Data science is becoming widely used by companies of all sizes, from local SMBs right up to huge corporations such as Amazon. But what exactly can retailers gain from this new technology? Product recommender systems are often the first thought but curating better product suggestions isn’t the only way to use data science in retail. Below are four other use cases worth exploring if you want to maximise conversions and generate more sales.
1. Data-driven price management
You may love ‘happy hours’ and hate Uber price surging, but both practices share the same baseline idea – bump up the profits depending on the market conditions. Can retail companies get as good at attracting customers during ‘off hours’ and capitalising on high demand when the need is there? Yes, if they know how to piece their data together.
According to Deloitte, price management initiatives can boost profit margins by 2%-7% in just 12 months, generating an ROI of 200-350% on average. But few retail companies are actually taking advantage of this opportunity, mainly due to:
Low data maturity and analytics culture
Lack of visibility into all of the channels, product portfolios and customer segments.
Both issues are relatively easy-to-fix if you have a data science team and, once your data is prepped for analysis, you can choose to experiment with several price management strategies:
i) Personalise your discount/pricing strategy
Data science allows you to map similar customers into clusters based on their past behaviours and determine the ultimate price/discount combo that will make them convert.
ii) Create segmented pricing
If you are not ready to go granular and personalise prices on a per-user level, you can still adjust your prices and offerings to cater to different audience segments. For example:
Value pricing – pitch a coupon or an extra discount to the bargain shoppers if they are buying at a convenient time (for example, they want winter boots in summer) or look at old inventory you need to clear out.
Standard pricing – pitched to the majority of your buyers.
Premium pricing – sweeten the deal for the premium-tolerant audience segment with an extra perk such as extended warranty (or another offer they are likely to respond to).
iii) Offer competitive real-time prices
Comparison shopping is at a peak with 87% of customers shopping on Amazon checking the price against the brands/retailer website. Considering how good Amazon is with price surging, manually benchmarking your prices with the competition is no longer viable. But with the help of data science and predictive analytics, you can create an advanced system that will help you automatically adjust prices depending on market conditions and competitors’ moves.
2. Data-driven attribution modelling
Conversions still remain a sore spot for retail companies. According to Wolfgang Digital’s new E-commerce 2019 KPI report, the average conversion rate in the EU retail sector is a meger 1.7% – the UK scores top.
And yet, despite the relatively low benchmarks, most companies still focus on traffic generation versus conversion optimisation. It may seem that playing the numbers game is a good strategy (more traffic = more sales) but you can actually get more with less by drilling down deeper into your analytics data and identifying which channels bring the best ROI and secure the highest conversions.
Data-Driven Attribution Modelling – a custom model mapping conversions and sales to respective touchpoints/channels in a customer’s journey – can help you with that. This model shows what kind of marketing sequences lead to the most sales, what creative assets play a major role in the process, and why some customers did not end up converting.
The kind of insights you can get include:
Determining whether Facebook retargeting ads or local inventory ads contribute more towards conversions.
Pinpointing campaigns that aren’t yielding any ROI.
Scooping the affiliates who increase the probability of conversions.
3. Intelligent cross-sells and up-sells
Effectively cross-selling even a small-ticket item can lead to massive profit increases. One furniture retailer decided to pitch an item worth 6% of their average order value (AOV) to shoppers. In just 41 days post-implementation, they had increased AOV by 4.6% and secured an additional $180,000 (£142,400.00) in monthly revenue.
Using data science in retail can help you increase profits without running numerous A/B tests. And you can even pitch personalised offers to different customer segments to further pump up conversions and sales. Adding predictive analytics to the mix will give you an extra edge: you will be able to see exactly when to upsell/cross-sell to meet your business goals.
For instance, you can create an algorithm to identify the key value items (KVIs) and key value categories (KVCs) that are making a big difference to your bottom lines and pitch them accordingly to different shoppers. Those can be:
Perceived value drivers: products that remain popular with all customer cohorts for a long period of time.
Traffic drivers: high-demand products that are purchased in high-volume all the time or short-term demand products that fly off the shelves (e.g. a trending accessory).
Basket drivers: items that are often purchased together with other products, such as an air mattress and a foot pump.
Assortment perception drivers: products that are likely to prompt the shopper to get related items in the store (e.g. a matching tie for a shirt).
4. Customer Lifetime Value Modelling
Determining who your most profitable and loyal customers are is relatively easy these days. But, traditional analytics fails to tell you when those shoppers are starting to purchase with less frequency, what leads to that and why they switch to a competitor altogether.
Data science can help you explore those root causes. You can identify the dependencies between different customers’ choices/behaviours and apply that data to predict their future actions. Here are a few examples of what you can achieve with CLV modelling:
Order attribution: Learn which marketing channels bring in the most loyal customers; and what campaigns contribute most to repeat purchases.
Cost of acquisition vs lifetime value: Identify the areas for trimming costs when targeting different demographics with high LTV potential.
Optimised retention offers. Reach out with the right pitch at the right time to re-activate individual shoppers.
Add more purchasable items to the mix. Use data to increase the selection of purchasable items in your inventory by knowing exactly what your shoppers need.
If you feel that your marketing budgets are growing, but your sales numbers stand still, exploring how you can use data science in retail may be the right option for you. Yes, it may sound like a lofty investment, but the ROI is definitely there.