In a recent report by Deloitte Digital and Salesforce, we surveyed more than 500 global brands to determine the impact of data not only on customer experience, but on the business performance of the brand itself. One of the key findings: brands that focus on data to a higher degree perform better than those who do not.
Still, many brands are not prioritizing data and AI. For a deeper look at the mindset of today’s data leaders, we chatted with Rama Ramakrishnan, Senior Vice President of Data Science for Salesforce Commerce Cloud.
Heike: Elite performers, which we characterize as brand leaders that reported a revenue increase of at least 10% in the past fiscal year, focus on data at nearly 2x higher rates than under-performers. Is that a surprise? How does focusing on data drive B2C brands to success?
Rama: This is not a surprise. If you step back and look at data, it’s an asset like any other business asset — your brand, your products, your culture, your retail stores. Any management team that knows how to get business value from what they own is going to have success. People who view data rightfully as a strategic asset, and give it attention, will unlock value.
This is true for all companies, of course. But for B2C brands, it’s even more important. These brands, by definition, serve consumers. We know consumer expectations of how they want to be treated (personalization, relevancy, immediacy) and those expectations are skyrocketing. If you do really well in harnessing consumer data and serve consumers, you will be rewarded. If you don’t, they’ll leave you for someone else.
Heike: While underperformers lag in all key areas of data management, the top three data deficiencies are
If you don’t have these three areas figured out, how does that pose a roadblock?
Rama: Let’s start with security, which has been in the news a lot recently. We’ve all seen some brands have suffer data breaches. There’s no question that these incidents can significantly damage your brand. That’s why it’s so important to treat consumer data very carefully. Companies must put in place a robust, secure way to collect, store, and use data.
By the way, I’m very proud of the way Salesforce approaches this. Trust is our number one company value and everyone takes it very seriously.
You asked if security is a roadblock for usage of data. Well, it’s not so much a roadblock as a requirement: you need to design it into your products and processes from day one, it can’t be an afterthought. If you don’t, you’re exposing your brand to problems down the road.
In terms of agility – obviously, it really matters for brands and B2C companies. I commented earlier about consumer expectations. On any given day, what your target audience wants and what it doesn’t want may change, and quite quickly at that. If you’re not in tune with this change, you’ll lose business, you will miss out on opportunities. For example, some of our customers tell us that when a celebrity is seen wearing a clothing item from their catalog, they seen an immediate uptick in site traffic and people searching for that item. If merchants were agile and tracking data in real time, they will spot this trend and modify the site in real-time to capture that surge in demand, driving more business.
Governance is important because if you don’t have it, your agility may be impaired. For example, if someone is analyzing some aspect of their business and need some key data, where do they get it from? Who knows where it is? Is it the right version? Is it up to date? If you can answer these questions quickly, you can analyze and drive actions faster.
Heike: Actionable consumer data is a brand’s superpower. It enables companies to know consumers in the moment and grow intelligence with every step. What do you think prevents data from being actionable? How can we make data insightful, not overwhelming?
Rama: It’s easy to create data and reports and dashboards, but it’s very hard to figure out what to do with it. It’s easy to create a bazillion reports, so they get created.
I think of this as the “last mile of actionability”. To bridge this, people creating reports need to work with business experts – the users of their reports – and a report should be declared “done” only if it recommends potential actions for the user to consider. If the dashboard says “traffic is down, we suspect these three reasons, therefore go check the following …”, that’s actionable.
It is not easy to drive reports all the way to recommendations but if you can do it, you will unlock a lot of value from the data.
The bottom line is that data that you don’t act on is as useless as data you never saw in the first place.
Heike: The potential business impact of AI is massive, but adoption rates are still low. On average, just over one-third of brand leaders have adopted any given AI use case, with more popular applications in tailored pricing and promotions and relevant search results. Why isn’t AI used more often?
Rama: The first problem is that, this is hard to do. There’s been a lot of focus on the fact that AI tools are readily available for download. But to build a compelling AI-based solution using these tools, you need AI talent, which, of course, is in short supply.
This is exactly why Salesforce is baking AI directly into all its products and solutions. By doing so, we are removing the need for our customers to go out there and try to hire AI talent. We are removing the need for ad hoc data acquisition projects. In short, we are de-bottlenecking the adoption of AI.
The second problem is that AI is still new and this leads to mis-steps. Best practices haven’t crystallized quite yet. Furthermore, some things are easy to do with AI, and some are hard and it is tricky to figure out which is which. For example, classifying images into a few different categories (e.g. shoes vs handbags) is an easy AI problem. But using AI to optimize search results for each of millions of shoppers is not easy. All this said, I think this is a short-term problem since companies like Salesforce are infusing AI into their products directly so that customers can get value without staying on top of all the latest trends or hiring a huge team of data scientists.
Also, the startup ecosystem is exploding with AI startups. Kevin Kelly (the founder of Wired) wrote that, “the formula for the next 10,000 startups is that you take something, and you add AI to it. We’re going to repeat that one million times, and it’s going to be really huge.” I see numerous startups and you can bet that they are knocking on the doors of B2C brands. Some will succeed, and those successes will dictate what we see in the next generation.
Heike: The number one use case for AI is tailored pricing and promotions. Any thoughts on brands doing this really well today?
Rama: One really compelling use case is dynamic pricing, as opposed to personalized pricing. Dynamic pricing is real time changes in pricing – while prices may go up and down over time, at any given time, all shoppers will see the same price. It has become popular, particularly with the rise of marketplaces. Many third party sellers on Amazon are competing with similar or even identical products there, so those prices are changing dynamically. If your price is static, you lose, so folks that sell through marketplaces adopt this.
Promotions are another great use case because you can target different segments of your customer base with things like free shipping and buy one/get one, without changing your list price. It can be tailored to people with psychological preferences and so on. Overall, as we infuse personalization into the fabric of the shopping experience, everything becomes more relevant, useful, and friction free for the shopper. It’s a no brainer, which is why we’ve heavily invested in it.
Heike: Brands plan to hire 50% more data scientists over the next three years. Given what we’ve heard about data problems, is this wise? What should they do beforehand, to not put the data scientist cart before the horse?
Rama: If you want to get value from AI, it’s four-legged stool. You need data, technology, data science talent, and all important business talent. Having a lot of data science talent is great, but if you don’t have data, business decision makers, technology, you won’t do much because one of those things will be the bottleneck. The technology is the least problematic aspect.
Data can be a problem but if you can get your hands on good-quality data, you just need a few data scientists and some talented people on the business side of the house to work together to get the job done. If you toss a problem and some data over the wall to your data science team and wait for magic, you will be disappointed. You need to free up business talent to work closely with data scientists.
Further, I’d recommend a crawl, walk, run, approach. Identify a problem or opportunity, hire one or two data scientists, find someone who knows the business really well, then have them work together and build a proof-of-concept, demonstrate that it works, rinse and repeat.
Heike: The promises of AI, data science, and machine learning loom large. But data alone won’t save the day without some serious collaboration with teams across marketing, commerce, and service. How can teams unify and align to set themselves up for the amazing potential for data science in the years ahead?
Rama: If I were the CEO of a B2C brand, I would form a cross-functional SWAT team across service, marketing, IT and data scientists. I’d tell them to choose one use case that cuts across functions, and work on that. Build a minimum viable solution quickly. Demonstrate success. Iterate, refine and scale.
The moment you demonstrate a cross-functional win, word gets around, good things start to happen. Everyone wants to get onto this bandwagon, and you’ll gain a lot of momentum.
Download the full report from Deloitte and Salesforce, Consumer Experience in the Retail Renaissance.