Home / Analysis / eCommerce  /  Moving Beyond 2% Conversion Rates: Introducing Big Data

Share This Post

Analysis / eCommerce

Moving Beyond 2% Conversion Rates: Introducing Big Data

Big Data

E-commerce accounts for around 6% of the entire retail industry, which is now at $2 trillion and growing steadily. Yet, this is tough ground, where each fraction of a second can make a difference between a sale and an abandoned cart. Big Data is the latest trend and is seen by some as the Holy Grail of conversion. Although it can make a huge difference, it’s still out of reach for companies not due to costs, as one might think, but due to insufficient understanding.

Conversion Rates Statistics

To measure the difference Big Data can make, an organisation needs to measure its initial conversion rate and set realistic goals. A recent report shows that conversion rates vary across industries, from around 2% for retail to 5% in travel and 10% for finance.

The same study shows that usually, visitors convert to paying customers on their third visit and the conversion rate does not depend so much on the traffic of the website since sites under 100K visitors, and those over one million, both have around 2-3% conversion rates. When it comes to the device used, there is a clear relationship between screen size and conversions, with a higher percentage in favour of desktops (4.3%) and a lower one for smartphones (1.5%).

Improving Search Engines Through Big Data

In this case, each e-shop owner can ask what is to be done? There are several ways to move the conversion gauge, at least marginally, with the help of Big Data. These include a customisation of the offer by looking at the user’s profile, automatising search result listings for a better usage of the space provided by each device, and closing the gap between what the client is looking for and what the website can provide.

Personalisation

Selling more boils down to the simple act of providing the customer with what they need, when they need it and at the right price. Big Data can help answer these questions by providing context for the search. Just focusing on the query gives a limited perspective, as seeing a picture crop. You need to see more to answer it correctly and comprehensively.

First, a valuable piece of advice is to abandon generalisations, statistics, and analytics and use data in its raw form, as entry logs, to capture the real interaction of the user. What words did they use, at what time, how long did they spend on page, did they add anything to the cart just to abandon it a few seconds later? It’s all about understanding micro-gestures to work back towards motivations and triggers.

By identifying the visitor’s characteristics, they can be compared to similar clients and served in a more tailor-made way. Also, by gathering data from the visitor’s own browser, they are not just thrown in their respective demographic bin but are treated as an individual, increasing the chances of revenue.

Real-Estate Usage

As seen before, screen size has a huge influence on the conversion rate, therefore the smaller the real-estate available, the more relevant the result displayed needs to be. Big data needs to be used to highlight those products with the highest estimated conversion rates, and display them first. The main problem here is to avoid generic top-performing products and dig deeper into the long-tail options that could be relevant for the particular user.

By comparing previous interactions of the visitor with existing inventory, an appropriate list can be created on the spot and displayed in descending order. To maximize revenue, a mathematical model will create a likelihood/income matrix and list on top not only the most relevant, but also the highest margin products. Even smaller companies can benefit from this groundbreaking technology through a pay per use system like big data analytics services from InData Labs.

Bridging The Language Gap

If you showed the same picture to 10 different people, most likely they will describe it in 10 different ways, and only a handful of their words will be the same. The problem faced by e-commerce search engines is that they must understand millions of queries and match them with appropriate products.

By incorporating natural language processing built on Big Data, you can transform your search box into a search assistant. To improve the process, the raw data retrieved from real customer searches should be matched by a human with the most appropriate product and the resulting tables used as training tools for the deep learning algorithms.

Not only different words denoting the same object should be considered, but different ways of spelling a word and even incorrect variants. Adding a real-time spell checker to your search can save customers the ordeal of seeing 0 results.

Dealing With Abandonment

Every abandoned query or even an abandoned cart means money left on the table. Studies show that more than 80% of discarded questions are never investigated. Big Data has the power to analyse each of these interactions automatically and provide an answer to the behavioural path and a possible reason for abandonment.

In the case of the abandoned cart, the most common strategy is re-targeting the user on other channels such as social media or e-mail. Before doing so, it would be useful to know the reason for the half-purchase to address that first. Most abandoned carts would turn into a purchase with an appropriate incentive: free shipping, easier checkout or a discount.

Limitations Of Big Data For Conversions

There are several limitations to using Big Data which should be addressed. Since the models rely on probabilities and past experiences, although they have predictive power, are by no means an accurate description of the future, since they don’t account for outliers.

Focusing too much on automated responses can take the human factor out of commerce and frustrate the client. Even if you have in place the best chatbot systems, automatic responders, and deep-learning algorithms, at the end of the day, don’t forget to offer the client the option to talk to a human operator for a situation that was not built in the system.

Share This Post

Jasmine Morgan is a senior BI/DW specialist with 8+ years in software consulting. Since 2013, her advanced focus is IT solutions for financial sector.

Skip to toolbar