Omnichannel, Digital Technology and AI

Introduction

Visual merchandising and space planning were once largely concerned with the physical store: the arrangement of fixtures, products, graphics, lighting and circulation paths within a defined floor area. Contemporary retail has expanded this responsibility. Customers now move between websites, mobile applications, social platforms, marketplaces, stores, kiosks, fulfilment centres and customer-service channels. Consequently, the retail space is no longer an isolated environment. It is one node in an integrated customer and fulfilment system.

An omnichannel approach seeks to provide a coherent customer experience across channels while allowing each channel to perform the role for which it is most effective. A store may function as a showroom, experience centre, service point, local fulfilment location, returns facility and brand community space. The website may offer greater assortment, richer information and personalised recommendations. Mobile technology may connect the two through digital navigation, stock visibility, loyalty benefits and payment.

For visual merchandisers and space planners, this transformation creates a strategic challenge. The task is not simply to place more screens in stores. It is to design a connected environment in which physical and digital touchpoints reinforce one another, while technology remains useful, accessible, secure and commercially justified.

Learning Objectives

By the end of this chapter, the reader should be able to:

1. Explain omnichannel retailing and distinguish it from multichannel retailing.
2. Design the roles of stores within an integrated physical–digital customer journey.
3. Evaluate click-and-collect, endless aisle and experiential retail models.
4. Explain applications of computer vision, IoT sensors, RFID, digital twins, AR/VR and dynamic signage.
5. Assess AI-assisted forecasting, localisation and planogram optimisation.
6. Select relevant performance measures and identify implementation risks.
7. Apply governance principles concerning privacy, fairness, accessibility, cybersecurity and responsible automation.

1. From Multichannel to Omnichannel Retail

A multichannel retailer operates through several channels, such as stores, a website and a mobile application. These channels may have separate inventories, promotions, customer records and performance targets. An omnichannel retailer coordinates channels so that the customer experiences one connected brand and can move between channels with minimal friction.

The distinction can be expressed as follows:

Omnichannel integration has three layers. The first is experience integration: consistent brand meaning, product information, service standards and navigation cues. The second is transaction integration: shared baskets, payment, promotions, loyalty and returns. The third is operational integration: coordinated inventory, order management, fulfilment, workforce planning and data governance.

A useful managerial principle is that integration does not require every channel to be identical. A store should not imitate a website, and a website should not attempt to reproduce every sensory quality of a store. Instead, each touchpoint should have a distinctive role within the total journey.

2. The Store as a Networked Experience

The contemporary store may perform five overlapping functions:

• Discovery: enabling browsing, inspiration, product trial and social interaction.
• Conversion: supporting immediate purchase through conventional or assisted selling.
• Fulfilment: providing click-and-collect, local delivery, ship-from-store and returns.
• Service: offering repairs, consultations, demonstrations, installation advice and customer support.
• Data and relationship: generating consent-based behavioural insight and strengthening loyalty.

This broader role changes space allocation. A retailer must plan not only selling floor area but also collection counters, staging areas, returns processing, consultation zones, digital service points and employee circulation. Back-of-house capacity becomes strategically important because online orders may be picked from store inventory.

A simple space-allocation framework is to classify each area according to customer value and operational necessity. Customer-facing space should support discovery, trial and service; operational space should support accurate and rapid fulfilment. Excessive back-of-house expansion can weaken the shopping experience, while insufficient staging space can create congestion and poor collection service.

Customer Journey Orchestration

A customer journey can be mapped as a sequence of tasks rather than channels:

1. Recognise a need.
2. Search and compare alternatives.
3. Assess fit, quality, price and availability.
4. Try, configure or discuss the product.
5. Purchase or reserve.
6. Receive, collect or use the product.
7. Seek support, exchange or return.
8. Re-engage when a relevant need arises.

The retailer should identify which channel is best at each task. For example, online search may be efficient for comparison, while the store may be superior for tactile assessment. A customer may reserve online, test the item in store, purchase through an associate’s tablet and receive a later delivery. Visual merchandising must therefore provide cues that explain these possibilities clearly.

3. Integrated Retail Models

Click-and-Collect

Click-and-collect allows customers to order online and retrieve products from a selected store or collection point. Its advantages include reduced last-mile cost, convenience and additional store visits. However, it requires accurate stock records, disciplined picking, clear customer communication and a well-designed collection area.

Space-planning requirements include:

• a visible but secure collection point;
• sufficient staging capacity for orders of different sizes;
• clear separation between waiting, collection and returns;
• privacy for customer information;
• employee access that does not obstruct selling areas; and
• signage that explains identification and collection procedures.

The operational promise must reflect actual capability. If the retailer promises collection within two hours, the store needs reliable inventory data, labour capacity and a defined service-level process.

Endless Aisle

An endless aisle gives customers access to products not physically stocked in the store. An associate or customer may use a kiosk, tablet or mobile interface to view extended assortment, colours, sizes or complementary products. Orders may be delivered to the customer or collected later.

The endless aisle is particularly valuable when physical space is constrained or when products are highly variable. It also reduces the pressure to carry every variant on the shelf. Its effectiveness depends on accurate product content, real-time or near-real-time availability, transparent delivery dates and a simple payment process.

The visual challenge is to make the digital extension feel like a credible part of the store rather than an apology for missing stock. Product samples, comparison displays and staff expertise should connect the physical item to the extended digital range.

Experiential Retail

Experiential retail uses the store to create meaningful participation rather than merely provide product access. Examples include product demonstrations, workshops, customisation, sports testing, beauty consultations, community events and immersive brand storytelling.

Experiential space requires different planning logic from conventional selling space. It must accommodate dwell time, seating, facilitation, storage, safety and resetting between sessions. Its success should not be judged solely by immediate sales. Measures may include appointment attendance, conversion after experience, repeat visits, membership growth, customer satisfaction and content sharing, subject to privacy and consent requirements.

4. Digital Technologies for Visual Merchandising and Space Planning

Computer Vision

Computer vision uses cameras and machine-learning models to interpret visual information. In retail it may support anonymous people counting, queue estimation, dwell analysis, shelf availability detection, planogram compliance and safety monitoring.

A computer-vision system generally includes image capture, preprocessing, model inference, event classification, data storage and reporting. The model may detect an empty shelf position, but managers must still distinguish genuine stock absence from occlusion, poor lighting, misplaced products or a temporary replenishment activity.

Computer vision should be designed around data minimisation. If aggregate traffic counts answer the business question, storing identifiable facial images is unnecessary. Accuracy should be tested across lighting conditions, customer demographics, clothing, mobility aids and store layouts. Human review is appropriate when automated findings could affect employees or customers materially.

IoT Sensors

Internet of Things (IoT) sensors connect physical objects or environments to digital systems. Common retail applications include temperature monitoring, footfall measurement, shelf weight sensing, queue detection, energy management and equipment diagnostics.

Sensors are valuable when they produce an actionable signal. A temperature sensor in a food display may protect product quality; a queue sensor may trigger staff redeployment. Sensor proliferation without a decision process creates cost, maintenance burden and data noise. Each sensor project should define the event measured, the response required, acceptable latency, data retention and responsible owner.

RFID

Radio-frequency identification uses tags and readers to identify objects through radio communication. Compared with manual counting or conventional barcodes, RFID can improve item-level visibility and support faster inventory counts, location awareness and loss prevention.

RFID performance is affected by tag placement, metal or liquid interference, reader configuration, process discipline and environmental conditions. It is not automatically a substitute for good stockroom practice. The business case should consider tag, reader, integration, labour and maintenance costs against improvements in availability, accuracy and fulfilment.

Digital Twins

A digital twin is a continuously updated digital representation of a physical asset, space or process. In retail, a store twin may combine floor plans, fixture locations, inventory, traffic patterns, environmental readings and operational events. It can be used to simulate a new layout before implementation, evaluate queue changes or test the effect of relocating a category.

A digital twin is more than a static 3D model. Its value depends on data quality, update frequency, model validity and the ability to connect simulation results with managerial action. It should be treated as a decision-support model rather than an unquestionable replica of reality.

Augmented and Virtual Reality

Augmented reality (AR) overlays digital information on the user’s view of the physical world. Applications include virtual try-on, furniture placement, navigation and product information. Virtual reality (VR) creates an immersive simulated environment and may support staff training, store-layout evaluation or remote product experiences.

AR and VR can reduce uncertainty, but they also introduce accessibility, device, motion-sickness and calibration issues. They should complement physical evaluation where touch, fit, colour or material quality is important. The retailer should provide a non-AR route to equivalent essential information.

Dynamic Digital Signage

Digital signage permits content to change by time, stock status, audience, weather, location or campaign. Dynamic content can reduce the delay between merchandising decisions and in-store communication. However, excessive motion, frequent changes or poorly prioritised messages can increase cognitive load and weaken brand clarity.

A signage governance system should define content ownership, approval rights, accessibility standards, screen placement, refresh rules, fallback content and measurement. Dynamic pricing or personalised messaging requires particular care because customers may perceive inconsistent treatment as unfair or deceptive.

5. AI in Retail Space Planning

AI in retail usually refers to statistical learning, optimisation, computer vision, natural-language systems or generative systems used to support decisions. AI does not remove the need for merchandising judgement. It changes the speed, scale and evidence base of decision-making.

AI-Assisted Demand Forecasting

A basic forecasting formulation represents demand in period t as a function of historical and contextual variables:

Here, ŷ_t+h is the forecast for a future period t+h, y_1:t is observed demand up to period t, x_t+1:t+h contains known or forecast explanatory variables such as price, promotion, holidays and weather, and z represents product or location characteristics. The function f may be a statistical model or machine-learning model.

Forecasts must distinguish demand from observed sales. Sales may be lower than true demand when products are out of stock, stores are closed or capacity is constrained. Training a model on unadjusted sales can therefore cause systematic under-forecasting for frequently unavailable products.

Two commonly used accuracy measures are:

In these equations, y_i is actual demand, ŷ_i is forecast demand and n is the number of observations. MAE is expressed in the same units as demand and treats all errors linearly. RMSE is also expressed in demand units but penalises larger errors more strongly. Neither measure alone establishes commercial usefulness; availability, waste, margin and service outcomes must also be considered.

Localisation and Assortment

AI can estimate differences in demand by store, neighbourhood, climate, catchment, customer mission and local events. Localisation should not be reduced to stereotypes based on postcode or demographic proxies. A responsible process combines observed performance, customer research, local staff knowledge and model outputs.

The objective is usually not maximum local variety. It is an economically and experientially appropriate assortment. A localised plan must consider minimum display quantities, replenishment frequency, substitution options, supplier constraints and the customer’s expectation of core products being available everywhere.

Planogram Optimisation

A planogram specifies the location, facing, adjacency and quantity of products within a fixture. AI-assisted planogramming can evaluate many alternatives against objectives such as expected sales, margin, availability, brand rules, shopper convenience and replenishment effort.

A simplified optimisation problem can be written as:

where S is expected sales, M is expected margin, A is an availability or customer-service measure, L is labour or replenishment burden, and α, β, γ and δ are managerial weights. This is a conceptual objective function, not a universal formula. In practice, the model must include constraints such as fixture dimensions, shelf capacity, product compatibility, safety, brand commitments and minimum facings.

Optimisation results should be reviewed by experienced merchandisers. A mathematically efficient arrangement may be visually confusing, difficult to shop, inaccessible, inconsistent with brand architecture or impractical to replenish. Human judgement is especially important where the objective includes emotional response, cultural meaning or emerging products with limited historical data.

6. Measurement and Economic Evaluation

Omnichannel projects require measures that connect customer experience with operational and financial outcomes. Useful measures include:

Sales per square metre is often calculated as:

Both sales and area must use consistent periods and definitions. This measure is useful for comparing space productivity, but it can penalise experiential areas whose value appears through later online purchases or loyalty rather than immediate store sales.

A click-and-collect evaluation should include incremental sales, fulfilment labour, collection infrastructure, failed collections, returns and possible additional purchases. An apparently successful service may destroy value if it generates high handling cost or excessive customer waiting.

7. Illustrative Case: Northstar Home’s Connected Store Programme

Background

Northstar Home, a fictional mid-market home-furnishings retailer, operated 18 stores and an e-commerce site. Its stores carried popular products but had limited space for colour and size variants. Online orders were fulfilled from a central warehouse, while store inventory records were updated only periodically.

Problem

Customers frequently visited stores to assess products and then discovered that their preferred variant was unavailable. Online orders were sometimes accepted despite inaccurate local inventory. The retailer also observed congestion near the service desk because returns, collections and product advice were handled in one queue.

Analysis

A process review identified four causes: weak inventory accuracy, no dedicated collection zone, insufficient digital access to extended assortment and channel-specific performance targets. The retailer considered computer vision, RFID and an endless-aisle interface, but rejected facial recognition because it was not necessary for the identified business problems.

Approach

Northstar piloted the programme in three stores. RFID was introduced for selected high-value and high-variance products. A tablet-based endless aisle allowed associates to order unavailable variants. A redesigned front-of-store area separated collection, returns and consultations. A digital twin of each pilot store was used to test queue positions and staging capacity before construction.

The retailer established the following controls:

• customer consent for personalised digital communications;
• aggregate traffic analytics rather than facial identification;
• human approval of AI-generated planograms;
• a clear distinction between available stock and stock requiring confirmation;
• weekly review of forecast errors and stock exceptions; and
• accessibility testing for tablets, signage and collection procedures.

Outcome

The pilot improved inventory confidence and reduced customer confusion. The endless aisle increased access to extended variants, while the separated service zones reduced visible queue conflict. However, the retailer found that RFID benefits were strongest for selected categories and that staff training was more important than the technology interface alone. Some customers preferred associate assistance, so self-service was retained as an option rather than imposed as the only route.

Learning Points

First, omnichannel performance depends on process integration, not merely customer-facing technology. Second, store space must accommodate online fulfilment without allowing operational activity to dominate the customer experience. Third, a smaller, well-governed technology deployment can be more valuable than an ambitious but weakly specified digital programme. Finally, metrics should evaluate the total customer journey rather than assign every sale to the final channel used.

8. Governance, Ethics and Risk

Privacy and Consent

Retail technologies can generate location, behavioural and potentially biometric data. Organisations should specify purpose, collect only necessary information, provide understandable notice, obtain consent where required, establish retention limits and protect data against unauthorised access. Privacy compliance is jurisdiction-dependent; retailers must follow applicable data-protection and consumer-protection law rather than relying on a generic global policy.

Bias and Fairness

AI systems may reproduce historical inequalities or use proxy variables that disadvantage particular groups. Examples include unequal accuracy in computer vision, exclusionary product recommendations or localisation that systematically under-serves certain communities. Model evaluation should examine relevant subgroups, document limitations and provide escalation routes.

Transparency and Human Accountability

Customers and employees should know when significant decisions are automated or when data is being collected for purposes beyond immediate service. A manager remains accountable for planogram approval, workforce consequences and customer treatment even when a system generates the recommendation.

Accessibility and Inclusion

Digital signage, kiosks and AR tools should not replace essential human assistance or non-digital alternatives. Text must be legible, interfaces should support assistive technologies where applicable, and layouts must respect mobility, sensory and cognitive needs. Inclusive design is both an ethical obligation and a practical way to serve a wider market.

Cybersecurity and Resilience

Connected stores increase the attack surface. RFID readers, sensors, kiosks, point-of-sale systems and cloud services should be segmented appropriately, patched, monitored and governed through access controls. Retailers also need fallback procedures for network failure, incorrect stock data, system outages and cyber incidents.

9. Implementation Framework and Best Practices

A disciplined implementation sequence is preferable to technology-led experimentation:

1. Define the customer or operational problem. State the decision that must improve.
2. Map the current journey and process. Identify delays, failure points and ownership gaps.
3. Establish baseline measures. Record current availability, labour, conversion, queues and satisfaction.
4. Assess data readiness. Check accuracy, completeness, consent, interoperability and ownership.
5. Select the least complex effective technology. Do not use AI where a reliable rule or process is sufficient.
6. Pilot in representative locations. Include different formats and operating conditions where possible.
7. Evaluate causal impact. Use comparison stores, phased rollout or controlled tests rather than relying only on before-and-after impressions.
8. Train employees and redesign work. Technology changes tasks, authority and service expectations.
9. Govern and audit. Review performance, bias, privacy, security, accessibility and unintended effects.
10. Scale selectively. Expand only when benefits, costs and risks are understood.

Best practice is therefore socio-technical. Hardware, software, data, physical design, employee capability and governance must be considered as one system.

Contemporary Developments

Several developments are reshaping the field. Retail media increasingly connects in-store screens and digital advertising, creating opportunities but also risks of excessive commercial messaging. Generative AI can assist with product descriptions, campaign variants, scenario exploration and conversational shopping; its outputs require factual, brand and legal review. Edge computing can process sensor information closer to the store, reducing latency and the transmission of raw data. More retailers are also exploring energy-aware store operations, using occupancy and environmental data to reduce consumption without compromising comfort.

The strategic direction is not toward a completely automated store. Rather, it is toward an adaptive store in which physical space, digital information and human expertise are coordinated. The most successful retailers will treat technology as a means of improving relevance, convenience and confidence—not as a substitute for thoughtful merchandising.

Chapter Summary

Omnichannel retail integrates stores, websites, mobile applications, fulfilment and service into a connected customer journey. The store may operate as a selling environment, showroom, service centre, local warehouse and community space. Click-and-collect requires accurate inventory and dedicated staging; endless aisle extends assortment beyond physical stock; experiential retail creates value through participation and service.

Computer vision, IoT sensors, RFID, digital twins, AR/VR and dynamic signage can improve visibility, navigation, planning and engagement, but each requires a defined business purpose and appropriate governance. AI can support demand forecasting, localised assortment and planogram optimisation. Forecast evaluation must account for stockouts and should combine accuracy measures with commercial and customer outcomes.

Technology implementation is socio-technical. Privacy, consent, fairness, accessibility, cybersecurity, employee training and human accountability are essential. Effective space planning now coordinates physical layout, digital interfaces, fulfilment processes, data quality and brand experience across the entire retail system.

Key Terms

Omnichannel retail: Coordinated management of channels to provide an integrated customer and operational experience.

Multichannel retail: Retailing through multiple channels that may remain operationally separate.

Click-and-collect: Online ordering followed by collection from a store or designated point.

Endless aisle: Digital access to products or variants not physically stocked in the store.

Experiential retail: Retail designed around participation, trial, education, service or community.

Computer vision: Technology that interprets visual information using algorithms and machine learning.

Internet of Things: Connected physical devices that collect, transmit or act on data.

RFID: Radio-based identification technology used to identify and track tagged items.

Digital twin: A data-connected digital representation of a physical asset, environment or process.

Augmented reality: Digital information overlaid on the user’s view of the physical environment.

Virtual reality: An immersive computer-generated environment.

Dynamic signage: Digital communication whose content can change according to defined conditions.

Demand forecasting: Estimation of future demand using historical and contextual information.

Localisation: Adaptation of assortment, communication or experience to a specific location or market.

Planogram: A visual specification for product placement within a fixture or selling area.

Inventory accuracy: Agreement between recorded inventory and actual available inventory.

Data minimisation: Collection and retention of only the data necessary for a defined purpose.

Human-in-the-loop: A governance arrangement in which people review, approve or intervene in automated processes.

Sales density: Net sales divided by selling floor area over a defined period.

Review Questions

Short Answer Questions

1. How does omnichannel retail differ from multichannel retail?
2. Identify three functions that a contemporary store may perform besides direct selling.
3. What operational conditions are necessary for effective click-and-collect?
4. Give two applications of RFID in retail space management.
5. Why should AI-generated planograms receive human review?

Long Answer Questions

1. Explain how click-and-collect and endless aisle models change visual merchandising and space planning.
2. Evaluate the potential benefits and limitations of computer vision, IoT sensors and RFID in a large-format store.
3. Discuss how AI-assisted forecasting can improve localised assortment decisions while creating risks of bias or under-service.
4. Explain the role of digital twins in testing store layouts and operational processes.
5. Develop a governance framework for dynamic signage and personalised digital communication in stores.

Analytical / Application Questions

1. A retailer has frequent online cancellations because store inventory is inaccurate. Design a phased intervention using process redesign, RFID or other technologies, and specify the measures that would demonstrate improvement.
2. A fashion retailer wants to convert 20% of selling space into experiential zones while maintaining profitability. Propose a space-planning and measurement approach that captures both immediate and delayed omnichannel value.
3. An AI system recommends reducing assortment in a low-income neighbourhood because historical sales are weak. Critically assess the recommendation and propose a fairer decision process that incorporates demand constraints, customer access and local knowledge.

Multiple Choice Questions with Answers

Question 1

Which statement best distinguishes omnichannel retail from multichannel retail?

A. Omnichannel retail uses only physical stores and mobile applications.

B. Omnichannel retail coordinates channels into an integrated customer and operational experience.

C. Multichannel retail always provides better inventory accuracy.

D. Omnichannel retail eliminates the need for store employees.

Correct Answer: B. Omnichannel retail coordinates channels into an integrated customer and operational experience.

Question 2

What is a primary operational requirement for reliable click-and-collect?

A. Accurate inventory visibility and disciplined order staging

B. Removing all service counters from the store

C. Displaying only the retailer’s highest-margin products

D. Using virtual reality for every collection order

Correct Answer: A. Accurate inventory visibility and disciplined order staging

Question 3

Which technology identifies tagged items using radio communication?

A. Computer vision

B. Dynamic signage

C. RFID

D. Virtual reality

Correct Answer: C. RFID

Question 4

In demand forecasting, what does MAE measure?

A. The average absolute difference between actual and forecast demand

B. The percentage of shelf space occupied by a product

C. The total margin earned per square metre

D. The maximum error divided by inventory

Correct Answer: A. The average absolute difference between actual and forecast demand

Question 5

Why can observed sales understate true demand?

A. Sales are always higher than demand during promotions.

B. Stockouts or capacity constraints may prevent some desired purchases.

C. Forecasting models automatically increase recorded sales.

D. Digital signage removes all customer demand.

Correct Answer: B. Stockouts or capacity constraints may prevent some desired purchases.

Question 6

What is the principal purpose of an endless aisle?

A. To prevent customers from accessing online information

B. To provide digital access to products or variants not physically stocked

C. To replace all physical product samples

D. To reduce the need for inventory records

Correct Answer: B. To provide digital access to products or variants not physically stocked

Question 7

Which statement about a digital twin is most accurate?

A. It is only a static photograph of a store.

B. It is a data-connected digital representation used to analyse or simulate a physical system.

C. It is a legal replacement for a store manager.

D. It can produce valid results regardless of data quality.

Correct Answer: B. It is a data-connected digital representation used to analyse or simulate a physical system.

Question 8

Which is an example of data minimisation?

A. Retaining identifiable video indefinitely for possible future use

B. Collecting only aggregate traffic counts when individual identity is unnecessary

C. Sharing all sensor data with every supplier

D. Collecting biometric data for every store visitor by default

Correct Answer: B. Collecting only aggregate traffic counts when individual identity is unnecessary

Question 9

What is a major limitation of evaluating experiential retail only through immediate store sales?

A. Experiential areas can create later online purchases, loyalty and service value.

B. Experiential areas never affect customer behaviour.

C. Store sales are always unrelated to customer experience.

D. Experiential retail cannot be measured quantitatively.

Correct Answer: A. Experiential areas can create later online purchases, loyalty and service value.

Question 10

In AI-assisted planogramming, why are constraints necessary?

A. They ensure that recommendations respect fixture capacity, safety, compatibility and operational requirements.

B. They guarantee that every product will sell equally.

C. They eliminate the need for historical data.

D. They prevent any human from reviewing the result.

Correct Answer: A. They ensure that recommendations respect fixture capacity, safety, compatibility and operational requirements.