From Rock Musician to Optimization and Personalization Pro
Sigi Bessesen works as the optimization lead at one of the world’s largest banks, HSBC, in its retail banking side. This is the side of the bank that consumers like you and I engage with through products and services like mortgages, bonds, savings, credit cards, and checking accounts. Sigi recently shared with me how he got into optimization and personalization, projects he works on, challenges he faces, and advice to others based on his experience.
Finding the route to optimization and personalization
Sigi has always been a musician, but, in his past, he also designed websites for musicians, bands, and artists. In his early 30s, Sigi begrudgingly realized rock stardom might just not be his path. That’s when he decided to take his creativity and design experience into a new career. “Although my coding was not mind-blowing, some of the layout and functionality was unusual for the time. I wanted to use that experience to find my dream job,” he said.
He landed that job, working for Google to improve website layouts for their syndicated partners who were using ads and search results to improve revenue. Later he helped customers improve their websites through UX and conversion optimization. Sigi says, “I loved the logic and grounding in data that goes into optimization, but I also liked working on the user experience. I really appreciated that optimization made user experience design more of a democracy.”
From that job, Sigi moved to Cape Town, South Africa, where he once again developed websites. But in 2011, he became an Adobe consultant in the UK, a job he said was incredibly rewarding, but over time became increasingly technical and less creative. He already knew the technical aspects of his client HSBC, and believed that working for them directly would allow him to focus on the creative aspects of optimization. He took a job with the bank.
When Sigi joined HSBC, the bank had little optimization anywhere — he was the senior authority on optimization. The good news was that he had visibility into how over 30 of the UK’s biggest companies operated their optimization teams — how they were structured, backed by the organization, and solved operational issues. Plus, the bank was highly motivated to establish a strong optimization program.
Early wins made it easy to maintain that support and expand optimization. Sigi says, “Initial tests knocked the ball out of the park, so the doors opened fully quite quickly.”
Risk and fraud
Three years in, and the optimization program is also working with teams like risk, governance, and fraud — not typical areas for optimization. But with HSBC focusing on digital, everything is new. Words like “re-platforming” and “integrating” are uttered frequently. It makes sense that care must be taken with all this change. Sigi says, “You’ve got this huge ecosystem of tools that people feel they need, but each tool is part of a pipeline. Making changes in one place impacts things elsewhere. You have to consider the implications of each change.”
Sigi explains that they can use Target audiences to flag fraudulent behavior early on. He says, “For example, if someone fills out a form in three seconds or takes too long to fill one out, it’s likely fishy.” He says that there’s great potential for using audience identification with Target beyond addressing fraud; for example, to alert on anomalies in account activity, look at the analytics behind that activity, and use that information to mature the customer journey.
Challenges specific to financial services
One of the challenges to optimization in financial services stems from the two very different ways customers interact with the bank. In one, they want the bank to meet their more frequent needs like paying bills, checking account statements, adding to savings. The other are the less frequent bigger “life events” such as getting a mortgage, taking out a loan, opening a savings account, or making investments. We can impact that first set of more frequent customer needs by making tasks quicker and easier. We can improve our ability to meet customers’ “life events” needs by being personal and relevant.
Understanding these different types of needs is enabling the optimization team to find a variety of ways to help the bank’s customers. Sigi says, “With our Help and Support pages, we’re using Adobe Target to hone that content based on the page the visitor comes from. If you’re coming from the credit card page to the Help and Support areas, we’ll focus the topics on credit cards. Same for loans — we’d surface loan content.” These changes make it easier for users to find what they want, and produce substantial gains for the bank.
“I find it crucial to be able to stop and take stock regularly; to see what other companies are doing, regardless of industry. The Adobe Summit online sessions represent a vault of really valuable sessions. To mention just a few, at Summit I’ve been inspired by how the National Bank of Canada has used Recommendations and how SunTrust approaches using machine learning.”
Automation in Adobe Target
Sigi stressed the importance of using the information you have about your visitor to personalize and make it easy for visitors to find things on your website. He says, “If you’re running a car website, and a visitor looks at a particular model, on their next visit, show that model on the homepage. In banking, if a visitor looks at a specific loan or credit card, resurface it on their next visit. It will benefit the consumer and the bank.”
This is where machine learning and AI becomes valuable. It automates segmentation of those smaller and more specific audiences — long tail segmentation. Sigi says, “We have data that lets us know new versus returning visitor, accounts they have, browsing behavior, referrer, and so on. All this lets us help a visitor find the information they need.” But setting up highly specific audiences manually is impossible. And testing 80 to 100 banners compounds the problem. Sigi says, “The minute you apply machine learning and AI, you’ve removed a lot of the effort and resources needed to set up the activity and deliver the best personalized content.
AI also provides valuable learnings about the content that customers prefer. Sigi says, “If you identify a segment through the automated delivery of banners to individuals using Auto Target, you can zero in on what things to optimize. For example, if a specific credit card banner works well for a particular segment, we can target that banner more specifically.”
The optimization team is just starting to use the Personalization Insights Reports in Target. They’re getting some insights into audience segments, and then looking at their analytics reports to see how that audience performed and how, for example, visitors get to the application page for a product. Sigi says, “We use a combination of whatever we can get our hands on to get an inkling of the next step and iteration. For example, if we have an indication that a visitor likes a particular product such as a gold credit card, we can surface that card more prominently or in a menu item. That leads to higher conversions for that product.”
Advice when starting out in optimization
When hiring for the optimization team, Sigi explains that they look far less at experience level than who the person is. Then they train them. Sigi says, “We have a five to eight week ‘Onboarding Academy’ that takes new hires through learning about our optimization approach and processes. As they’re doing this, they’re getting one-to-one introductions to the team and our stakeholders.” In the first week they are mainly asked to just have coffee and talk to people. Taking this approach has allowed HSBC to build a highly effective optimization team.
Based on his experiences, Sigi offers this advice if you are starting off in optimization:
- Take the crawl, walk, run approach. Sigi says, “AB testing is often a little misunderstood. Just because you have a visual editor and 99 ideas for updating your website, doesn’t mean you should test all those ideas. Learning any new skill takes time.” He recommends methodically building a solid foundation of learnings, with the caveat that “Every now and then be daring and put your neck on the line. It can be worth it.”
- Establish a testing everywhere mindset. People can get hyper focused on a single big idea that they know will work. When the idea flops, you’ll hear them say, “It seemed like such a good idea at the time.” Sigi advises testing everything. For example, before you launch a new feature, test to make sure it works and is accomplishing what you wanted.
- Make friends, find a confidant, and be known. For optimization to succeed in a company, make friends and find someone you trust for sound advice. He says, “Do what you need to do to get going. Make sure people know who you are and what you’ve done. Spread the word about optimization.”
- No “land grabbing.” With optimization, protecting your turf really doesn’t work. Success comes from cooperation and sharing best practices — even externally. Adobe Summit is a great place for best practices sharing. Sigi backs up this advice with his own actions. He says, “One of my aspirations for optimization is eliminating the need for a dedicated team to ‘do optimization’ as the testing and experimentation approach becomes a part of the organization’s DNA.”
- Put data behind your hunches. Sigi admits that optimization is still a new area. “I think you would struggle to find an ‘Optimization Expert’ — I hope nobody calls themselves that. Nobody can be 100 percent certain of the impact of a change or addition. Optimization done well is, in my opinion, all about process and cadence — with a solid foundation in data.
Sharing to build the optimization knowledgebase
Sigi says, “There’s so much to learn about optimization, and because it’s such a relatively new practice, nobody can really claim to know the answer.” For Sigi, improving at optimization is much like improving as a musician. “Most musicians understand early on that you can’t improve in isolation. You have to talk to people. I know I’d be much better bass player if I’m surrounded by bass players than being alone.” Fortunately, people like Sigi are happy to share those best practices and build up the knowledge in this maturing field of optimization and personalization.