For the last two decades, ecommerce has been built around search, scroll, click.

But these foundations are beginning to crumble in the latest seismic shift in the commerce landscape.

Website traffic and clicks are declining for many brands as consumers shift toward conversational AI tools that help them discover, compare, and buy without visiting a traditional site.

Gartner predicts a 25% decline in traditional search volume this year, forcing businesses to rethink their approach to online visibility.

We’re now entering the era of agentic commerce, where AI goes beyond helping shoppers browse by actively participating in the journey: discovering, comparing, recommending, and purchasing.

What was only recently a concept is now being rolled out by some of the biggest players in retail, payments, and AI.

And while it’s launching in the US first, agentic commerce is set to be turned on globally.

Top agentic commerce solutions

Several brands and retailers are now enabling purchases directly through agentic AI tools like ChatGPT. For example, OpenAI’s Instant Checkout allows users to buy from select Shopify merchants as well as Etsy sellers.

Instant Checkout runs on the Agentic Commerce Protocol (ACP), built by OpenAI and Stripe.

Any brand can use it, whether they already work with Stripe or another payment processor, via Stripe’s shared payment token system.

In practice, this means when you click the “Buy” button in ChatGPT’s Instant Checkout, Stripe handles the checkout and payment behind the scenes.

You can pay with credit cards, Apple Pay, Google Pay, or other methods directly in the chat — with no need to leave the conversation or go to another website.

Within the payments ecosystem, PayPal became the first payments wallet embedded in ChatGPT.

Google is taking a different but complementary approach with its Universal Commerce Protocol (UCP). Rather than embedding checkout into a single AI product, UCP is designed as an open standard that lets AI agents, merchants, and payment providers speak the same language across any interface.

In effect, it turns existing commerce stacks into agent-ready infrastructure, so AI experiences like Google Search and Gemini can move from recommending products to completing purchases securely and at scale.

The early agentic adopters

US retailers including Walmart and Sam’s Club have also integrated with ChatGPT, letting customers complete grocery purchases natively in the conversation.

Initially confined to the US, agentic commerce is rolling out globally.

In Europe, Frasers Group brands, including Sports Direct and House of Fraser, are adopting agentic commerce using commercetools to enable in-chat purchases.

This marks one of the first mainstream rollouts of native shop-in-chat functionality outside North America.

Consumer confidence in agentic commerce

Despite the rapid rise of agentic commerce, businesses will need to ensure they’re bringing transparency to their customers.

In a Riskified survey, nearly 3/4 of international consumers say they’re using AI in their shopping journeys, with 70% saying they feel somewhat comfortable with an AI agent making a purchase on their behalf.

However, only 13% of consumers have completed a purchase after an AI agent sent them to a website.

This gap between interest and action highlights trust at the heart of agentic commerce.

Consumers like the idea of AI-assisted shopping, but they hesitate at the point of purchase.

Their biggest concern, according to the survey, is payment security, cited by about one-third of respondents, followed by worries about privacy, potential mistakes, and loss of control over decisions.

How brands can prepare for agentic commerce

Agentic commerce is changing the rules of digital commerce.

AI agents are becoming the first place customers go to decide what to buy, which means brands need to optimize not just for search engines but for decision engines.

Here’s how to get ready. 

1. Make your commerce stack agent-ready

Product data is how agents see your catalog.

AI agents rely on structured, complete data to represent your brand. This includes product information, pricing, availability, promotions, delivery options, and returns policies.

If your catalog only works for human browsers, agents won’t be able to accurately recommend or sell your products. Ensuring your data is organized and machine-readable allows AI to navigate your catalog, compare options, and complete purchases.

2. Rethink SEO as AEO

Traditional SEO (ranking for keywords) is still useful, but in agentic commerce, you need to focus on AEO: making your product and policy information clear and structured so AI can interpret it.

Agents need answers to questions like: What is this product? Who is it for? How much does it cost? When will it arrive? What happens if there’s a problem?

The clearer your answers, the more likely an agent will recommend your brand and complete a purchase.

KPIs also change. Traffic and pageviews become less critical. The key metrics are how often AI agents select your brand and how effectively they convert intent into purchase. Achieving this requires closer alignment across commerce, payments, data, and trust.

3. Create a single source of truth

Centralize all product data in one system that controls attributes, variants, languages, and updates. A product information management (PIM) system is key here: it ensures all product details are accurate, complete, and consistent across every touchpoint.

With a PIM, you can manage updates in one place, reduce errors or mismatches, and maintain reliable data for all your AI-assisted shopping experiences.

4. Design for trust and transparency

AI-assisted purchases will require consumer confidence. So make sure your policies, pricing, and payment flows are clear and easy to verify.

Offer familiar payment options, visible consent for purchases, and easy ways to review or cancel orders. The more transparent and reliable the experience, the more likely customers will let agents act on their behalf.

Your next steps for agentic commerce

If you want to understand how agentic commerce will impact your digital commerce strategy, now is the time to start the conversation.

The future of commerce won’t be built on clicks, but on how confidently AI agents can act on your customers’ behalf.

Tryzens, the global branded commerce agency, has partnered with JD Outdoors Group, part of the JD Sports Group, one of the world’s leading sports fashion and outdoor retail businesses.

As part of this collaboration, Tryzens has been supporting brands including GO Outdoors, Naylors, Tiso, Millets, and Blacks, as the group modernizes and scales its digital commerce ecosystem.

With a diverse portfolio and millions of customers engaging across channels, the next phase of growth demanded a platform strategy that is streamlined, resilient, and built for omnichannel at scale.

Unpacking complexity to drive transformation

Each brand brings its own architecture, operations, heritage, and customer expectations. They faced a complex legacy environment where changes took months to implement, slowing growth and limiting innovation.

Tryzens’ experience mapping complex enterprise requirements enabled a seamless transition for JD Outdoors’ brands onto the Shopify platform and its broader ecosystem. 

This delivered a solution that:

• Reduces time to market: what once took months to roll out now takes hours 

• Unlocks innovation: merchandizing teams gain control without IT intervention 

• Supports change management: from data migration to discoverability, we have been collaborating with JD Outdoors on every step 

• Integrated seamlessly: as an enterprise parter, our experience with integration layers, pricing nuancing, and third-party platforms ensured that a single partner could handle the end-to-end transformation 

Building a scalable foundation 

Our scope of work included the delivery of 5 Shopify stores for JD Outdoors and the implementation of OneStock across 4 JD Outdoors sites. 

We are also providing change management and training to embed long-term operational capability, ensuring teams can manage change confidently. 

Together, we’re helping JD Outdoors advance a broader digital and omnichannel strategy focused on core outcomes: 

• Modernizing customer experience by leveraging Shopify and its app ecosystem to enhance performance, UX, and conversion. 

• Creating a single view of inventory, consolidating stock and orders across physical stores, ecommerce, and future channels to improve utilization, reduce operational cost, and enhance service. 

• Enabling true omnichannel execution, unlocking capabilities such as click & collect, store fulfillment, and seamless customer journeys across brands and regions. 

 The result is a new foundation to support JD’s ambition for connected commerce across its growing brand portfolio. 

Multi-brand rollout at enterprise scale 

The transformation program is already underway. 

We have packaged Shopify into a recognizable enterprise framework, using modular blocks that can be reused across brands, which includes complex integrations that have been engineered for improved page speed. 

In October 2025, Naylors became the first brand live as part of the replatform program.  

The launch marked a major milestone, setting the foundation for a scalable, high-performance ecommerce model. Features such as live stock lookup and click & collect further strengthen the omnichannel experience.

Momentum continued into the new year when Tiso recently launched on Shopify, with orders flowing seamlessly and performance tracking strongly. 

The rollout accelerated mid-January with two further launches of heritage brands: Millets and Blacks. 

This was followed by the launch of GO Outdoors, the largest in JD Outdoors’ portfolio, and one of the country’s biggest outdoor retailers.

This was a large-scale project. Using our experience in mapping complex legacy systems to specific platforms, we built an enterprise architecture that is practical, scalable, and aligned with the high standards JD Outdoors is familiar with.

River Island, one of the most enduring names on the British High Street, has selected Tryzens Global as its digital transformation partner.

The partnership will focus on leading a major initiative to simplify its commerce stack, reduce operating costs, and accelerate innovation and trading agility.

Founded in 1948 by Bernard Lewis, the family-owned retailer is known for trend-driven womenswear with a global online presence and over 200 stores in the UK and Ireland.

River Island was also an early digital adopter, launching its online store in the late 1990s — one of the first major High Street brands to do so.

This milestone helped the brand set the pace for online retail in the UK.

The foundations for future transformation

As many retailers are realizing, operating a digital commerce architecture at scale can come with increasing complexity and cost.

River Island’s digital ecosystem has naturally evolved over time, and the business is now exploring ways to simplify and modernize its technology stack to enable the business to focus its efforts on and expanding its customer base, enriching the product experience, and providing an efficient and agile trading model across their core engagement channels.

Working with Tryzens, River Island aims to deliver on its strategy to reduce their cost to serve while protecting brand equity, enriching customer experience, and providing the trading team with flexibility to adapt to market changes.

The program starts with discovery to shape the digital transformation of the wider commerce stack, including ecommerce and product information management.

The first phase of delivery centers on a roll out to the international operations, leveraging the Shopify, Global-e, and Akeneo platforms.

Why River Island chose Tryzens

River Island selected Tryzens based on our proven experience delivering enterprise-scale omnichannel transformations for major brands.

Tryzens operates the breadth of the digital commerce ecosystem, covering all major ecommerce platforms, as well as PIM, OMS, and POS solutions with the intent to align the technologies that best fit the specific business needs and ambitions of each client.

Equally as important, River Island wanted a partner who could augment and support the team already in place with depth of experience to de-risk lead large-scale programs. Our operating model is built around co-ownership of delivery: collaborative governance and delivery aligned to commercial outcomes.

Finally, Tryzens brings value-added services that reduce delivery risk and strengthen in-life operations, giving River Island confidence not only in how the transformation will be delivered but continuous guidance on in-life operations.

Retailers often claim they’re data-driven, yet most are quietly fighting fires caused by weak data foundations.

Duplicated customer records, mismatched product attributes across systems, inconsistent content metadata… bad data costs retailers an average of $13 million each year.

Businesses no longer have the luxury to ignore poor data.

And that number is almost certainly higher once you account for lost conversion, failed personalization, bloated operational workloads, and AI initiatives that never take off because the inputs are fundamentally unstable.

The consequences of bad data

Integrations are the backbone of digital commerce operations; a smorgasbord of acronyms – PIM, DAM, CMS, OMS, CRM, ERP, CDP, loyalty platforms, commerce platforms – highlight this interdependency.

When the data is inconsistent or poorly governed, every integration becomes a mess. 

• IDs don’t match across systems. 

• Attributes appear, disappear, or change names. 

• Stock feeds vary depending on the source. 

• Media assets use different naming conventions. 

• Customer identities can’t be stitched across channels. 

Beyond operational pain, weak governance also creates exposure on the regulatory side. Inconsistent or inaccurate data can trigger compliance risks across global and regional frameworks, putting both the business and its customers at risk.

Every one of these issues creates unnecessary rework, regression bugs, and builds up technical debt.

Retailers feel that impact every time they launch a new experience and data issues surface. 

This is where governance becomes an accelerant.

When data across systems is consistent, integration becomes faster and faster means cheaper.

How data governance drives revenue

Too often data governance is seen as a compliance box-tick. But in modern retail, governance can become a performance lever.

Research from McKinsey found that data-driven organizations are 23× more likely to acquire customers and 19× more likely to be profitable. And those outcomes hinge on good governance.

Clean, structured, consistent data delivers measurable upsides that include:

• Higher conversion: standardized product attributes improve search accuracy and reduce null results, helping to drive more confident buying decisions. Crystallize research shows that richer, more consistent product information correlates with a 20–50% boost in online conversion rates.

• Lower returns: customers return fewer products when the metadata, descriptions, sizing, and imagery are accurate and consistent. In fact, an Akeneo survey found that 62% of consumers said more accurate product information would make them less likely to return items.

• Hyper-personalization: good customer data leads to smarter recommendations and more relevant campaigns. According to Cension analysis, hyper-personalized offers can yield around a 30% average lift in conversion rates.

• Operational efficiency: merchandisers, content teams, and engineers stop wasting hours fixing issues downstream in CMS, PIM, and commerce platforms. According to a report by CSS Commerce, time spent searching for correct product data is significantly reduced, with some saving as much as 2 hours per week per employee – scale that across teams and the productivity gains compound quickly.

• Faster delivery: when downstream systems receive predictable data, releases stop breaking and sprint cycles speed up. A CDP Institute report found that brands are achieving up to a 75% reduction in time to market for marketing initiatives after implementing a CDP.

• Product syndication: when feeds to Google Shopping, marketplaces, social platforms, and affiliate networks are consistent and complete, ads rank better and product visibility increases. The result is higher-quality traffic and stronger ROAS.
   

Where AI fits into good governance

Brands and retailers have been rushing toward AI-powered capabilities. Almost every platform in the digital commerce ecosystem now has its own integrated AI features.

From generative content for product descriptions and email campaigns to predictive personalization and automated recommendations, AI is increasingly embedded into workflows that drive conversion and loyalty.

But while AI promises speed and scale, its impact depends on the quality and governance of the data used. 

That’s because AI is an amplifier.

Feeding AI with low-quality inputs only amplifies inconsistencies and multiplies mistakes. What was once a trickle of errors, now a flood that drowns teams.

Put another way, if your product attributes are inconsistent, AI enrichment models will be inconsistent.

If your customer data is fragmented, personalization models will be incoherent.

If your inventory and pricing data aren’t trustworthy, forecasting will be wrong.

A study in the Engineering Management Review found that most AI projects fail, and one of the fundamental reasons for this is “poor data quality” and “a lack of proper data governance”.

A data governance blueprint for retailers

Retailers need a practical approach to governing their data, one grounded in the realities of digital commerce.

A retail-first governance model prioritizes:

1. Product data quality 

Product data touches everything: PDPs, search, recommendations, SEO, returns. Priorities include: 

• Standardized attribute sets across all categories 

• Clear ownership between merchandisers and content teams 

• Consistent product hierarchies 

• Automated and rule-based enrichment workflows 

• Standardized media naming and tagging

2. Customer data accuracy 

With customer identity at the heart of personalization, omnichannel, and loyalty, unified profiles are essential. Priorities include: 

• Unified customer ID and identity resolution 

• Consent and preference governance 

• Clean integration into CDPs, marketing automation, and loyalty platforms

Good governance also makes the customer journey smoother. When data points can flow cleanly between systems, retailers can surface information a shopper has already shared or shown interest in — instead of forcing them to re-enter it.

For example, if a customer has explored a specific category, configured an item, or provided details earlier in the funnel, governed data ensures downstream experiences can recognize it, adapt instantly, and remove friction.

3. Operational consistency 

Shared standards for naming, formatting, enrichment, and validation across content teams, merchandisers, trading, engineering, and marketing. Priorities include: 

• Reusable content taxonomies 

• Structured tagging across CMS, DAM, PIM 

• Global vs local content ownership 

• Metadata rules for imagery, video, copy, and UGC

4. Platform alignment 

Governance rules ensure your PIM, DAM, CDP, CMS, commerce platform, and marketing stack all speak the same language, so data moves without breaking. Priorities include: 

• Versioning rules for how systems talk to each other 

• Standard formats for events and clarity on which system owns which data 

• A single source of truth for every key data point 

• Visibility into where data comes from and how it moves through the stack 

• Consistent rules for IDs, timestamps, and how attributes are named and formatted 

Next steps

Most retailers aren’t suffering from a lack of data, but rather, they’re suffering from the cost of unmanaged data.

And the performance gap between those with strong governance and those without is widening fast as AI, personalization, and omnichannel expectations accelerate.

If retailers want personalization to scale and teams to move faster, governance is the unlock.

Tryzens works across platforms — commerce platforms, CMS, CDP, ERP, PIM, and others — so we know where data breaks, how to fix it, and how to mitigate it.

If you’re looking to build a governance framework that aligns all platforms in your tech stack, then let’s talk.

A customer data platform (CDP) collects customer data from multiple sources to create a unified customer profile.

It centralizes information from websites, mobile apps, CRM systems, and point-of-sale to build a 360-degree view of each customer, helping marketing, sales, and service teams deliver personalized experiences.

CDPs focus on four core tasks: collecting data, harmonizing data, activating data, and pulling insights from data.

Do those well, and you get faster personalization, fewer data silos, and clearer measurement of customer-driven revenue.

The market’s growing fast: the CDP market was valued at about $7.39 billion and is forecast to reach roughly $23.98 billion by 2029.

Gartner analysis forecasts two trends emerging within the CDP market: data-sharing innovations like zero copy (which allows assessing data across multiple databases without needing to copy or reformat it) and embedding artificial intelligence to make predictive analytics more accessible.

Below we compare six leading CDPs used in retail tech stacks.

For each, we state what it does best, the features that matter, and the kinds of retailers that use it.

Salesforce Data 360

Salesforce Data 360 (formerly Data Cloud) unifies large enterprise data estates into the Salesforce ecosystem and makes that data actionable across Sales Cloud, Marketing Cloud, Service Cloud, and Commerce Cloud.

Key features 

• Hyperscale data platform built into Salesforce with low-code/no-code tooling. 

• Zero-copy integrations to major data lakes and warehouses (Snowflake, Databricks), letting teams query data where it lives without wholesale copying. 

• Tight integration with Salesforce automation and AI tools, so you can turn unified profiles into journeys, offers, or agent workflows without heavy engineering.

Data processing 

Data stays in your existing systems (data lakes or warehouses) and is queried as needed. Updates happen near real-time, letting teams act across Salesforce tools without heavy data movement.

Customer profile 

Enterprise retailers with existing Salesforce investments, complex omnichannel operations, and a need to put data to work across many internal teams.

Some brands and retailers who use Data 360 include Pandora, PepsiCo, R.M.Williams, and Living Edge.

Adobe Real-Time CDP

Adobe Real-Time CDP creates real-time unified customer profiles across Adobe Experience Cloud for content-driven personalization and cross-channel campaigns.

Key features

• Built on Adobe Experience Platform, it unifies known and anonymous data to create real-time profiles for both B2Cs and B2Bs. 

• Native activation into Adobe Campaign, Target, and Experience Manager for tightly coordinated content and campaign delivery. 

• Expand and enrich first-party audiences with privacy-safe data collaborations to drive more effective acquisition, personalization, and loyalty across channels.

Data processing

Collects data continuously from multiple sources to create up-to-the-minute customer profiles. Profiles can be used instantly in Adobe campaigns and content.

Customer profile

Brands and retailers that prioritize owned content, digital experience management, and complex cross-channel personalization, for teams that already use Adobe for analytics, content, or campaign management.

Some retailers that use Adobe Real-Time CDP include Coca-Cola, The Home Depot, Walgreens Boots Alliance, and Prada.

Tealium

Tealium connects data across sources to create cross-channel customer profiles in real time, helping businesses better understand and engage their customers.

Key features

• AudienceStream connects customer data from different sources, builds unified profiles in real time, and sends that data to marketing and analytics tools. 

• Broad tag and integration ecosystem that supports customer journeys across marketing, analytics, and commerce tools. 

• Activate machine learning insights from high volume customer data with AudienceStream and Tealium Predict ML.

Data processing 

Captures events and updates customer profiles in real time. Works with both streaming and batch data to feed analytics and predictive models.

Customer profile

Mid-size to large retailers that want a flexible, easy-to-connect platform for unifying customer data and working smoothly with their existing tech stack.

Retailers that use Tealium include Penfolds, Dunelm, and L’Oreal as well as businesses that require strict data governance, such as Bupa.

Twilio Segment

Twilio Segment is a composable platform that collects event data at scale and routes it to hundreds of integrations, making it a strong choice for teams that want a developer-friendly CDP and easy activation across tools.

Key features

• A single view of each customer with data from every touchpoint, including the data warehouse. 

• A composable framework with flexible APIs, developer tools, and hundreds of integrations that let teams customize, extend, and scale their customer data infrastructure. 

• AI-powered workflows bring predictive insights and automated personalization.

Data processing

Collects data from every touchpoint and sends it to multiple tools at once. Supports both real-time streaming and syncing with warehouses for analysis.

Customer profile

Retailers that operate microservices or third-party tools, and those that want to centralize event-level data.

Brands and retailers using Twilio include Coca-Cola, Philips, and Panera Bread as well as digital-first, platform-based businesses like Spotify, Lyft, and Uber.

Lexer

Lexer tailors CDP capabilities specifically for retail, combining customer insight, RFM-style models, and automation workflows aimed at trading and merchandising teams.

Key features

• Retail-focused metrics and models that predict customer behavior and lifetime value. 

• Consolidates years of transactional and profile data from multiple systems into a single, clean, de-duplicated customer record. 

• Using AI and marketing models like RFM, Lexer creates intelligent metrics and predicts what customers will do next.

Data processing

Focuses on purchase history and customer profiles. Combines past data in batch and applies predictive models for merchandising and retention insights.

Customer profile

Retailers and brands that need a CDP focused on merchandizing, promotions, and shopper behavior, rather than general marketing.

Brands using Lexer include Zimmermann, Rip Curl, O’Neill, and Cotton On.

Ometria

Ometria is a built-for-retail CDP that blends customer analytics with orchestration tools aimed at email, automation, and lifecycle marketing.

Key features

• Capabilities that include exclusion rules, frequency capping, and commerce data activation. 

• Supported by a single, consistent source of predictive insight that’s trained on retail data. 

• Adapts to existing tech stacks, integrating with current marketing tech and data sources.

Data processing

Merges real-time behavior with historical purchase data to power automated marketing campaigns and predictive insights.

Customer profile

Ometria is designed for retailers that want to turn customer data into meaningful marketing action without heavy technical lift. It’s ideal for CRM, marketing, and ecommerce teams focused on driving retention and revenue through personalized, data-led campaigns.

Brands that use Ometria include Sephora, Foot Locker, Fred Petty, Hotel Chocolat, and Whittard.

How to choose a CDP

Here are some key considerations when it comes to choosing the right CDP for your business.

1. Review your tech stack. Do you need a CDP that works within a specific ecosystem like Salesforce or Adobe, or a neutral platform that connects easily with all your tools?

2. Consider where your data lives. If most of it sits in your warehouse, choose a platform that connects directly without moving or duplicating data.

3. Check governance and privacy. You need a CDP that balances speed with consent, data retention, and clear compliance controls. 

4. Test AI features. Many CDPs include built-in predictive models, but you should see how they perform, how they’re updated, and whether they can be exported or adjusted.

Customer data platforms are at the heart of modern retail tech stacks.

Choosing the right CDP is the one that fits your data landscape, your activation needs, and the operations of your marketing and merchandizing teams.

If you’re looking into CDPs but not sure which is best, we can help identify the right fit for you.

AI is not a tool for the future; it’s a tool for now.

In a short space of time, AI has become a vital part of the CRO toolkit, with teams adopting that mindset. 

That said, AI isn’t a replacement for human insight. People need to feed the right data and prompts, still need to interpret findings, and still need to apply them in a way that’s strategically aligned with broad brand goals.

AI tools in digital commerce experimentation

There’s now a universe of AI tools available, particularly for ecommerce merchandising. There’s everything from product recommendations and sort rules on PLPs to smart search optimization.  

When it comes to CRO, however, AI is particularly valuable in supporting experimentation, making it easier to design, analyze, and optimize tests. 

CRO itself is a broad umbrella that covers user testing, data analysis, funnel analysis, and optimization. Experimentation sits within that umbrella.

It’s the process of testing ideas and theories to improve website performance. The typical experimentation workflow includes several stages: analysis, hypothesis definition, building, testing, and interpreting results.

Using AI to define hypotheses

One of the most impactful ways to use AI is in the early stages of experimentation, specifically during hypothesis definition. 

Defining a clear hypothesis can be time-consuming and challenging. So using AI tools like ChatGPT or Gemini can take your data insights, funnel analysis, and research inputs to generate well-structured hypotheses.

AI can also work in reverse.

If your business has a particular strategy or focus area on your website, you can feed a potential hypothesis into the AI and ask where your data insight efforts should be concentrated.

Doing this, AI accelerates the manual work of identifying patterns and consistencies while helping you focus on the areas most likely to deliver impact.

Remember, though, that a hypothesis is an educated guess, not a guarantee. And here lies the importance of human interpretation.

Choosing the right CRO metrics with AI

AI also plays a role during the testing stage. It can help identify the most appropriate metrics to measure success.

For example, when testing changes on a PDP, like moving the “Add to Cart” button, refining product descriptions, or reducing scroll length, AI can suggest which metric best reflects the impact of those changes.

In the CRO world, much of the work involves managing stakeholder expectations and showing progress on big-picture goals, like moving the revenue needle.

But stakeholders and CRO leads often have different perspectives on what to measure. Some focus on revenue or conversion rate, while others might track add-to-cart rates. 

AI can analyze all the data, the hypothesis, the test details, and sample sizes to recommend an optimal primary metric. But again, human judgment is necessary to validate AI’s recommendation and ensure it aligns with business goals.

In this sense, AI functions as a virtual sounding board. It helps identify overlooked metrics, interprets what they could mean for the business, and suggests broader applications. 

For example, instead of tracking total revenue, focusing on add-to-bag metrics might make more sense if the test involves moving an add-to-cart button. These insights can make a test applicable across multiple pages or sections, increasing its value. 

AI also provides a safety net during experimentation. 

Feeding AI information about your test, the planned changes, and what you’re measuring can help flag risks, suggest adjustments, or indicate if your hypothesis needs refining.

It acts like a virtual colleague, helping ensure nothing is overlooked.

Supporting iteration and optimization

Experimentation rarely ends with one test. You often need to iterate, re-test, or make minor adjustments. AI can support these post-test activities.

Whether a test is a winner, loser, or inconclusive, AI can help interpret the results, suggest variations, and provide insights to drive better conversion or revenue outcomes.

This approach ensures your experimentation program continues to deliver value and evolve over time.

Top tips for using AI in experimentation

Finally, some top tips for using AI effectively in CRO and experimentation. 

1. Don’t take AI’s word as gospel

Maintain a healthy sense of skepticism and ensure human oversight. It can easily generate fake news, fabricated quotes, misinformation, and false statistics.

AI does not understand your customers, your users, your brand, or your brand’s tone of voice — the nuances driven by marketing, creative, and brand teams require human judgment. 

2. Use traditional CRO and AI-driven CRO together

Use a combination of both approaches.

Choose what is best for your testing path, your brand, and you as an optimization specialist. AI can provide shortcuts, safety nets, and sounding board moments, but it should enhance, not replace, human decision-making.

The key is to focus on doing what’s best for your customers while using AI to make experimentation more insightful and effective.

AI in CRO

AI is transforming CRO and the broader experimentation landscape, and those who embrace it thoughtfully will gain a competitive advantage.

As AI continues to evolve, it will play an increasingly central role in shaping experimentation strategies and optimizing digital commerce experiences.

Amazing things are possible in digital commerce when you look up.

If you’re looking to implement AI in your CRO practice, reach out to us.

Generative AI unlocks enormous value for brands and retailers, but it also raises very real privacy and data security risks.

To unlock hyper-personalization, accelerate content production, and unleash smarter customer service, businesses are increasingly relying on AI tools to access and analyze vast amounts of data.

Yet recent incidents show how quickly well-intentioned AI experiments can turn into breaches.

Earlier this year, a large retailer accidentally exposed the personal information of 1.8 million customers. It had used an AI tool to create personalized offers, but because it was missing access controls, hackers were able to access customer data. 

In March, a global tech company left a cloud storage area containing sensitive company data and customer information publicly accessible. Hackers used automated tools to quickly harvest the data, forcing the company to halt AI projects and conduct an emergency security review. 

In October, the government of New South Wales, Australia, accidentally exposed the personal information of up to 3,000 flood victims to ChatGPT. Data from the breach could potentially be used to train the AI platform, making sensitive information publicly searchable.  

With organizations holding more data than ever, their power comes with greater responsibility. And when a breach occurs, the consequences are costlier.

The cost of breaches

As a new IBM study has found, companies are failing to protect their AI tools from compromise, leading to an average cost of $670,000 in data breaches. 

The report also found that while only 13% of organizations had breaches involving AI, 97% of them lacked proper AI access controls. 

One of the key findings was that the most common entry point for these attacks was hackers accessing AI tools through compromised apps, plug-ins, and APIs. 

In this environment, privacy is an imperative that protects customers and the brand’s reputation.

Practical steps for protecting brand privacy in AI

Protecting brand privacy is more than avoiding breaches. It helps to build trust with customers while confidently using AI. Brands can take several concrete steps:

1. Control what data is uploaded

Only provide AI systems with the minimum data required. Avoid uploading sensitive personal information. And consider anonymizing or pseudonymizing datasets to reduce risk. 

For example, instead of uploading full customer profiles with names, email addresses, and purchase histories into an AI system, replace names with random ID numbers and mask email addresses.  

This way, the AI can still analyze purchasing patterns without exposing personally identifiable information.

2. Choose AI providers carefully

Compare privacy policies of different LLMs or AI platforms (but more on this below). 

Favor providers that offer strict access controls and no long-term retention of uploaded content. Understand whether your data might be used for model training or shared with third parties.

3. Implement strict access controls

Limit AI system access to authorized personnel only. Use role-based permissions and multi-factor authentication. And monitor usage logs to detect unusual activity quickly.

4. Secure integrations and APIs

Audit every plug-in, API, or third-party connection linked to AI tools. Ensure encryption of data in transit and at rest. Regularly test endpoints for vulnerabilities.

5. Establish internal governance and audits

Define clear policies for how AI can be used across teams. Conduct regular compliance audits to ensure AI practices meet privacy standards. Train employees on risks and responsibilities when using AI tools.

6. Plan for breach response

Have a clear incident response plan specifically for AI-related breaches. Ensure fast detection, containment, and communication to minimize impact on customers and reputation.

Comparing privacy policies of gen AI platforms

As businesses increasingly integrate generative AI into their operations, understanding the privacy policies of these platforms is crucial. 

Here’s how the major platforms handle user data:

ChatGPT 

The creators of ChatGPT, OpenAI, typically retain user data for 30 days. However, due to a court order related to a lawsuit from The New York Times, it’s currently required to store all ChatGPT conversations indefinitely, including those users request to delete. 

In terms of data usage, OpenAI uses user interactions to improve its models, with users managing their data settings through the platform’s privacy settings. 

Following the backlash of publishing conversations to search engines, OpenAI has now disabled the option to make them publicly discoverable.

Claude 

Claude, developed by Anthropic, has extended its data retention period to five years for users who opt in to model training. This applies to new or resumed chats and coding sessions. Users can change their data sharing preferences at any time in their privacy settings. 

By default, Claude uses user data to train its models. Users must manually opt out if they wish to prevent their data from being used.

Gemini 

Google retains user data for different periods, depending on the user’s settings. Users can choose to have their data retained for 3, 18, or 36 months, or opt out of data retention entirely. 

Google uses user data to improve its services and for personalization. Users can manage their data settings through their Google Account’s activity controls.

What this means for brands

These differences define how much control a brand has over its data. Choosing an AI platform isn’t only a question of capability or cost but of data sovereignty. 

Retailers and brands should ask: 

• Who ultimately owns the data once it’s uploaded? 

• Can that data be used to train someone else’s model? 

• How easily can it be deleted or audited later? 

Platforms that prioritize enterprise-grade privacy, such as offering isolated environments, encryption by default, and zero data retention, will increasingly become the gold standard for brands handling sensitive customer data. 

As privacy regulations tighten and consumer awareness grows, brands that lead on responsible AI use will differentiate themselves as trustworthy.

Building a culture of responsible AI

Data protection isn’t just about the platform you use but the people who use it. 

And that comes down to company culture. 

Every team using AI, from marketing to customer service, needs to understand the privacy implications of the tools they use. That means: 

• Embedding privacy-by-design principles in every AI project. 

• Vetting third-party tools before deployment. 

• Appointing cross-functional privacy leads to ensure alignment between legal, tech, and commercial teams. 

When privacy is treated as a shared responsibility, AI can become a force for innovation rather than a source of risk.

Looking ahead

AI’s potential is only beginning to unfold, but so are the privacy challenges that come with it. 

Brands that act now by tightening data controls and building privacy into their AI strategy will be the ones who earn customer trust and future-proof their digital operations.

If you’re looking at ways to integrate AI into your tech stack, contact us for guidance.

AI dominates conversations in retail right now. From headlines predicting mass automation to promises of limitless personalization, it’s easy to see why expectations are high.

Generative AI alone is poised to unlock between $240-$390 billion in economic value for retailers.

But the reality is that AI isn’t a magic solution; it’s a tool. One that, when applied with purpose, helps retailers reduce manual tasks, increase speed, and make better use of data.

AI’s not the solution; it’s a tool

Retailers face challenges that no algorithm can solve outright: supply chain disruption, rising customer expectations, the complexity of omnichannel, and pressure on margins.

AI won’t remove those realities. What it can do is equip retail teams with sharper tools to tackle them.

Reducing repetitive tasks

Repetition is often the hidden cost in retail operations. Catalog teams manually tag thousands of SKUs with attributes like color, size, and material.

Customer service agents handle endless variations of “where’s my order?” Merchandisers constantly refresh product data to stay aligned with availability.

These tasks are important, but they drain resources and pull teams away from work that drives growth.

AI tools can step in here without compromising accuracy. For example, computer vision models can auto-tag products with multiple attributes in seconds, ensuring consistency across catalogs.

Natural language processing can power chatbots that resolve the most common customer queries instantly, freeing human service agents to focus on more complex interactions.

Automated product enrichment engines can pull and clean product data from multiple sources, reducing human error and speeding up time-to-market.

The benefit isn’t just saving time. It’s reallocating that time. Teams can shift their energy from administration to innovation: developing campaigns, building partnerships, or testing new customer experiences.

Speeding up workflows

Retail workflows often grind down when demand spikes or when multiple teams need to collaborate under tight deadlines.

Creating seasonal product descriptions, adjusting merchandising rules, or running demand forecasts can take days or weeks. AI accelerates these processes without replacing the expertise of the people who own them.

In content production, generative AI can draft product descriptions or promotional copy at scale, while human editors refine them to ensure brand tone and compliance.

Merchandising teams can use AI-driven recommendation engines that dynamically adjust based on real-time shopping behavior. This removes the need for constant manual rule-setting.

In planning, predictive analytics can generate rolling forecasts based on live data rather than static historical reports, which helps teams adjust faster to market shifts.

The real advantage is agility, as speed means doing things faster and responding sooner to opportunities and risks. For example, launching a product range earlier, pivoting a campaign mid-flight, optimizing stock allocations before peak periods.

Unlocking hidden value in data

Data has always been one of retail’s strongest assets, but the volume and variety now exceed human capacity to interpret at scale.

Every customer touchpoint (browsing online, engaging on social, or purchasing in-store) creates data. But without the right tools, most of it sits unused.

AI helps surface the value buried in that data.

For instance, algorithms can segment customers based on actual behavior, not just broad demographics, enabling more precise targeting.

Machine learning models can spot early signs of churn in loyalty programs and trigger tailored re-engagement campaigns.

AI-powered analytics tools can process millions of data points across sales and returns, surfacing patterns that humans may miss, like identifying which products are most likely to sell together in different regions.

The impact goes beyond marketing. Finance teams can use AI to model margin pressures more accurately. Operations can optimize logistics by predicting demand at store level.

The result is not more data, but more actionable insights.

Let’s distill this into what AI can do for specific teams:

For customer service teams 

Chatbots powered by AI handle routine queries like order tracking and returns. They don’t replace service teams, but they cut down response times and free human agents to deal with complex cases.

For merchandizing teams

Visual search and personalized recommendations use AI to connect customers with products faster, driving conversion without demanding extra manual input from merchandisers.

For operations teams 

Demand forecasting powered by AI reduces stockouts and overstocking. It doesn’t eliminate supply chain risks, but it improves accuracy in planning.

For marketing teams  

AI enables dynamic content and audience segmentation at scale, with AI handling the volume while the brand defines the strategy.

Why it matters that AI is a tool

A big part of the narrative is that artificial intelligence is a solution in itself, a plug-in-and-play human replacement.

But this mindset creates two risks: an overreliance on technology and undervaluing human expertise. 

1. Overreliance. Retailers who expect AI to deliver transformation on its own will be disappointed. AI is only as effective as the data it is fed and the business goals it supports. 

2. Undervaluing human expertise. From buyers curating collections to associates building trust with customers, people have always been at the heart of retail. And AI doesn’t replace that role but helps to amplify it. Consumers continue to seek out human connection when interacting with their favorite brands, especially when it comes to complex decisions or queries.

Final thought

AI won’t “fix” retail challenges. It won’t erase supply chain challenges, automate loyalty, or replace the creativity and judgment of human teams.

However, what it does do is enhance the tools.

What it will do is give retailers better tools to meet the challenges of today and build the customer experiences of tomorrow.

Retailers that apply AI with purpose and a clear understanding that it’s a tool, not a solution, will have the best chance of long-term success.

Looking to embed AI into your technology stack? Contact us.