Merchandising in the era of artificial intelligence (AI)

Introducing the Modern Merchandiser

The digital merchandiser’s role has evolved over the last couple of decades beyond just promoting products online. The merchandiser of today shapes customers’ perceptions and guides their purchasing decisions. Modern digital merchandising teams are usually dedicated to specific sectors of an ecommerce store, but they collaborate with a vast network of other stakeholders including buying, marketing, and trading in order to reach the wider organization's retail sales goals.

Operating in a more challenging retail environment

During this period of economic uncertainty, companies are increasingly conscious of costs. In order to minimize expenditures and maximize ROI, businesses will gravitate towards automation, data-driven processes, and AI. In order to accommodate these new processes and technologies, merchandisers will need to add new functionality to their teams – in addition to their already considerable daily responsibilities. Merchandising teams will be expected to demonstrate their impact and be under pressure to deliver more value with less resources.

Merchandising teams can no longer expect to drive engagement and revenue solely through sales and offers.

Ecommerce businesses need to adapt their overarching strategies to effectively integrate AI into their workflow and processes. Using AI successfully in a visual merchandising “arts” strategy will require an assertive blend of AI and human intervention.

Leveling up for the new NOW

On top of essential market, product, and customer knowledge – merchandisers will also need to understand how AI models work.

A basic understanding of AI is required to explain AI behavior to stakeholders, build the necessary workflows, make corrections where needed, and incorporate AI input into existing workflows.

On the brink of seismic change

It’s likely that changes in skills and processes will generate resistance within some existing merchandising teams. To alleviate workers' concerns about being replaced by AI, companies will need to align their merchandising team to the overall organization’s KPIs and strategic business initiatives as well as provide the AI tools required to achieve their retail goals. This approach guarantees that the merchandising team will view the transformation and its subsequent benefits in a positive light.

Further, fostering a culture of testing, measuring, and iterating will be key. AI is unlikely to replace the creative elements of merchandising but rather enhance them by automating and streamlining the data-driven processes. It allows merchandisers to devote more time to high-impact activity such as human-curated collections or campaigns that emphasize the visual brand experience.

Expanding what’s possible

AI enables retailers to actively merchandise using a much broader array of pages and collections – with far greater responsiveness.

Algolia customer Steve Madden recently experimented with AI to generate a series of capsule collections based on the colors of college football teams and used them for locally-targeted marketing campaigns.

Without the AI, the creation and management of such a large volume of customized collection pages at this level of granular segmentation would be unscalable.

A balanced alliance

Integrating AI requires a balanced approach that automates certain tasks, while allowing human intervention when needed. In the future, we may even see new AI/human collaboration methods such as using a conversational interface to create AI generated collections micro-targeted to highly-specific groups.

For example, a merchandiser will instantly set up a collection by verbally asking the AI, “Algolia, make me a page of budget equipment for someone who is looking to take up fishing, prioritize products with low returns rates and high margins, and optimize the page for conversion rate among first-time customers.”

Latest isn’t always greatest

In digital merchandising, it’s a common assumption that you always want new products at the top of your results. A large sportswear retailer ran a sales analysis of new products during the first 14 days after launch and found that:

A small segment of highly engaged customers accounted for the majority of the sales of newly launched products.
Almost all sales of brand new products either came from the "new in [category]" pages or from targeted landing pages.
Product-type pages like “men's t-shirts” or “women's leggings,” drove very little sales of new products. In fact, burying proven best sellers beneath new products reduced the overall performance of these pages.

Data-driven evolution

Merchandising teams often commit considerable effort to manually curating pages and collections in response to external factors such as stock positions, seasonality, or requests from stakeholders.

For example, merchandisers commonly promote a product with the goal of moving overstocked products. But once the inventory is liquidated, the team often moves to the next “crisis” without a chance to review the analytics reports or the insights that inform future decisions such as:

Did the product sell due to the promotion?
Were there other factors that contributed such as a social media trend or paid advertising?
Was there an impact on sales of other products as a result of the promotion?

In the traditional model, these questions are seldom formally raised and rarely answered. AI-enabled merchandising teams using automated functionality can more easily assess the impact of their curation decisions retrospectively. Through ongoing data analysis, teams can generate valuable insights to continuously improve their planning, execution, and optimization – often in real time, thereby effectively “closing the loop” between merchandising efforts and revenue generation.

Not all data sources are created equal

In an ideal world, ecommerce businesses would have a constant influx of high-quality fresh data to drive their merchandising strategies. The reality is most merchandising teams presently rely on a patchwork of spreadsheets and reports – often generated from various systems built originally for non-merchandising purposes like web analytics, stock management, or financial reporting.

The bygone days of spreadsheets

With AI-powered merchandising, manual tweaking ranking order on product detail pages will soon be a thing of the past.

Instead of updating and reviewing spreadsheets, AI will collect a wider range of data through events and use it to automatically re-rank products with user-set rules and relevance algorithms.

With AI handling the calculations, merchandisers will have the freedom to experiment and test new formulas in real-time. Merchandising teams can then corroborate formulas using data-driven metrics.

Anchor on reliable frameworks not ‘secret sauces’

Even where good data is available, teams often use it to validate their “educated guesses” rather than design a consistent data-driven operational strategy. They prioritize products with their own internal scoring recipe or “secret sauce” to blend and weight data. Initially, this approach might prove effective, but over time performance will degrade as the secret sauce is passed down from legacy teams to new employees who may not understand how it works or how to modify it.

Additionally, the positioning of products is often updated infrequently and the data and weightings are applied globally rather than at the level of a specific category, customer segment, or individual customer. Because of the resources required – tools, people, skills and volume of traffic – detailed positioning can be cost-prohibitive to all but the largest retailers.

Iterate for maximum optimization

Because shopping behaviors change over time, visual merchandisers need to continually improve their ranking strategies to ensure they’re consistently aligned with the changing dynamics of the market and the business. AI-driven tooling makes it easier to constantly test and adjust winning formulas.

For example, a retailer has a goal of pushing all winter coats out the door before the season ends in a month. This would require the merchandiser to forecast the sell-through rate and periodically adjust the promotion or pricing over the next few weeks. AI takes the workload of these mechanical tasks off the merchandiser. It can continuously analyze, re-forecast, and automatically adjust in real-time throughout the entire month until the last jacket is sold on the last day of the season at the highest possible price.

Manual tasks don’t scale well

The manual nature of current merchandising methods, even in the most data-focused teams, becomes challenging when scaling operations. Companies with extensive catalogs, many collections, and multiple pages will need to either grow a large team or limit activity to their high-value areas.

There tend to be very few shared best practices in digital merchandising. Most practices that do exist have been carried over from in-store merchandising, especially with established brick and mortar retailers, or are centered around visual appeal and aesthetics, particularly in the fashion retail space.

Data is liberating

Fed with a consistent stream of fresh data, AI tools can adapt merchandising strategies that react in real-time and respond to variables such as fluctuations in buyer behavior, stock, and pricing.

If these AI capabilities are properly integrated, merchandisers will have the opportunity to focus on broader trends and leverage their market knowledge to capitalize on emerging opportunities. Instead of working at the level of individual products and pages, merchandisers will be free to set objectives, refine aesthetics, and enhance the customer experience. The balance between art and science will remain, but the repetitive and mechanical tasks of the science side will increasingly become more automated and driven by AI.

Less pinning, more AI automation

Merchandisers using AI-powered tools will rely less on rules to reach specific goals. Pinning will be used sparingly to override the AI only in special situations such as a limited time promotion of a new brand launch or sponsored product placement.

Instead of intervening at the level of individual products, merchandisers will define goals and objectives at a broader strategic level. For example, a merchandiser may command the AI to “adjust the ranking to promote the knitwear collections to clear down discounted stock, prioritize gross margin on outerwear, and give increased exposure to new dresses over the weekend.”

Building a data-driven merchandising strategy

All merchandisers want an end-to-end merchandising strategy, but it takes time and resources to get it right. Creating a plan that fits an individual business’s unique circumstances requires an iterative approach – testing and tweaking to make gradual improvements.

Predicting customer behavior with AI

AI can analyze historical sales data to more accurately forecast the performance of new products before they are purchased to help determine optimal quantities and price points. AI can also better understand customer intent as they browse your site and tailor their search results – thereby improving conversions and sales. Additionally, revenue analytics data can then be used by the B&M team to accurately predict stock that drives the highest revenue.

3 ways AI is transforming merchandising teams

Automation will allow merchandising teams to be less compartmentalized by category
Instead of team members specializing in a single niche area, merchandisers will have broader ownership of categories. Businesses with larger catalogs will be able to optimize across all areas of their digital retail space, not just the top level and high traffic areas.
Automation will give merchandising teams more time to strategize and iterate
With the AI doing most of the manual work, merchandisers will concentrate more on the broader buying and merchandising (B&M) strategy and data-led projects. Analytical skills will become more important along with the ability to prompt AI tools for optimal outputs.
Merchandising teams will focus more on identifying broader trends or wider campaign-led strategies
With AI ranking automation, merchandising teams will only need to make small adjustments to finesse the ranking. Team roles will become more proactive and connected to the latest social and cultural trends. Merchandisers will step in to steer ranking and create pin/boost rules with an expiry time to give trending products greater exposure. The tooling will then be able to read the signal and naturally boost or bury the product depending on its performance.

Integration: The promise and challenge of AI

AI is evolving rapidly with new technology being launched almost every day. In the next few years, retailers will be bombarded with a myriad of AI product choices from both new vendors and established corporations.

These AI tools will need to be integrated into existing ecommerce tech stacks. Retailers using all-in-one ecommerce platforms will have limited choices but a simpler implementation; those with fully composable tech stacks will have access to a wider array of solutions including MACH (microservices, API-first, cloud-native, and headless) offerings.

Most AI tools require reliable data to work effectively, such as customer, products, stock, transactions, and clickstream analytics. Other data valuable to the AI may include returns and competitor pricing.

These data sets frequently reside within organizational silos not easily accessible to third-party systems. Retailers with existing data warehouses and event driven architecture will be at an advantage. While others will need to define clear data pipelines and architecture prior to integrating AI.

With an AI-powered tech stack combined with a data-driven strategy, merchandisers will be able to craft unprecedented customer journeys such as an omnichannel retailing experience in the ‘phygital realm’ – a convergence of physical and digital marketplaces where customers seamlessly shop online, in-store, a combination of both, or through other channels like phone and email.

A greater choice

In a headless ecommerce system, retailers are no longer shackled to the predetermined options of a SaaS ecommerce platform. They can choose the top APIs in the market, create a technology stack tailored to their specific business needs, and deploy in a shorter time-frame.

For example, a business with a large and varied catalog whose customers are search driven could prioritize world class search. While a business with a smaller catalog but high propensity to repeat purchase could instead prioritize a customer data platform (CDP) to personalize offers and maximize retention.

As well as reducing vendor lock-in, the headless approach allows for a clear division of responsibilities among different product teams, thereby supporting parallel workstreams. This means the search and discovery teams aren’t constrained by requirements from CRM, while front end engineers are free to focus on site performance.

The omnichannel story AI merchandising at work

At 8:15am, while eating breakfast, Sarah receives a mobile push notification from her favorite fashion brand. They’ve just released their latest shoe collection and want to know if she’s interested in trying on a new pair of beige shoes that match a jacket she bought just last week. They have her preferred size available at their shop downtown, close to her work place. Would she like to come in and try them on?

Sarah zooms in on the catalog image of a fashionably-attired model showcasing the shoes. She clicks “yes, make an appointment’’, and a calendar pops up in the fashion brand’s mobile app. She schedules an appointment for noon.

While at work, Sarah receives a second notification reminding her of the appointment along with a limited-time offer to get double loyalty points on purchases.

During her lunch break, Sarah heads over to the shop. The in-store team is waiting for her with not only the new shoes in her preferred size, but also the other apparel that completes the full look from the catalog image.

The shoes run small. Sarah needs one size down. Unfortunately, the store doesn’t have the correct size in stock. Would she like to have the shoes delivered tomorrow to try on at home? She agrees.

Before leaving, Sarah logs in to the in-store kiosk and browses the upcoming summer collections releasing next month. She finds a swimsuit that will be perfect for her vacation in June. She saves it to her wishlist.

The next day, the shoes are delivered to her house. They fit! She opens the mobile app and confirms her purchase.

The future is now

This is not a roadmap for some future scenario. The technology to create similar bespoke customer experiences exists today.

Sarah’s data such as purchase history and preferred store location are saved and updated in the company’s customer data platform (CDP).
Marketing orchestration software is used to schedule appointments, send notifications, and coordinate the in-store team activity.
Electronic point of sale (EPOS) and customer loyalty systems capture the data from Sarah’s store visit.
The company’s martech stack continuously updates Sarah’s preferences. The data loop captures her recent purchase journey in order to tailor her next customer journey.

This is just one example of omnichannel orchestration.

Currently many businesses lack the integrated systems to capture and process data to achieve a fully immersed omnichannel customer experience. But with the right APIs, data, and strategy, it’s inspiring what can be accomplished.

The new horizon of merchandising

AI represents a game-changing opportunity rather than a threat to merchandisers going forward. Proficiently using AI to personalize the customer experience will help build stronger brand loyalty.

AI will continue to reliably automate more low-value, mechanical tasks that merchandisers still need to perform today. This automation will allow merchandisers to elevate their role and be more strategic. It will enable merchandising teams to better align with the buying, sales, and marketing teams for the benefit of the overall organization. Those who learn how to use AI to their advantage will be more likely to succeed and win.

Combining the art of merchandising with the science of algorithms

Merchandising Studio: powerful capabilities in one dedicated interface.

Tailor-made for retail ecommerce
Zero code and 100% business user friendly with no developer dependencies
Based on a unified interface with all the functionalities in one place
Full human control, you decide which AI capabilities to use
One-stop-shop for automations, rules, and analytics
Built for data-driven merchandising strategy

Future-proof solutions for scaling your ecommerce site

Merchandising Studio is about to get even better with new AI-powered features and dynamic capabilities to optimize your data-driven merchandising strategies.