as from your own stored data. Platforms like Facebook, Twitter and LinkedIn are giving us unprecedented access to billions of bits of consumer data. And by linking social media profiles to our customer database we can gain valuable insights, such as which customers are key influencers, which have large followings and what other brands they like. All of this insight can feed into and enrich the segmentation model.
For this client, our approach was to model the engagement data into a matrix, placing every customer into one of nearly 700 boxes depending on email and web engagement behaviour. We weighted behaviour depending on its value to the business, so an email click was worth more than an open, but not as much as viewing a product page.
Finally, we rationalised this into a 25 segment model to make it more ‘useable’ and less granular. This could then be overlaid with the existing purchasing data model to reveal the insights the client was looking for.
The results were fascinating. There were large groups of people that rarely (or never) shopped but religiously opened emails, forwarded them to friends and browsed online. There were also customers that spent large amounts but hadn’t opened a single email and browsed only to purchase. There were also reassuring correlations between engagement activity and purchasing behaviour, as you would expect, but the question remained – how do you treat those that fall outside the expectation?
High value/low engagement customers are reasonably simple as they are clearly happy with very little contact, coming only at crucial times. Propensity modeling to determine when they are likely to want to shop could help to fix purchasing habits, as would targeting them only with high-value offers and promotions to increase basket size. Likewise, these high-value customers might benefit from a more bespoke service, such as personal shopper advice or even telephone contact.
However the more troubling group are those that don’t shop but remain fervently engaged. Every brand has some of these – aspirants, fans, call them what you will.These people love the look, feel and image of your brand. They love browsing products and creating wish-lists. They’re just never likely to buy. So how do we facilitate and encourage this behaviour, how do we track it and how do we place a value on it?
Tools for trade
Facilitation and encouragement means giving these fans and advocates the tools to share and promote whenever they want to. Enable emails and product pages with sharing buttons and social comments. Allow users to log-in to your site using Facebook and link their wall posts to their activity. Create sharable wish-lists and look-books. Furthermore, consider incentivising this activity to attract more advocates. Perhaps sharing could become part of your loyalty or rewards scheme.
Tracking this behaviour is far more complex and involves using the right software tools. There are some free tools to measure general amplification of messages, such as TweetReach, but to link this back to individual profiles you will need to pay for it.
Platforms such as HootSuite Pro, Brandwatch and Adobe CRM all have facilities to measure activity against individuals in the social space, but all have strengths and weaknesses so it’s important to have a clear idea of exactly what you want to track. Remember, to be able to measure the ROI you will have to be able to accurately track the actions you’ve incentivised, so getting this right is really important. Which leads to the last question – what value can we assign to this activity?
Creating an ROI for social sharing is a separate article (or book!) in itself and not something that can be addressed in full here. However there are some ideas that can help in developing an attribution model for determining sharing value.
You might want to start by looking at the price of purchasing promoted advertising through social platforms – Cost Per Click on Facebook or Promoted Tweets, for example. This can give you some idea of how much you could spend on encouraging existing customers to share on your behalf. By measuring the effectiveness (in terms of clicks, shares, purchases) of promoted messages versus earned messages, you can establish what kind of budget should be assigned to encouraging sharing.
Another approach might be to compare the cost of acquiring customer data or making a sale through traditional channels with the cost of doing it through social platforms. This is an easy way of comparing social sharing with other, more established and accepted numbers within the business. But bear in mind social networks are a notoriously tricky place to make a direct sale from, but are more suited to capturing data using games or competitions.
Whichever approach you adopt, bear in mind what you’re looking to achieve and make sure you’re capturing the right data in the right way. That way you’ll be able to prove there is quantifiable value in customers who never buy a single item in your product range, as long as they’re supporting you in other ways. That’s what your bean-counters will want to know.