5 Steps to Personalization — In Months Not Years
Personalization is all about tailoring the shopping experience for each customer to offer the right products at the right price. But how do you make the right offer when you don’t know who the customer is? Most visitors to a website are anonymous, and you may have no prior knowledge about them. But perhaps even more fundamentally, how can you help your customers (anonymous or otherwise) find the right products in the first place?
This blog outlines 5 steps to starting your on-line/digital personalization journey that can be achieved in months, not years. The grand picture of personalization requires board level buy-in, company wide strategies, and large transformation projects. But by focusing on the start of your customer’s online retail journey, and taking a few innovative approaches, you can achieve valuable results in the near term with no invasive work required.
Step 1 – Begin digitizing your retail knowledge
Many personalization strategies involve capturing large volumes of data (Big Data) in order to extract patterns, which can be used to create insight into your customers. It takes time to gather the information, and specialized data science skills to extract the patterns, but, as experts in your retail vertical, you already possess a wealth of information that can quickly be applied to improving your customer’s online experience.
I’ll cover what this means in more detail in the subsequent four steps, but it’s important to note that your industry knowledge is actually a differentiator! What customers are searching for, or what social media post they’ve responded to, or what items they’ve been browsing, gives you an incredible insights into those individual. This is important because it means you can use that information to personalize their experience, and you don’t even need to know who they are.
Customers have demonstrated a willingness to trade personal details for better experiences and offers. But by demonstrating you understand them immediately (before they give you any personal details), and proving you can use that knowledge to improve their experience, they are far more likely to share further information with you. It’s at that point, that your Big Data strategies can be really maximized.
Step 2 – Drive more traffic to your web site
This is really the start of many of your customers’ online journey: Finding the products they are interested in. The greatest personalization engine in the world is no use unless your customers have found you first.
The two most common methods people use to find new products online are through search engines, and social media.
Search engine optimization is a rich and complex topic, and worthy of a full book – not a tiny section in a blog post. But a simple rule is that adding extra meta-data to your product pages will help search engines pick them for relevancy. However, the more focused and topic-based that information and meta-data is, the higher search engines will rank that page (for a particular set of search phrases). This is the first example of where your industry knowledge comes into effect. By understanding what people are likely to be searching for (and the terms they might use), you help make your product pages highly relevant for key searches.
Social media has quickly become an equally important strategy in driving traffic to websites. Sponsored ads can be a highly effective tool. But like any advertising, knowing who to advertise to, where you can find them, and what messaging to use, requires industry knowledge.
Step 3 – Guide the customer to the right product
Once your customers have found your website, the next step is quickly guiding them to the relevant products. If they’ve come from a sponsored product link they may already have a head-start, but typically users will look to use the website’s search facilities.
Many online retail websites provide navigation trees, but people are trained to use search interfaces. Moreover, search bars offer the potential for the customer to describe in detail what they are looking for – far more richly than a set of static product categories. However, when search doesn’t present the correct products, customers will often assume they don’t exist. Equally important, a search result that returns incorrect results can leave a bad impression and make finding the relevant results harder.
There are two problems most website search functionalities face:
- Product descriptions often vary widely between individual products. Due to the IT processes (and technologies) that are involved in moving supplier data into the customer facing website, a lot of this rich information is often lost or normalized, and thus the website can’t include this data in it’s search. A “schema-agnostic” database with search capabilities, however, can load all product data with little or no effort required, and can ensure all this information can be included in the search, ensuring products won’t be excluded.
- Slightly more challenging, is that search algorithms don’t understand what the terms being entered by the customer mean. If a customer searches for “high-definition TV,” a search algorithm doesn’t know that “high definition” is a phrase that has meaning and shouldn’t be broken down into separate searches for “high” and “definition.” The question is: How do you extract this kind of knowledge you have in order to inform the website’s search results? Using a technology called “semantics,” it’s possible to encode this kind of information and match not only the search phrases, but also the product data. For example, a TV might have an attribute called “1080p,” which could be matched as a synonym for “high definition.”
By choosing a schema-agnostic database with search and semantic support, it is possible to take your product data and start to embed your industry knowledge and start to provide a complete and accurate search. This doesn’t have to replace any existing technology (at least initially), and could just be used as an improved search service for your website.
Step 4 – Offer relevant recommendations
At this point, you now have a lot of context and information about your customers, without needing to know who they are. If they clicked on sponsored ads, or searched for particular terms, they have given you insights into their interests. The best part is that the same knowledge you digitized in the previous steps can be used to turn this insight into smart recommendations.
For example, if a customer followed an ad on Twitter referencing an item of clothing a celebrity recently wore you can leverage that fact to offer other items that are associated to that celebrity. If customers searched for particular items that are in fashion this season, you might be able to offer them other items that are currently in vogue. A search for “high definition TV” might offer an up-sell to a “4K TV”
Again, this is using the semantics approach described in the previous step. Semantics works by allowing you to capture simple facts (or triples, as they’re referred to):
- “HD” – “is also known as” – “high definition”
- “4K” – “is an upgrade to” – “high definition”
Now, a customer who searched for “HD” will not only get all TVs with “high definition” in its description, but can also be offered a up-sell for “4K” TVs.
This creates a highly flexible system that is superior to a rules-based approach (that requires you know know all possible combinations ahead of time) and is more agile than a big data approach (that requires large datasets to be collected first).
Step 5 – Feed personalization insights into your Big Data project
Now that your customer’s actions are now carrying and recording context, you can feed this into your Big Data project to improve its insights, and create a context driven feedback loop.
Moreover, an improved customer experience will increase the probability a customer will start to provide more details about themselves. With this you’ll be able to better identify your high value customers, and correlate this rich contextual information against these important individuals.