Navigating Big Data

Tesco famously has ‘segments of one’. Which is lovely of course - but they had to buy a data company just to make sense of the data so they could get there. Most of us don't have that luxury. But it doesn't mean we can or should ignore data, even if it looks like it might become unwieldy.

Some brands haven't yet realised that the power in a brand/customer relationship has shifted from the marketer to the marketee. Clearly however social media and the ability to share every thought, spoken or unspoken, with friends and peers and even the whole wide world means that the brand perception is out in the wild. It's been let loose. No longer is the way your brand is represented in your control. It's in the expressions of passion, ire, indifference and ephemerality of the digital ecosystem: Facebook, Pinterest, Snapchat, Twitter, Vine, even email. It's transmitted by mobile, stored on the web, and available to the world.

Your job as a marketer is to understand that this revolution has already happened. And to take advantage of it. If you can do it successfully you can catch up with the wild thing your brand has become, and even gain competitive advantage while your peers wrestle with boards who just don't get that they're no longer in control.

Scary thought?

So what do you need to do in order to flip the situation around? Well, part of the problem is the notion that we can regain control. I don't think we can. What we can do however is map how consumers behave, and indeed how their attitudes will shape how they behave in the future. By going down this route rather than trying to gather the brand in, you can extend the brand into the customer's territory, give them more control by enabling free interpretation of the brand's essence. And that takes not only courage, but data too.

Customer insight is the product of data. The three dimensions of segmentation (what we call 3D Segmentation) are:

• Demographic - who the customer is;
• Behavioural - what they do and have done;
• Motivation - why they do it. 

Demography is slow moving, so we use it as a kind of snapshot to describe people. It means we can target them accurately. Behaviour is retrospective, but we can observe behaviours and trends and make extrapolations based on probability and this gives us propensity models. This means we can target them efficiently. The final dimension is about motivations, attitudes and 'need states'. Sports brand ASICS leverages this in its MyASICS loyalty programme: by understanding why a runner runs, we can talk to them in terms that resonate… the desire to be fitter, or to win, or to raise money for a cause. By talking to its customers about those things that address their motivation, ASICS creates extreme loyalty, increasing sales. Worldwide. And MyASICS is served by a website, and emails, and mobile. All of which feed back data so we can hone the programme.

These days the various digital channels are so well established that the mechanisms that allow you to track a customer in their journey in one can easily be joined with the mechanism in all the others. It means we can effectively create a joined-up process to track a customer across all digital channels as they weave about their daily lives. This ability extends even to the real world - we work with clients who have incorporated data from electronic point of sale (EPoS) systems into their customer view, so we can attribute till sales to pay per click (PPC) campaigns and journeys via every imaginable digital touch-point.

And it's not that difficult, and you don't need to buy a DunnHumby or a data team to do it. The concept of rapid prototyping has been very successfully applied to creating online customer labs and pilot programmes. For instance, brands like Bupa have used it incredibly effectively to build online communities at very low cost before making decisions about major investment (my agency, Underwired, created Bupa's Carewell using this rapid prototyping approach – saving the client around £150,000).

Forget the Single Customer View and its squillions in Capital Expenditure; rope together several separate systems based only on those components you actually require to do the job of proving return on investment (ROI) and use it to monitor customer behaviour in response to the insights you generate from simple data analysis. In my experience six or seven segments gets the job done - segments of one are for when you're already at the outer extremes of wringing profit from data and not when you're mid-shift towards putting your customers at the centre of the brand universe.

Big Data: Why more data is better for brand loyalty and customer experience

We've recently started talking to a brand which has around 700,000 customers in its database. They have collected lots of behavioural data, by which I mean transactional data - recency, frequency and value (or RFM) - and response data. This response data is all about what happens when the customer is sent a piece of communication, in this case an email. What they do, when they do it, where it leads. Say the database contains 30 fields. That's 21 *million* pieces of information, all tied together to create a big fuzzy room we can in effect walk around, try to make sense of, and manipulate to achieve commercial goals. 21,000,000.

Everyone talks about Big Data as if it were some kind of technological nirvana. The reality is you can gather data from a whole lot of sources and stick it all together more or less by hand, if you need to. In practise, Big Data is shorthand for the notion that if only you could mine, interpret and extrapolate all the data you could get you'd have some kind of joined up living solution to customer engagement, almost a mindmeld between your brand and a collective representation of your customer base in its entirety. Nice.

The reality is that data is an enabler, something you can make use of - not something that should make your decisions for you.

So how does this pragmatic approach work? There are a number of critical steps to take you  from having on the one hand a commercial goal and on the other some customer data. First, make sense of the data. Customer insights start with understanding what kind of data you have. In our CRM terms this information breaks down into three broad groups:

Demographic - who the customer is
- Gender, age, life stage
- Location
- Income
- Status
- Family make-up
- Education etc.

Behavioural - what they do
- What they have bought
- When
- In response to what
- How much do they spend
- How long is their 'customer lifetime'
- What channels do they use
- When do they respond most

You can see already that by combining some of this information you can infer quite a lot about the way you might want to talk to some of your customers. It is obvious that you can start to create segmentation based on demographic and behavioural data. However, this approach to segmentation may help you to be efficient (behavioural) and accurate (demographic) in who you talk to, but it often does not tell you what to talk to them about.

Taking the classic example of customers of a prize-based fantasy football league, segmenting by these two dimensions might lead you an easy segmentation based on whether the customer buys one or twenty teams (behavioural) and jump to conclusions about their financial status (demographic).

3D segmentation adds a new aspect, motivation, to the mix. If you can divine what motivates your customers then you can speak to them using motivation-based segmentation and that may actually provide the cut through that's required in a highly competitive environment.

Motivation - why they do it
- Need state
- Environmental factors

This dimension can change based on changes in the other two dimensions; for example changes in family make-up or life stage may radically alter someone's drivers for engaging with your brand.

In the case of the fantasy football league, by looking not at behaviour or demographics (which didn't appear to correlate) but by motivation, through the simple expedient of a brainstorm with everyone we could find near the meeting room we reached an insight we could test - first by checking the correlation with the behavioural data, second by sending a brief questionnaire to a standard sample. The insight was that customers bought principally because they were either motivated by passion for the game (bought a single team) or by the desire to win the prize pot (bought twenty teams).

By using this simple insight we created two segments serving two types of (relevant) content. These were then split into time-based sets based on where the customer was in the product lifecycle (new joiners, mid-season etc.) so we had six or seven simple segments.

Revenue went up 93% in 90 days. The client was The Sun.

The job of data is not to confuse or confound. The job of data is to allow you to extract simple insights that allow you to run singleminded campaigns that tap into your customers' motivations so that they want to engage with you. As we start to think beyond the age of CRM and focus on rapid growth, it is imperative that Big Data doesn't become an encumbrance. Data should be there to provide insight so you can get on with the engagement - because how you engage with your customers is the only thing that will drive your success.

Customer engagement for bars (or, teaching granny to suck eggs)

In the olden days (pre-1995), when customers knew the name of the person serving them and vice versa, life was good. Then along came the internet. People stopped going out for a drink to socialise, catch up or find a friend. We were told then that the power had to transfer to the brand, or at least the vendor. Marketing became about how you presented your brand and how you attracted people in. All the while customers were losing focus on what was attractive and turning their attention to what met their needs best… and online met their needs pretty perfectly.

So what are the best decisions that need to be made to address this change? First ask yourself: can we compete without going online? If your answer is ‘no’ then you need to assess how best to target your customers and by what means.

Social media, used largely (at least in my own life) to work out where to go for a celebration, catch-up or a noisy bit of fun. In other words, the perfect media platform to reach customers to start / maintain relationships. It’s pretty simple to make work in fact: get to know your customers by observing their behaviour (on your website, in response to your emails and/or tweets) then make sure you use those channels to say things that they want to know or hear.

However, social media isn’t the only option available. Email – today’s postcard – cost pennies to generate and send, even in relative bulk. For this kind of approach of course you need data. You could gather this at the point of sale and add it to a centralised database – which would be as simple as an Excel spreadsheet. By offering some kind of value exchange when you visit the website - perhaps a free drink next time you visit - you will be able to further your data capture which will allow you to create targeted, timely and relevant campaigns to drive sales and support your ongoing relationships.

Where once in the golden days a bartender would know each customer by sight, today, bar owners can know the customer through digital tracking. The internet has enabled a one-step removal of customer engagement and, in turn, this means bartenders can engage with many more customers than before. As you can gain much more information through this channel than if you were to try and speak to each one on a busy evening. Twitter, Facebook and email will become your friends – and your new way to make new friends – because by being your customer’s friend you’ll take them back to the good old days, and this could be the difference between fading away and reinvigorated, transformational growth

Evidence-based marketing: myth or reality?

Apparently a Facebook ‘Like’ is worth £7. Isn’t that brilliant? If you have a hundred thousand followers that makes your social media strategy absolutely rocking news. The question is, says Felix Velarde of digital CRM agency Underwired, how many of you really believe it?

We ran a campaign for an FMCG client a while ago, in which we emailed a segment in the customer database using very specific messaging – it was relevant, timely and well-designed. The click-through rate (CTR) of 88% was staggering in anyone’s book. I’d like to say every campaign we run gets that kind of result, but occasionally we crash and burn too. Anyway, wow, what a great result. But what does it mean?

The digital marketing industry seems to be founded on metrics and key performance indicators (KPIs) of variable – if not downright dubious – quality. If you judge any of the industry awards you will know that half the entries benchmark against industry standards which in a lot of cases appear to have been made up on the spot: “Campaign X achieved a CTR of 12.3%, beating the industry standard 8% and therefore deserving of a Gold.”

This is where I have the first of my issues: even if there *is* an industry standard click-through rate, does it really apply to your campaign? Our campaign got an 88% CTR because it was highly targeted to a known audience expecting the email with a fantastic proposition. If it had achieved 12.3% or even 24.6% it would have been a failure. The truth is that no matter what the CTR is, one thing it is not is evidence of anything that isn’t either subjective (“I say, that’s awfully good isn’t it?”) or comparative (“Blimey, did much better than last time!”).

The second of my issues is that all these benchmarks aren’t evidence of sales. And hard revenue really is the only figure that ever matters (and before you ask about brand consideration or sentiment, these are both abstractions of sales – higher brand consideration may imply higher resulting sales, though you’d have to test it to find out if it’s really useful). So what is the *evidence* behind the assertion of success? How do you track from the planning phase through the campaign, to the reality of the effect on sales?

Well, it’s pretty easy in e-commerce. Straight-through processing (which is what the financial industry calls tracking the customer from first contact to order) in digital is quite simple these days. Ad tracking (display, affiliate, PPC and natural SEO) is relatively easily linked to website analytics, thence to email service providers (ESP), back to websites and even passing through couponing and promotion systems. It’s been done plenty of times by the big brands, and even small brands can buy integrated services. So if you’re an online retailer it’s pretty straightforward.

If you’re an FMCG brand you can still do it, although you may need to be a little more sophisticated about it. For example, this might mean matching your customer segments to commercially available purchase data and running regular surveys to track buying patterns in your base and cross-checking them against buying behaviour. If you’re a retailer you can find mechanisms to collect customer data and purchase behaviour at Point of Sale and match this data back to the single customer view.

The trick here is to get real data about whether your campaign actually worked, as opposed to seeing floating KPIs that may or may not indicate the same thing – but which provide no scientific basis for decision-making. How many times have you seen “successful” campaigns stopped? There’s a simple reason: there was no evidence to show they generated a profit. If there had been and they had produced £26 for every £1 spent, nobody in their right mind would stop them. Evidence is critical. Myth or reality? Evidence-based marketing is the only thing you should be *allowed* to do!

2013 will be the year of Total Customer Engagement

The year ahead will be marked by a number of really interesting developments in how we engage with our customers. One that has caught up with us – and which marks a sea change – is the advent of the Generation Y customer base. Gen-Y uses mobile as its main medium of interaction; if you’re not using mobile to engage with younger customers then you’re probably missing the biggest trick available, though it has been a very slow start since the trend first became apparent five, or so, years ago. Just in passing, to give you an idea of how trends break, let me illustrate this briefly…

Imagine your customer base is, potentially at least, a million strong. And say on day one a single person uses mobile as their main device for browsing the web, and every day that number doubles. By the end of the first week, 64 people use mobile. By the end of the second week that’s 8,192. That’s the point at which as a business you might start thinking there’s a trend. In actual fact it only takes another seven days and that’s your entire customer base. If you didn’t engage in a mobile strategy after two weeks, you missed it. Trends accelerate in an exponential curve – and in the internet age a trend can look slow for a few years, when in fact the numbers are doubling every month. Mobile is one of these, and if you target people born after 1982, that means you’ve got to jump, now!

There’s another trend that’s been a few years in the making, but in 2013 it will have gathered a critical mass, and it’s going to affect you – like it or not. The trend is for Total Customer Engagement. In plain English this means joined-up marketing, the opposite in fact of the “silo” thinking that has driven marketing during the first couple of decades of digital.

Most companies today have a web strategy, which may include e-commerce. They will also have a nascent social media strategy, a Google Adwords programme, and maybe an eCRM programme serving segmented comms to different sets of customers, split by value, behaviour, demographics and motivations. All of which have given businesses learning and insights into how they engage with customers using ‘new’(ish) channels. What’s lacking is integration. The sea change in 2013 requires a fresh way of thinking about customer engagement, which puts the customer at the centre. The trends feeding this are the shift in the balance of power from brand to customer, driven by the shift towards peer decision facilitated by social media, and the power of collective reviews.

So how does a business tap into this before it’s too late to do anything about it? A shift of focus is required, meaning the marketer needs to understand where the consumer is going to be, on- and off-line, when they are at a moment when they are likely to change their view of a brand, ideally positively. If you can identify that, Customer Type A is likely to be on a mobile, using Twitter, when considering whether to shortlist your brand, then you can target them with the right message at the right time and in the correct format. By mapping where the customer is at each critical point in their relationship with you, and mapping a rational sequence of nudges to take them from pre-custom to loyal customer, and then creating a matrix of comms against medium for this map, then you have a plan for engaging them. This Total Customer Engagement plan gives you as a business several things: a plan that can be tested, benchmarked and improved; a brief for your team and their suppliers so they have specific tasks to achieve against your business’s Key Performance Indicators (KPIs); and a framework that can be adapted when new trends become apparent.

This kind of approach is essentially multi-channel and channel-neutral. It’s also measurable and of course by its very nature future proof. By investing in developing this kind of (actually very simple) framework and marketing architecture, it will protect you against the overwhelming trends that are often nearly impossible to spot early, but which end up rather too quickly having strategic impact. Do it now, before it’s too late!

Press release – F Hinds appoints Underwired

F Hinds, Britain’s leading independent family jewellers with over 100 high street shops, has appointed specialist digital strategy agency Underwired to build its online customer engagement programme. The retailer, which was established in 1856, serves around 2 million customers every year.

Underwired will develop the brand’s digital marketing strategy and will work closely with the board of the brand.

F Hinds’ Andrew Hinds said of the appointment, “We have a close relationship with many of the customers who use our shops and we want to be able to extend that to provide the same quality of service to them online. Our outlook is long term and hard sell is an anathema, so we only want to be in touch if we have something relevant to offer our customers as individuals. We chose Underwired because their approach mirrors ours and because they are straightforward and focussed on the simple things which matter the most.”

Underwired’s Marketing Strategies Head Matt Button commented, “We’re very excited to be working with such a well established retail business. We will build a long term strategy using our experience and expertise from working with other retailers that remains true to the values and ethos of F. Hinds.” Button was previously Head of CRM at retailers Waterstones and HMV, where he led award-winning customer engagement and loyalty programmes.

Underwired’s clients include Sony, ASICS and Bupa.

http://www.fhinds.co.uk
http://www.underwired.com

Making sense of the marketing maze

FMCG Magazine, issue 11 volume 13

Why is – and why should it be that – FMCG marketing feels so vague? Everyone thinks the be-all and end-all is the big TV campaign. It takes six months to write, three months to edit, a whole bunch of illustrators, animators, storyboardists and directors’ assistants, a hundred grand in production and a hundred more grand for a slot in Corrie. This elephantine effort would be fine, were it not for one tiny flaw: you know fifty per cent of it works, but you don’t know which fifty per cent. The elephant in the room is accountability, and in today’s climate marketers must, must be accountable, as must their marketing.

We’ve come out of a recession, and are wobbling around the edge of a second iteration. Consumer confidence is low, partly driven by what people see in the financial sections of the TV news and partly driven by the doom mongers in the red tops. Businesses like yours aren’t bonkers so there is close scrutiny of every budget from procurement to overheads to manufacturing to marketing. The days of the “Let’s do TV, it works for the big brands” being sufficient justification for the board are over.

There is a second driver at work here: FMCG brands don’t sell to consumers, you sell to intermediaries – Nisa, Bookers, Asda, Tesco. You might argue that they should be doing your marketing for you (in fact, perhaps you should argue; if you spent your TV budget on selling more lines into Tesco they would have a vested interest in pushing your products hard, and yet we have devised ourselves an effective if expensive pull-oriented strategy, presumably because such tactics are too much like hard work). Yet we compete for shelf-space, PoS, the (very) occasional appearance in a promotion or an ad. FMCG brands focus their marketing towards the end consumer, essentially cutting out the middle man and appealing for share of basket. Works beautifully for the supermarkets.

So, two drivers: bypassing the supermarkets to get straight to influencing the consumer, and a terrible lack of clarity about whether this consumer focus can become more focused. Oh, and a third factor: the supermarkets sneakily deciding to launch their own brands (as distinct from own-brand) competing directly with yours. Anyone would think you were in trouble.

Over the past sixteen years – in fact ever since the Snickers® MegaBite online community was created for the brand’s Euro96 sponsorship – digital has become a valid and very useful channel for reaching and engaging consumers. Multi-brand FMCG companies have created websites, communities, games, multimedia, email and mobile campaigns very successfully, if success is measured in awards, media exposure and word of mouth. Over the past ten years, online channels have become properly measurable. The rise of analytics, and analytics specialists, has allowed marketers to track users’ online behaviour in great detail. Marketers are familiar with terms like UX (User Experience), IA (Information Architecture) and User Journeys (a term we appropriated from the supermarkets as it happens). We can drive people to websites, deliver appropriate experiences that support the brand architecture (brand onion, pyramid, pretzel... your ad agency will have its own version), and increase dwell time (the amount of time a consumer spends wandering around, through engagement or confusion, your website).

Digital can track absolutely everything. So it’s slightly surprising that most FMCG brands have not, because they believe they cannot, tracked the value they get from it. In the days when nobody knows which fifty per cent of the advertising works, you would think that having such an auditable medium would be a lifeline.

Digital means a consumer’s activity can be tracked all the way through to a sale. For example, if you sell a tin of beans on your website, we can track a visitor from before they get to the site (their first click on a Google Adword or a banner) through the site, around the site, to the basket and to a successfully concluded sale. We can attribute sales value to visits, which in turn means we can optimise campaigns, spend more on the sources which produce the highest sales, and generally be pleased that you know which fifty per cent is which. We can distinguish good from bad and make commercial decisions based on evidence. And evidence-based marketing is what your board wants.

FMCG doesn’t trust digital in the same way. It’s why, for example, most brand campaigns have a limited shelf-life online, and why websites get replaced with alarming frequency. You’re marketing to the end consumer, but you’re selling indirectly. This perception is common among FMCG marketers: indirect means indistinct. Decision making is therefore down to gut instinct – and how many awards the campaign wins. For me, that makes online marketing for FMCG brands a hopeless case. I want to know how to attribute value, no matter how indirect the sale is.

So let’s discuss a method which means that indirect doesn’t necessarily make it quite so hopeless. We’ve used it over the last four years for McCain Foods.

We started with a database of customers, acquired from a number of sources: bought lists, competition entries, newsletter opt-ins; in fact anywhere we could find data. We cleaned it up, got rid of the stale, unidentifiable, lapsed and suspect data, and created a robust base of legally opted-in people. We put together an email programme. This was pretty simple, consisting of product descriptions, recipe ideas, offers and simple calls to action. This gave us a backbone we could measure, and measure we did.

You will be familiar with the normal email marketing metrics: Open Rates, Click Through Rates and dwell time. We benchmarked the programme, making sure we had some consistency to start off with, so we could run some experiments. The first experiment: when should we send these emails? We sent the same email every couple of hours to a different section of the database to establish which time of day got the best open rates. At this best time of day we sent an email every day of the week to see which day of the week got the best open rates. Inside eight days we had the optimum send time.

The second experiment involved benchmarking against the real world. The assumption was that anyone in the database would be more engaged with the McCain brand than the general population (for obvious reasons: these people have opted in to regular emails, and they are getting regular brand exposure). What we wanted to do was to see if we could affect behaviour over time. Working alongside the brand tracking studies already being performed by Hall & Partners, Underwired created a comparable set of questions to mirror the study, in effect asking the same questions of the database so we could compare database versus general population at start, then after six months.

The results at the start were entirely predictable: 61% of people in the base loved the brand versus 20% in the general population. By the end, after the email marketing programme had been in action for six months, that score had risen to 64%, and in fact the gap had widened to 11%. The programme was clearly driving changes in perceptions of the brand, against a general fall in the advertising-only scores. But still, indirect and indistinct. How do we change this?

The next step for the campaign was to find a value benchmark. This consisted of two distinct phases: first find the comparison data, and second find a way to accurately measure any changes wrought by the digital activity. By using email only, the customer journey was kept very simple, and there was a built-in mechanism for running surveys so we could establish consumers’ shopping behaviour.

First we sourced a chunk of useful data. This came by way of Dunnhumby providing real-world shopper behaviour from Tesco customers; we sought out product choice, average purchase value and purchase frequency.

We continued this stage of the journey into attribution by refining the segmentation of the McCain database. The segmentation was fairly simple: brand engagers, brand resistors, category resistors, neutrals. This was also split demographically. The segmentation was tweaked to exactly match the Tesco shopper profiles so we could accurately compare one with the other.

So what have we found? We have discovered that when we put a person into the eCRM programme, in the first six months their purchase frequency goes up by 3%. Knowing what we know about average purchase value (in £s) and frequency for each segment, we can therefore easily find out not only how much the change is worth within the base, but also how much we should invest in acquiring more people into the database in order to drive ever-increasing incremental profit.

So what does this mean? Well, for one it means we know precisely which segments are worth investing in, how much to invest, and what the sales volumes we drive will be. This makes TV seem vague indeed – we do, now, know how to attribute value even when we’re marketing directly yet selling indirectly. We can justify every penny of the digital marketing budget (or at least that portion that’s spent on auditable campaigns) and, in a recession or in a post-recession world, that means we can be certain that what is being done is being done right.