How to Use Customer Data to Anticipate What Customers Want
Written by Tom Peace, CEO & Founder of Sway Outcomes
Tom Peace is the CEO and founder of Sway Outcomes, a global customer engagement consultancy, and the former leader of The Loyalty People. He has advised brands including Sally Beauty, Wingstop, Vans, The North Face and Saudi German Health.
Most brands already know more about their customers than they'll ever use. The problem isn't a shortage of data. It's that the data sits in five systems and only ever gets used to send another discount. Here's how to move from knowing your customers to acting on what you know and eventually getting there before they ask.
Why do discounts beat data in most brands?
Because discounts are easy. Pick a percentage, pick a date, press send. Everyone gets the same thing, and the sales line moves for a week.
But customers have moved on. McKinsey's 2021 research on the value of getting personalisation right or wrong found that 71% of consumers expect companies to deliver personalised interactions and 76% get frustrated when this doesn't happen. The same research found that companies that grow faster drive 40% more of their revenue from personalisation than their peers that grow more slowly.
A blanket discount of 20% is the opposite of personal. It tells the customer you don't know who they are.
As I wrote recently, our maturity model at Sway Outcomes works as a staircase. At Level 3, you build the data infrastructure to know your customers. At Level 4, your customer relationship management (CRM) system uses that data to drive action. At Level 6, you use it to predict what happens next.
Not sure how your data stacks up? Email me at tom@swayoutcomes.com and we'll send you a free Customer Engagement Accelerator (CEA) report showing where your data is helping you and where it's holding you back.
What a single customer view means
Almost every brand says it wants a single customer view. Fewer agree on what one is. It isn't a giant database with everything in it. It's the ability to recognise the same person wherever they turn up and see the handful of facts that actually change how you treat them.
In practice, that means you can answer the following questions about any individual customer in one place.
What have they bought, where and how recently?
How do they like to hear from us, and what have they said yes or no to?
Have they contacted customer service? If so, was their issue sorted?
What have they told us about themselves?
If your online team, your stores and your contact centre would each give a different answer, you don't have a single customer view. You have three partial ones.
Ask customers, then give something back
Most brands rely on what they can observe. Purchases, clicks, visits. That's first-party data, and it's valuable, but it only tells you what someone did, not why.
The richer source is zero-party data, also known as declared data. It's what customers tell you on purpose. Their preferences, their plans, their sizes, their dog's birthday.
It comes straight from the person. It's given with permission, so using it feels helpful rather than intrusive.
The catch is that nobody hands it over for nothing. There has to be a fair value exchange. Here are a few that work well.
A short style or taste quiz that returns genuinely better recommendations.
A birthday or anniversary field that leads to something worth having on the day.
Early access or member pricing in return for completing a profile.
If the customer can't see what they get for answering, they won't answer. If you ask and then ignore the answer, you've made things worse than not asking.
Segmentation, personalisation and anticipation
These three words get used as if they mean the same thing. They don't. Segmentation is grouping customers who look alike. Lapsed buyers, high spenders, new joiners. Useful, but everyone in a segment still gets the same message.
Personalisation is responding to the individual. The email reflects what this person bought, and the offer matches what they told you. It's still reactive. Something happens, and you respond.
Anticipation is acting before the customer does. You spot that a regular is going longer between visits and get in touch before they've gone. You know a refill is due and remind them the week before, not the week after.
Segmentation spans Levels 3 and 4. Good personalisation is solid Level 4. Anticipation, using predictive analytics in real time, is Level 6.
You can't jump straight to anticipation. Prediction is only as good as the data and the habits underneath it.
How to start anticipating customers
Here's the order we'd suggest. Each step pays off on its own.
Pick three decisions to improve: Don't start with the data. Start with the decisions. Which three customer moments matter most? A first repeat purchase, a lapse in visits, a renewal date. Everything else serves these.
Map the data each decision needs: For each decision, list the few data points that would change what you do. You'll usually need far less than you think, and most of it already exists somewhere in the business.
Join it up around the customer: Connect those data points to a single customer record. You might use a customer data platform or make a sensible fix to your existing CRM. The tool matters less than agreeing on one customer identifier and one owner.
Fill the gaps by asking: When the data you observe can't tell you what you need, ask. Keep the questions few and the reward obvious.
Turn insight into triggered action: This is Level 4. Set up journeys that fire when a customer's behaviour changes, regardless of the date. For example, you might start a campaign to win someone back when their normal rhythm breaks. Measure each journey against a control group.
Add prediction once the basics hold: Only now does Level 6 make sense. With clean data and working journeys, predictive models can score how likely customers are to lapse, buy again or move up a tier. Artificial intelligence (AI) sharpens something that already works. It won't rescue something that doesn't.
The data mistakes we see most
In our CEA work, the same problems turn up across sectors.
Collecting everything, using nothing. A registration form with twelve fields and a CRM programme that uses only one of them. Every unused field is friction for the customer and a liability for you. Collect less and use all of it.
Creepy personalisation. Using data the customer doesn't remember giving you. Being relevant is welcome. Being watched isn't. If you'd struggle to explain how you knew, don't use it.
Treating the discount as the personalisation. Putting someone's first name above the same 15% discount everybody else got isn't personal. It's a mail merge.
Buying the platform before fixing the plumbing. A new customer data platform won't solve a problem nobody owns.
None of these are really about technology. They're about ownership. I talk more about the ownership gap in my interview on why customers stay and what brands often get wrong.
Keep trust at the centre
Every step up the staircase asks the customer to trust you a bit more. That trust is easy to spend and slow to earn back.
Salesforce's 2023 State of the Connected Customer research, based on a survey of 11,000 consumers and 3,300 business buyers across 25 countries, found that 67% of customers expect companies to understand their needs as those needs change. It also found that when companies clearly explain why they're asking for personal data, customers are more likely to trust them with it.
So say why. Say what customers get. Make it easy for them to change their minds. The brands that do this well treat privacy as part of the proposition. That means thinking beyond the legal box at the bottom of the form. For a retail view, this piece on how to use data to personalise the customer experience in retail is worth a read.
Frequently asked questions
What is zero party data? It's information a customer chooses to share with you directly, such as preferences or plans. First-party data, by contrast, is what you observe from their behaviour. Information customers share directly can give you a clearer picture of their preferences and feel less intrusive to use.
What is a single customer view? It's one shared record for each customer. The point isn't to store everything. It's to recognise the same person across every channel and treat them consistently.
How is predictive analytics used in marketing? Predictive analytics uses past behaviour to estimate whether a customer is likely to buy again, lapse or take another action. Brands use those estimates to act before the moment passes. Predictive analytics works best with clean, connected data.
Start with what you have
You don't need more data to start anticipating what customers want. You need to use the data you've already got to guide a few decisions that matter while keeping the customer's trust intact. Get that right, and the discounts can start doing less of the heavy lifting.
If you'd like an outside view, my team at Sway Outcomes offers a free Customer Engagement Accelerator (CEA) report. We'll score your brand across all six levels, looking at your data infrastructure, your CRM and your readiness for predictive work. Then we'll show you the quickest wins. To get your report, email me at tom@swayoutcomes.com or message me on LinkedIn.
Read more from Tom Peace
Tom Peace, CEO & Founder of Sway Outcomes
Tom Peace is a customer engagement strategist who helps brands turn transactions into lasting relationships. Under his leadership, The Loyalty People became a globally recognised brand within the loyalty industry, with a rapidly growing and highly engaged following. He went on to found Sway Outcomes, a consultancy that goes beyond loyalty programme strategy to cover every aspect of customer engagement. Its maturity model has six levels and shows brands exactly where they stand and what to do next. By combining deep sector expertise with diagnostics enabled by artificial intelligence (AI), Tom delivers strategic clarity in weeks rather than months for brands such as Sally Beauty, Wingstop, Vans, The North Face and Saudi German Health.
