Perhaps the single most important algorithmic distinction between âborn digitalâ enterprises and legacy companies is not their people, data sets, or computational resources, but a clear real-time commitment to delivering accurate, actionable customer recommendations. Recommendation engines (or recommenders) force organizations to fundamentally rethink how to get greater value from their data while creating greater value for their customers. In other words, theyâre a terrific medium and mechanism for transitioning traditional managements to platform perspectives.
âBuild real recommendation engines fastâ is my mission-critical recommendation to companies aspiring â or struggling â to creatively cross the digital divide. Use recommenders to make it easier to gain better insight into customers while theyâre getting better information about you. Making recommendation an organizing principle for digital design distinguishes leaders from laggards.
Recommendersâ true genius comes from their opportunity to build virtuous business cycles: The more people use them, the more valuable they become; the more valuable they become, the more people use them. This embodies Tim OâReillyâs brilliant Web 2.0 ethos, which he articulated a decade ago. Recommenders continuously learn and improve from user experience. That means organizations can continuously learn and improve from their recommenders. Recommenders are digital platforms for virtuous cycles.
In my experience, legacy managements too frequently misunderstand recommendersâ role in driving innovation and cultural change. They treat recommendation engines more as e-commerce sales and marketing gimmicks â another feature to add to the site â than as crucial investments in virtuous cycle platforms. Recommendations are seen as a technique to sell more online, instead of a renewable resource for relentlessly improving customer insight and their own. These organizations donât design with virtuous cycles in mind.
Platform companies invest in virtuous cycles to solve problems; more-traditional firms invest in process improvement to solve problems.
At one multibillion-euro industrial equipment firm, for example, sales, marketing, and maintenance managers simply couldnât get beyond the belief that a smart online parts catalog was all they and their customers needed. âLetâs make search faster and simpler,â they said. âLetâs make it easier for customers to find what theyâre looking for.â So they did. Their working presumption and assumption was that customers knew what they wanted. Identifying purchase patterns and correlations that could algorithmically inspire recommendation or further inquiry was an afterthought.
Minds changed only after a respected rival and two of their biggest value-added resellers launched their own recommenders. The firm soon discovered that, in fact, many of their customers and most prospects didnât know what they needed. Online recommendations, divorced from high-pressure sales pitches, proved an excellent gateway for industry customers to get up to speed on complex product lines. Search improved recommendations; recommendations improved search.
One reseller posted a maintenance and repair content recommender, including video, on its site that bundled product and service advice. The reseller further opened up the site for curated comments from customers. Although the reseller was much smaller than its biggest supplier, its recommendations-oriented site enjoyed 50X more web traffic. More importantly, the supplier acknowledged, the reseller typically knows more about equipment status and planned purchases than it does.
âWe didnât think a recommendation engine was necessary at the time,â said one of the companyâs web developers. Indeed. A recent site visit confirms the companyâs recommendations are little more than repurposed catalog descriptions with links. Itâs now striving to catch up.
By contrast, a financial services firm grappling with the robo-advisers quickly agreed that its clients wanted more than just good advice; they expected financial recommendations that they felt empowered them. This created internal conversations the company had never had before. The robo-adviser team literally asked itself, âWhat would Amazon do?â
Should, for example, the robo-adviser display Amazon-esque suggestions like, âInvestors like you also consideredâ¦.?â and âInvestors who bought this fund/ETF/instrument also boughtâ¦â? The design team went through hundreds of mock-ups exploring not just how best to present recommendations but also how to identify how they wanted to learn from online interactions.
The financial services firm quickly realized that its robo-adviser needed to be a recommendation platform â the more people used it, the better the recommendations needed to become. Customer focus was key. Every effort should be made to personalize the recommendations; the robo-adviser should make exploring the recommendations simple and easy. The virtuous cycle had to be baked in.
The robo-adviser team understood that success required greater data visibility and access firm-wide. As a large legacy firm with hundreds of regulated products, the company was operationally siloed, with data scattered throughout the enterprise. The robo-adviser recommender became an opportunity to facilitate cross-functional collaboration between product groups and IT managers who literally had never communicated before.
Several groups resisted and pushed back against the robo-adviser teamâs requests for information and cooperation. But, to the C-suiteâs great credit, leadership insisted that everyone cooperate.
Why? As the CFO told me, âWe now understand this is what our future is going to look likeâ¦. Weâre going to have to share data across silosâ¦. We have to have digital processes that are data-driven by customers, not just our own people.â The robo-adviser recommender became template and test for the companyâs digital transformation initiative.
That represents an organizational, operational and cultural breakthrough, not just a digital one. Bluntly, recommenders make the best âcanaries in the mineâ for digital transformers. They inherently raise all the right questions and challenges any serious innovator should have about crossing the digital divide. They force organizations to think less about process orchestration and more about virtuous cycles. Digitalization requires revisiting business fundamentals, not simply upgrading IT.
Here are five recommendations for organizations aspiring to leap across digital divides:
Whom Do We Want Our Recommendations to Be Like?
Donât reinvent the recommender. Would your customers prefer an Amazon-like or Netflix-esque recommendation experience? Perhaps Quora or Stack Overflow content recommendations would add the most value fastest. Stitch Fix and Pinterest represent different genres of visual and curated recommendation experiences.
You should explore and experiment with the recommendation approaches that best reflect and communicate your organizationâs brand values and user experience aspirations.
Thereâs no shortage of real-world recommenders to emulate and innovate on. Similarly, thereâs no shortage of software â much of it open source â to build it with. The challenge is in how your organization should integrate and align recommendation experience and user experience.
What Do We Want to Learn from Recommendation Experience?
Companies like Amazon, Netflix, and Quora derive quality data and tremendous insights from tracking how their users observe, follow, and ignore recommendations. How do you want to act on what you learn from recommendation experience? Do you want to price differently? Bundle products and services better? Add or remove specific features and functionality? Identify emerging customer segments? Personalize offers? Test innovative ideas? What metric matters more: engagement or inquiry?
Properly structured recommendations generate at least as much information as they provide. Define what kind of learning you want your recommendations to generate for you.
How Could We Make Our Recommendation Cycle More Virtuous?
Answering the first two questions clarifies what kind of data youâll need to launch your prototype recommender. But are there other data sets, algorithms, and interactions that could improve the virtuous cycle? Should your recommender offer a âclick to chatâ option to facilitate an interpersonal connection? Might curated comments and reviews add measurable value to the recommendation experience? Could you ask customers if they want email follow-ups or white papers on products or services theyâve expressed behavioral interest in? Would a LinkedIn linkage make sense?
Recommendation engines arenât just about recommendations; theyâre platforms enabling all manner of digital informational interaction. Appropriately designed, these interactions create value for your customer and for you. Virtuous cycles can become more virtuous and valuable.
What More Should We Be Recommending in a Year?
As recommenders become more virtuous and valuable, the opportunities for novel and innovative recommendations arise. Product companies can recommend services, both their own and third partiesâ; service companies can recommend products. Some recommenders might even recommend introductions to fellow customers and clients. Look at how Amazon bundles its vast array of products and services into multiple recommendation formats. More data inevitably leads to not just better recommendations but also broader varieties of recommendation. That, of course, depends on what you want to learn and how you want to evolve your virtuous cycles.
It is a mistake to confine recommenders to the e-commerce domain. As services like Quora, Stack Overflow, and LinkedIn affirm, recommenders can do far more than recommend products and services for customers; they can provide content, advice, and information for decisions. Why not create recommenders to suggest people for new product teams? How about recommenders for business partners and channels?
Recommenders can be useful, useable, and used wherever people need suggestions and advice. Virtuous cycles shouldnât belong to customers alone. Wherever digitalization and data exist in abundance, the power and potential for recommenders shouldnât be far behind. The best recommendation for enabling digital transformation is to enable the digital transformation of recommendation.