Personalisation for SMBs – Types, Technology and Strategy

The landscape of ecommerce personalization tools is rather complex and all the sales reps of personalisation vendors promise you at least 20% sales uplift for no implementation time. On my quest to to shine light on Personalisation it makes sense to outline and differentiate the most important types, technologies and strategies to start a successful personalisation implementation.

Types of Personalisation

There are four key areas of ecommerce personalization tools, each one of the has distinct advantages depending on your strategy, website and resources:

• Versioning tools – different executions of site elements such as the home page to different customer segments: returning vs. new visitor or Google search term referrer vs. traffic from Social Media
• Recommendation (cross-/up-sell) tools – can be simple widgets displaying on a product detail page the most commonly products bought together up to fully integrated solution for email programs, POS systems, or call centres based on complex algorithms (see also technology)
• Filtering solutions – display specific information based on customer’s directive: customers answer questions to narrow down a set of products. Empowers customers to find products that suit their needs and enables web sites to tailor broad
  assortments to customers who may not know what they want.
• User-generated applications – display such as “top rated” or “most recommended” products based on community feedback and actions
• A combination of all above

Types of Recommendations

Typically rather well suited for SMBs are recommendations since the effort/time to implement is lower than for other solutions, hence often provide a higher ROI.

Based on complex algorithms these applications deliver personalisation like:

Item-to-item recommendations – based on relationships between products

• Typically seen as: people who bought / viewed this also liked… (typically collaborative filtering – see Amazon ones)

User–to-item recommendations – based on user characteristics – the visitor’s expressed likes and dislikes, browsing patterns, location etc as well as shopping history

• Typically seen as: “you might also like…” or “based on what we know about you we recommend…”

User-to-user -virtual communities of like minded peers inspire each other with new products. Her the wisdom of the crowds surfaces e.g the long tail of the product catalogue (pretty much the quintessence of right time – right user – right content)

• Typically seen as: people like you recommend… (often referred to as ‘social recommendations’ or discovery)

I could go on forever here but one important thing to bear in mind is that these types of recommendations work differently depending on the page type (e.g. check-out vs. category page) or the stage of the customer purchase cycle and/or the type of goods a retailer is selling.

Technology of a personalisation solution

Personalisation is all about data! The more data points available the better the results are.

Source, Forrester research Inc.

Hence inputting the right data into the system is crucial; data points can be implicitly observed customer behaviour on a web site (a.o., time on site, click stream
behaviour, purchase history) or explicit customer preferences.

Proprietary algorithms are then used to find relationships, commonalities, associations, and cause-and-effect relationships.

Outputs differ from straightforward recommendation widgets to changes in navigational structures of the site. Depending on the scope of the implementation and the the technology used. One thing to consider here is the importance of real-time and intent driven recommendations because of changing user preferences: just because you bought something yesterday, doesn’t mean you want to buy a similar thing today (e.g. I might buy a birthday present for my sister one day, and a new CD for myself the next). See also 5 Problems of Recommender Systems.

For small and medium sized businesses it is advisable to make use of the hosted solutions available (software as a service – SaaS) and if possible of a revenue share pricing model (a fee is only charged if a item was purchased as a result of a click on a recommendation).

Personalisation strategy

Effective personalization strategies are undeniably successful in increasing the bottom line profitability for ecommerce sites. However, it is important to make sure that  privacy issues are also addressed as well as the effective handling of usability and relevance issues.

For SMBs to start planning a personalisation strategy they must first answer a range of over-arching questions in order to develop and prioritize their business strategy to support their personalization goals.

What is the degree of internal resources available? Different Personalisation technologies have different requirements on maintenance.

What is my depth of assortment? Personalization works best when you either have a large number of different types of products (and plenty of long tail products) and services to sell, or your customers come from many different walks of life.

What do I want to accomplish in terms of marketing? E.g. what is most important: customer acquisition or customer retention? What types of customers have the highest value: frequent shoppers or biggest spenders etc…

When it comes to choosing the technology the following is of importance to consider:

• The amount of data points available
• Is sufficient reporting delivered
• Is the solution a black box or can I adjust it to my marketing strategy
• etc…

For a complete and comprehensive discussion in terms of strategy, technology, implementation and vendor selection please fill out this formfor a free consultation or just give me a call. Click here for more internet marketing münchen