Personalization is table stakes for any successful e-commerce website that offers more than a handful of products. As the diversity of products grows and each item becomes more targeted in use, it’s not practical to present one’s full selection to all customers (i.e., “Results 1-25 of 151,432 products”). It falls to site merchandisers to refine those results and decide which products to present to each customer.
In this article, we’ll explore the complexity of personalization and the role of contextualization in driving impact and revenue in B2B e-commerce.
The Evolution of Product Discovery in E-commerce
In the early days of e-commerce, product discovery was fashioned by “overall” metrics: “best-sellers”, “most viewed” and the like guided users to products that were broadly popular. In the consumer world, where best-selling products often imply quality or style, this was effective. But it did little to help merchandisers manage more niche products which were popular for specific, highly targeted customer types.
This led to a more refined approach, where sellers would watch patterns of traffic and purchases within categories and aggregate customer histories into an understanding of their preferences. By observing what customers viewed and bought over time, merchandisers could infer those customers preferences about products, and use that information to target campaigns. For consumer products, this was quite effective since personal preferences are not usually volatile, so a preference documented today is likely to be accurate next week.
Why Personalization and Targeting Efforts in B2B E-commerce Fail
Yet this is also a significant difference for B2B – and it helps explain why personalization efforts built on technology made for consumer products often fail when applied to B2B e-commerce. In B2B, purchases are made because the buyer has a specific problem to solve, but it is not a *personal* problem. Their purchase reflects an urgent need to solve an immediate problem, and for customers who are companies, that problem may be very different from one faced last week – even when the products needed to solve each are the same kind of problem.
Here are some examples:
- Consider a commercial electrical contractor who is bidding on a renovation project for a large multi-unit hotel in the downtown area one week, but the next week has a different order for a job to expand a local car wash to add extra bays. Both orders may call for switches, but the kinds of switches the buyer may search for while building the hotel proposal are likely to be quite different from the searches done for the car wash project.
- Similarly, a machine shop that specializes in manufacturing metal components may purchase end mills or threading taps every week, but it is all driven on their queue of jobs and customer mix. Say that machine shop foreman has two jobs coming up in a week – one for an aerospace customer who wants a part machined out of nickel alloy and another for a customer who refits and repairs engine blocks for commercial trucks for parts made from cast iron. It is unlikely that the same end mills or taps would be used for both applications; indeed, carrying over preferences from one of those projects to the other could lead to inaccurate recommendations and dissatisfied customers.
Personalization Versus Contextualization
Today’s leading e-commerce sites must account for this additional context of application just as effectively as they do for the contexts of product and customer. And it can be challenging for a site to infer application from customer behavior. Instead, application can be found in search terms used, faceted search selections, or accounted for in custom product configurators. It is critical for the site to quickly and accurately recognize the means used to identify applications and flag them as such but only for the duration of that session, so they are not “held over” – as with consumer preferences in B2C – into subsequent sessions.
Transforming in B2B with Contextualized Experiences
This is one area where machine learning and AI can deliver real results for companies seeking to differentiate their commerce experience from the competition — and raise the bar. Machine learning grants the ability to observe and learn from each and every instance of a term’s use during discovery. AI can observe how well certain terms serve as signals for applications by tracking what products are searched for and purchased when they are used.
It can take time to build up enough data for the results to be meaningful, but some platforms have an advantage in that they already have large data sets of transactional data that have trained their algorithms. It’s therefore faster to deliver meaningful results in a “contextualized experience” that works even in the kinds of complex technical applications found in B2B.
With these capabilities now available, we are entering a fresh era for B2B e-commerce where the clumsy, often inaccurate merchandising that characterized the earlier days of e-commerce is quickly replaced by new approaches. Successful companies are moving quickly to use this technology to create commerce experiences that account for all three contexts – product, customer, and application. They use contextualization to build experiences that are more helpful and more trusted by buyers while also being more differentiating and more profitable for the sellers.
It’s a true “win-win” for B2B.
