WIXLIB

RETAILEXPERIMENTATIONTHE NEW FUNDAMENTALSBig changes in retail require a new approach tocustomer-focused experimentation. Here’s howto make it work.Start reading

ABOUT THIS E-BOOKALIGNMENTACCESSAGILITY AND AUTOMATIONFacing growing pressure to compete for customers on the quality of their customer experience, many retailers are takingan even more rigorous, scientific approach to improving that experience through experimentation, focusing on threecritical test-and-learn principles: AlignmentAccessAgility and automationIn this e-book, we’ll show how Accenture Applied Intelligence, using SAS and Intel capabilities, is helping retailers mastereach of these principles on their way to achieving competitive differentiation through an advanced ability to experiment.Read more:NEW APPROACH1 THETO EXPERIMENTATION2 ALIGNMENT3 ACCESS4AGILITY AND AUTOMATION5 NEXT STEPS

THE NEW APPROACH TO EXPERIMENTATION IN RETAILFrom operations to core customer strategyBlame it on fickle customers, lightning-fast shifts in trends, and cutthroat competition: It’s a constant, day-to-day, quarter-to-quarter battle for retailers to remain relevantand profitable. At every moment, they must find new ways to stay one step ahead.Retailers have made big investments in major strategic initiatives suchas innovation and digital transformation, reengineering their approachto customer engagement, sales, marketing, product development, youname it. That’s also why retail organizations today bear little resemblanceto their former selves, from ten or even five years ago. Their survivaldepends on their ability to continuously adapt and change.One constant throughout all this change has been experimentation,which has played a central role in retail operations from the start.“ What happens if we change this productin this way? How will customers respondto this new product? What is the effect onsales if we change our store layout? Thisseasonal promotion worked in one region– how will it work in another? ”Asking – and answering – questions like these is in the DNA of successfulretailers. It’s how they continue to deliver new products that meet theneeds of their customers, and how they drive ongoing innovation toimprove existing products.NEW APPROACH1 THETO EXPERIMENTATION2 ALIGNMENTRead more:Faster, more efficient approach3 ACCESS4AGILITY AND AUTOMATION5 NEXT STEPS

There is nothing shiny or new about experimentationin retail. It’s just what retailers do, and have doneforever. This is exactly why today, many takeexperimentation for granted, treating it as a secondtier strategic focus rather than the top-level strategicimperative it should be. What’s the difference? Ifexperimentation were approached as a top-levelpriority, just as digital transformation and innovationare in many organizations, it would land on theagenda of every strategy meeting, it would be builtinto performance assessments, and it would be widelyunderstood to be a growth driver. More importantly,retailers would be able to improve and refine moreaspects of their businesses more quickly – not justreduced cycle times in product innovation, butimprovements in areas such as customer experience.With a better, more efficient approach to experimentation,brands can improve the customer experience byconsistently developing the promotions, products andexperiences their consumers are seeking – faster and more profitably.NEW APPROACH1 THETO EXPERIMENTATION2 ALIGNMENT3 ACCESS4AGILITY AND AUTOMATION5 NEXT STEPS

ALIGNMENTSuccessful experimentation strategies in retail – not just individualexperiments, but the entire portfolio of experiments underway atany given time – require a level of consistency and alignment acrossexperiments that can be difficult to achieve. In practice, that meansdifferent experiment groups across the organization should useidentical data types, and should apply the same set of standardsand processes for analyzing the data, in order to achieve “applesto-apples” interpretations of the results. This is most easilyaccomplished with a single solution that is shared acrossexperiments – the same tool used by Marketing, Innovation,Store Operations, and so on. This makes it easier for the organizationto align on goals, processes, and handoffs, from test design throughtest execution, measurement of the results, and data analysis.Similarly, Accenture has helped this retailer use the same solution totest new products, using preset attributes to generate the optimallist of test stores and control groups. This approach not only reducesthe time and effort required to run experiments but also lends a levelof rigor, consistency, and alignment to the experiments that result inmore objective, scientific, trustworthy experiments.This end to end alignment results in tailored customer engagementsto specific audiences, e.g., based on brand sentiment, geo-location,preferred shopping category or current loyalty tier. Retailers canoptimize each potential interaction and decision point in the buyingjourney – allowing brands to improve the customer experience andreduce costs.This is the approach that Accenture Applied Intelligence uses whenworking with a leading fashion retailer to establish control groupsfor store operations-focused experiments. When the companywanted to find out how moving a prominent fixture in stores wouldaffect sales, rather than hand-selecting stores in which to run theexperiment, as well as the control group, it used a centralizedexperimentation solution to generate the store groups based ona set of attributes established by the team running the experiment.1THE NEW APPROACHTO EXPERIMENTATION2 ALIGNMENT3 ACCESS4AGILITY AND AUTOMATION5 NEXT STEPS

ACCESSRetail experimentation thrives when all stakeholders have readyaccess to experiment-related data. That’s how you improveexperiments, how they are measured and interpreted, how mucheffort they require, how many you can run simultaneously, andthe amount of time it takes to successfully launch and run them.In many retail organizations today, access to data is managed onan experiment-to-experiment basis, leading to lots of duplicatedeffort. Just as important, this type of decentralized approach rulesout the opportunity to share, integrate, and analyze data acrossexperiments, which holds the potential to uncover unexpectedinsights and patterns.A better approach is to create a central repository of experimentdata – one that applies a consistent set of data managementstandards and processes – to allow different users to tap into thedata as needed, on demand. Role-based user profiles determinewhich information is available to individual users based on theirneeds, preferences, and roles. This is less about restricting accessto data than it is about ensuring the steady delivery of relevantdata to users so that it is useful in practice, recognizing that C-levelexecutives need access to different types of data comparedwith their counterparts who are marketing analysts, or midlevelmanagers.When the delivery of stakeholder-specific insights is part of the plan,these individual stakeholders are poised to make more informed decisionsregarding their experiments – in a more connected, coordinated way thatcan add up to a bigger impact for the retailer. This also contributes morebroadly to the culture of empirical, evidence-based decision making thatretailers have been seeking to cultivate for years, moving beyond gutdecisions to those informed by objective, data-driven insights.“ Retail experimentation thrives when all stakeholdershave ready access to experimental-related data.”When Accenture Applied Intelligence works with retailers todevelop this type of central data depository, it uses a single SASbased, Intel-powered solution to give them a centralized view acrossall tests. In site testing, this allows the leaders of experiments tosee (and eliminate) store overlaps, to avoid competing tests acrossstores. Similarly, leaders can see whether customer segments arebeing triple-treated with tests. Without such a centralized view, thesetypes of important details may be uncovered long after they areuseful, or not at all.1THE NEW APPROACHTO EXPERIMENTATION2 ALIGNMENT3 ACCESS4AGILITY AND AUTOMATION5 NEXT STEPS

AGILITY AND AUTOMATION“How quickly can we see results from this experiment? ” Retailersask “how quickly can we do [X]? ” across their businesses every day,with good reason – lightning-fast shifts in consumer preferences andcompetitive pressures create new opportunities (and perils) thatdecision makers must respond to just as quickly. This type of agilityshould be required of all test-and-learn experiments, as well, tomaximize and accelerate their impact.Automation technologies play a large and growing role in achievinggreater speed and agility in retail experimentation. Retailers relyingon heavily manual approaches to experimentation find it moredifficult to remain competitive against peers who are adoptingautomation principles. Analytics capabilities just like those usedby retailers to manage their relationships with customers, analyzeand improve campaigns, and more, are equally valuable inexperimentation – constantly updated cloud-based data collectionand analysis that allow retailers to model, execute, iterate, andimprove their tests at the speed of retail.Retailers relying on heavily manual approachesto experimentation find it more difficult toremain competitive against peers who areadopting automation principles.When Accenture Applied Intelligence recently worked with a leadingretailer on marketing and promotions-focused experiments, its centralizedsolution allowed the retailer to automatically draw data from a range ofinternal and external data sources from the cloud. In-store promotions datasupplied by one department automatically populated the solution as newdata appeared, alongside related data from an external email marketingprovider, as well as relevant data from web-based marketing tools. Withouta centralized solution to do the heavy lifting, this would have been a highlymanual process – conducted in three different silos.Automation also enables many more data inputs, delivered morefrequently, into the experimentation process. Rather than having aweekly meeting to review updates on experiment data, imagineseeing a steady stream of this type of data populating a dashboardas it is generated, for example. This has far-reaching implications,allowing for more iterative work plans, accelerated release cadences,and a sustained emphasis on continuous learning across experiments.1THE NEW APPROACHTO EXPERIMENTATION2ALIGNMENT3ACCESS4AGILITY AND AUTOMATION5 NEXT STEPS