You Don't Know Your Customer Anymore
Written by Francesca Danzi, Chief Client Officer / Digital Transformation / Startup Advisor / Keynote Speaker
Traditional algorithms cannot help in predicting purchase behaviors today because they are based on historical data. Behavior change brought about by Covid19 invalidates much of this data. How can brands effectively reach customers today and personalize engagement across channels? Reflecting on this challenge, my thought went to ActionIQ - intelligence hub that enables the delivery of “personalized experiences at scale”. Tasso Argyros, its co-founder and CEO, thinks differently and thinks ahead. Here the highlights of our conversation last week.
FD: In a recent article Mark Payne, co-founder and president of Fahrenheit 212, urged businesses to “take all the insights you’ve gleaned from years of mining the gray matter, hopes, needs and behavior of your customers, shove them in a proverbial file, and start over”. What do you think about it?
TA: I think it's a great point. Think about AI today: it’s data science algorithms. Algorithms need to be trained on historical data. What’s happening is that all the historical data reflects behaviors that are no longer there. Hence, all the algorithms based on that historical data are useless right now. It is scary but that’s the reality. The beauty of a customer data platform (CDP) is that the data and the models live in the same place, therefore the models can be retrained with data from the previous month or even week, in a very agile way. If your customer behavior changes, you can press a button and we re-train your models. If you don't trust the models, you can look into what's informing them and draw your own conclusions. The sooner people accept the fact that all the models they have deployed are useless, the better it is, because they need to get going, shift their focus from the models to the new data and use a more agile platform to action it.
FD: Tasso, can you clarify what customer data platform are and why there is so much interest around them today?
TA: A CDP covers the last mile: it makes the customer data that is inside a company visible and accessible for marketing, sales and customer servicing purposes. Leveraging this data to improve customer experience is a massive challenge. The growth of the costs of customer acquisition and the shift towards increasing customer lifetime value have made CDPs very ‘hot’.
FD: Absolutely, building customer relationships is paramount and crisis add a sense of urgency. How does ActionIQ stand out, what makes you different from other providers out there?
TA: A lot of companies claim to be CDPs because they leverage some customer data, but they are more ecommerce specific companies or point solutions like web and email personalization platforms. Their main limitation is that they take data from just one channel to personalize only that channel. The difference with ActionIQ is that we take data from everywhere, any system and source a company may have, online and offline, and we can then push data into web solutions of any kind, but also email, direct mail or even analytics systems to enable further analysis. We are the only company that can pull data out of data lakes in the order of terabytes. It is a very complex task that requires a product that can scale and that understands how the data is structured. We are totally channel and function agonistic and believe that this is important because to provide a truly better experience you have to be wherever the customer is and take into account all of their past interactions with you.
FD: And the platform learns from activations and campaigns, correct? Can you tell me more about AIq Artificial Intelligence and your vision for it?
TA: AI is a big part of our platform; it is essential to enable personalized experiences. For us personalization means taking into consideration the individual journey and intent of every customer. Think about Amazon, that uses a traditional algorithm for recommendations: people who bought this product also bought these other ones. The personalization happens at product level, which means that you and I can have a very different purchase intent, very different journeys and destinations linked to this purchase, but we both get exactly the same product suggestions. What we offer with ActionIQ is the ability to understand the full picture and predict where each individual customer wants to go next in their journey. This allows the diversification of suggestions / communications / activations.
FD: How can you claim that “Action IQ is the fastest path from customer data to personalized experience at scale”? It is quite a bold and enticing statement.
TA: What takes time and constitutes the biggest risk when you are deploying a solution is data not being in the right format. We remove that risk as we can process data in any format. We work in a very agile way and can deliver results very fast, 6-8 weeks. We have the best time to value in the market. As a response to Covid19, we have developed a plan that can be executed even faster.
FD: That's fantastic. What is the effort required, company side?
TA: It's actually not significant. We have shaped a team of experts both on the use case, marketing side and on the data science side. We run a value optimization workshop for prospect customers, free of charge. We help them identify their primary goal, e.g.: improving retention, moving more people from physical to digital channels, trying to sell the same number of products but with better margin. Then we look at the channels that can help them get there and focus on the required data, wherever it may be. Having clarified goal, channels and data, we develop a plan that can be executed in 4 to 6 weeks.
FD: How can your platform take into account external data, like macro-economic or environmental data?
TA: AI cannot replace the marketer. There is always business context that is not reflected in the data. The future of business is in the interaction between humans and AI. What we have developed, to facilitate this interaction, is what I call a white box model, in contrast with the traditional black box model algorithms. ‘Black box’ refers to the fact that they can predict what your customers want to buy next, but there's no visibility into how the algorithm is making that prediction. The engineers who built the algorithms are the only ones that can influence them. Transparency + flexible user interface are the principles on which we have built our AI, a white box model. With ActionIQ a businessperson can understand why an algorithm is driving certain conclusions, what data and patterns inform it. They are also able to influence the algorithm with business context e.g. bad weather coming, inventory shortage.
FD: That’s refreshing, I felt the frustration in using black box models myself in the past. Let’s talk about your fashion and retail customers. Some have been using your platform for a while, yet I see untapped opportunities in their CX personalization. Is it fair to say that the capabilities of the platform are often underutilized and, if yes, why is that? (Strategy, talent, processes)
TA: First of all, companies need to use all the data. You can have a CDP deployed but still only testing a limited amount of data. The second thing is to have all the channels connected to the CDP, not just the one or two biggest revenue drivers. Channels that don't provide immediate revenue still affect the customer. This requires clear ownership of the customer experience. I do rejoice when I see customer experience consolidated under a single person or a couple of people. Vera Bradley, one of my clients, has a Chief Customer Officer, a title that you don’t see very often and sometimes it's more of an influencing role rather than a functional one, owning a PNL etc. Clear ownership for all channel covering orchestration is instrumental to success. The final thing is ambitious leadership. One of my clients, who runs sales and marketing at a Fortune 50 company, said to me: I have about 200 campaigns going at the same time today, I want them to be 20,000. That means moving from a one-to-many approach to one-to-very few. It means having clusters of 50, 100 customers, that for a Fortune 50 company is very granular. That requires extreme efficiency, automating content, audiencing, activation, journeys… but it's doable with the technology we have today. What you need is ambition.
FD: Beyond retail, what other industries are ramping up in designing multi-channel, personalized customer experiences?
TA: Banks and financial services woke-up one or two years ago and realized that they are multi-channel businesses. They acknowledged that even people who have the time to speak to a wealth advisor, still prefer to communicate via text messages sometimes. The blending of digital and physical is happening everywhere, in every industry where companies communicate with customers. The crisis is accelerating this change. Every industry who had intermediaries is now trying to go direct to consumer. Wherever that's happening, CDPs become top priority, because customers are used to a highly personalized, high-touch model, and that has to be somehow replicated digitally.
FD: Let’s talk about how you are reacting to this crisis. Are you looking at your business in a different way, have you shifted your priorities?
TA: We called all of our customers and asked how we could help. We understood that the need was to reduce costs. Now a CDP adds value by improving customer engagement, but it can also help reduce acquisition costs by being smarter about the customers you have: where am I wasting money? You know the famous saying that half my marketing is wasted but I don’t know which half? If we can help figure that out, there's a massive savings proposition. In addition, we can help consolidate a lot of point solutions into one CDP, something that may not be worth looking into in good times but is valuable when budgets are tight. We came up with budget optimization workshops, that we are running with our customers.
FD: I mentor and advise fashion tech startups through the NY Fashion Tech Lab and other programs; in this time of uncertainty, they have lots of questions. You are both an entrepreneur and an investor. What message would you give to fellow entrepreneurs and startup founders as they navigate this crisis?
TA: It really depends on the stage of their company. For people who were thinking to start a business this would not have been a better time because when you start a company, you spend the first year or two on market research, product development … you can still raise money and your burn is very low. That's why a lot of tech companies were founded in recession time. When your burn is slow you go through the recession, then when you're ready to sell, the recovery has happened. The biggest problem is for companies that were just ramping up their growth phase: they may need to take some really hard decisions, make sure that they extend their runway even if that means scaling down and waiting. I talk to investors all the time and they are looking at companies that add a lot of fundamental value and are looking for ways to slow burn, meaning that with a relatively low amount of money can last for a couple of years. At the same time, founding shouldn’t be a problem for solid companies, since there's a lot of VC money raised in 2018-2019 that's waiting to be deployed and needs to be deployed, I wouldn't be too concerned with the availability of capital.