Join our newsletter
#noSpamWePromise
Oops! Something went wrong while submitting the form.
Actionable retail analytics case studies that drive growth.
Retail analytics is transforming the bricks and mortar and e-commerce landscapes.
From Amazon drones delivering your favorite cupcake the moment your sweet tooth starts to tingle to your local shop stocking the new GoPro just before you set up on a new adventure.
In this article, we will explore the guiding principles of how data can be used to improve your retail business.
But we will also make it actionable.
By diving into real-life case studies, we’ll show how the principles can be applied to your retail business to drive growth.
Retail data analytics is the process of collecting retail data (Point of Sale or POS, stock and inventory, market trends, customer feedback, …) and transforming it into insights - trends, predictions, and other deliverables that help you make better business decisions for your retail business.
When used correctly, retail analytics can help you acquire more customers, stock the best-selling products, improve your vendor network and drive higher revenues and profits for your store.
There are four crucial steps to make retail analytics work:
Are you seeing a trend in customer product returns that needs turning around? Do you need to bring more clients into your shop? What aspect of your retail business do you want to make better?
The more, the merrier. Usually, retailers do not collect all the data they already produce in their stores. Such as Points of Sale (POS) and e-commerce transaction data, customer feedback, in-store (physical or web) browsing behavior, etc. But also make sure to collect additional data that can help you answer your business question. For example, if you would like to predict what clothes are going to sell better next weekend, you might want to collect weather data, to build a model of winter vs summer clothes that need restocking.
This step might require a bit of technical know-how. Model your data via data engineering practices, to have all the data centralized in one database. Apply advanced statistics and analytic techniques, to get the trend and predictive insights that can help you answer your business questions.
Knowing what to do wins you half the battle. Executing it is the other half.
What kind of benefits can you expect after implementing retail analytics?
Retail analytics can play a vital role in elevating efficiencies in everyday business management. And when successfully managed, you can keep track of everything - from stock supply to store behavior, to market trends and demands.
But unless you have clear baselines of how you are performing right now, you cannot build advanced models to predict if the customer acquisition is increasing or decreasing. Without a fully integrated product catalog with warehousing data, you cannot anticipate if you are going to run out of skis just before the winter season.
The first step is to collect and validate all data into a single source of truth.
Such collected data can be used to build business intelligence reports - anything from simple Excel graphs showing stock data by product SKU to complex BI dashboards in Power BI.
Ascend Climbing is primarily a two-location climbing business offering vertical fun to 16.000+ climbers. Covid hobbled the company’s business model, so Ascend Climbing responded by expanding its services to offer e-commerce products, such as merchandise, gift cards, virtual fitness training sessions, and online yoga classes.
With the rapid expansion and digitalization of the retail business, they faced a dire challenge. As a one-man-analytics-band with the help of an external developer, they had to take care of all the retail business analytics needs.
They spent hours on manual data tasks every week - joining data sets from disparate data sources (Google Ads, Facebook Ads, Emailing Software, Transaction data …) into Google Spreadsheets where 99% of the business intelligence reporting took place.
There was no single source of truth, and they needed to understand the big-picture metrics and the KPIs behind customer behavior.
They decided to deploy Keboola to automate their ETL processes across the business, and with the use of numerous pre-built extractors and the Generic Extractor, they were able to collect data from different data sources at scale resulting in improved email validation. They were also able to connect the payroll and check-in data with other sources, to create a single source of truth.
In a matter of days, they were able to set up data pipelines that automatically collected all the relevant data from all the relevant data sources and better understand their e-commerce and physical shop performance.
With fast insights, they were able to determine which advertising channels bring more customers, what products perform better on the e-commerce, and other quick wins for improving their business efficiencies.
Retail data is not just a foundation for building business intelligence. It can also be used to drive growth.
Sure, you can use BI to gather insights that improve efficiency and therefore the bottom line (“oh, that product is really selling just in one location, no need to restock it globally!”).
But data can be used to inform your marketing strategies. And it can also be piped directly into advertising software to lower your advertising costs and increase your marketing ROI.
By implementing Keboola as the data engineering and analytics solution, Ascend Climbing was able to collect and refine more customer data. But the data was not just used for business intelligence.
Joining data from disparate data sources gave them the power to build a Customer Data Platform (CDP) from the customer data they collected and owned themselves.
This data was then piped back into advertising software to build custom ad audiences and remarketing campaigns that were enriched with data to allow a more laser-focused targeting on the best performing customer segment and had a more tailored messaging to perform better across social media.
The move from “spray-and-pray” targeting across broad demographics to advertising tied to customer behavior and enriched data improved Ascend Climbing’s ROAS across marketing channels significantly.
Inventory management is a juggling act.
If your inventory levels are too low for a product in demand, you miss an opportunity to sell. If you keep too much of a product on stock, and it does not sell, you are eating away warehousing resources from other products that could actually move the bottom line.
Retail analytics is a great tool to anticipate future demand and optimize your supply chain of vendors to meet the forthcoming surge.
With the use of historical customer data, seasonality effects, market trends, and customer behavior, you can build models that predict your stock needs for the near and far future.
Let’s look at an example of how this is done in practice.
Sportissimo is a CEE-based sports retail business selling sporting goods across 5 countries, covering 190+ physical stores and multiple digital e-commerces.
They were facing a constant challenge of how to optimize stocking and manage the supply chain. Each locality (including the digital stores) has a different set of customers, driven in their shopping habits by different patterns of demand.
For example, the weekend forecast predicts a winter wonderland in one location, so snowboards are flying off the shelves faster than they can be restocked. Another location, on the other hand, is barraged by customers demanding cycling equipment for the sunny weekend ahead.
Sportissimo used Keboola to collect data from all its various data sources in near real-time: point of sale (POS) terminals in physical shops, weather data from various APIs, online retail data analytics tracking consumer behavior on the e-commerce sites (search, products added to cart, …), and a dozen more.
Keboola joins the current and predicted data with historical data points collected from customer loyalty programs and past sales data.
Together, the data is used to build machine learning algorithms that make over 12 million predictions daily - one for every product in every location, estimating how much will be sold per day for the next fifteen days.
This allows adjusting stock dynamically, always ordering items that will soon be sold out without keeping on inventory products that would collect dust.
Usually, we receive customer complaints once it’s too late. The customer already decided not to do business with us.
With the multichannel approach of digital retailers, it is hard to keep track of every touch point. Customers give us feedback by replying to marketing emails, opening technical support tickets, writing reviews of our products, sharing their (dis)satisfaction with their friends, and finally, by voting with their hard-earned dollars.
But when you operate at scale, it is hard to keep track of all these touchpoints. It gets even harder to not mistake the trees for the forest. The amount of customer feedback collected via different sources can quickly turn overwhelming.
Retail analytics helps us collect data from multiple sources fast. But also build composite metrics that cut through the noise, and give us quick insights into customer (dis)satisfaction.
Once we understand what makes the customer happy, we can increase that operation to delight them further. If we measure and monitor customer satisfaction metrics, we can quickly discover when a fire is starting and take it off before it burns out more customers.
So how to go beyond email surveying and build a retail analytics machine that churns out actionable customer satisfaction insights?
Delivery Hero is the largest delivery company in the Czech Republic and the wider European area, reshaping the world with quick-commerce or q-commerce.
With over half a million vendors using their platform, and a high-paced order demand, Delivery Hero is also in the business of using predictive analytics to improve their routing and product. But the biggest challenge is guaranteeing a strong customer experience across the varied customer journeys.
With every vendor being slightly different, Delivery Hero needed to collect a wide range of metrics to measure customer satisfaction: from the conversion rate on the website and app to the NPS score, satisfaction ratings of the ordered items, speed of delivery, packaging, and a dozen more.
This is a huge engineering challenge. But also a challenge for retail analytics. Putting a number on customer satisfaction helps evaluate which vendor partners need assistance and communicate potential issues with them before the delivery problems cause them to lose business and their reputation.
By using Keboola’s ETL, Delivery Hero collected over 60 customer satisfaction metrics from the entire customer journey - from NPS to delivery speed ratings. Using Keboola’s Transformations, Delivery Hero joined the 60 metrics into one. A single metric that helps them gauge the overall customer satisfaction on a scale from 1 to 5.
The metric is widely used by both vendor partners and account managers, to more easily spot when something goes wrong. The average rating suddenly fell under 4? Better check what caused the sudden fall - is it speed delivery? The niceness of the waiters? Food taste? The single metric acts as an early warning signal and allows everyone involved to dig into the other 60 to quickly determine the cause of issues.
Take a look at your e-commerce and physical stores. What is the crucial business information you need to run it better? Write your questions down. These will form the backbone upon which you can build your retail analytics.
Collect all the internal (SOP, transactions, customer feedback ..) and external (market analyses, weather forecast, …) data into a single source of truth. Use tools like Keboola (try it out, it has an always-free tier) to automate the heavy lifting. No need for developers or data engineers.
Analyze the data you collected in the previous step to answer your most pertinent business question.
Your analysis can be as simple as joining data across spreadsheets to gauge the overall business trends, or as complex as building a machine learning algorithm from scratch.
Apply the insights from step 3 to your retail store today.
Contact us to schedule a chat about how Keboola can support your analytics journey.
