What happened in the UK’s Golden Quarter
Salesforce’s end-of-year Shopping Insights report indicates that global online sales during the 2025 Golden Quarter reached a record 1.29 trillion dollars, with the UK accounting for 38 billion dollars (around 28 billion pounds) in ecommerce revenue. UK online sales grew 5.5% year on year, driven by a 9% increase in average selling prices and a 10% uplift in traffic, despite an overall cautious consumer environment. InternetRetailing.
A key finding of the report is the disproportionate role of AI. Around 20% of retail sales globally were influenced by AI and agents, representing 262 billion dollars in spend. Third‑party AI shopping channels and AI-powered search showed markedly higher intent: traffic coming via these sources converted roughly nine times better than social referrals. Retailers deploying their own AI agents saw revenue growth close to 60% higher than peers, and AI agents also absorbed a 142% surge in operational tasks such as returns and shipping updates. InternetRetailing; Salesforce.
At the same time, UK physical retail underperformed. According to the British Retail Consortium, December footfall declined across high streets and shopping centres, and in-store non‑food sales grew only 0.4% year on year, while online non‑food sales rose 11.1%. InternetRetailing. Parallel ONS and industry commentary on the Golden Quarter highlights subdued overall retail growth, selective promotional response to Black Friday, and a continued structural shift toward online, with online penetration in November climbing to the highest level since late 2021. PwC; FashionUnited.
The picture is more nuanced in grocery, which is largely outside Salesforce’s non‑food lens. Grocery was one of the few bright spots in UK retail over Christmas, supported by festive food spending and inflation; online penetration here remains around the low‑ to mid‑teens during peak periods while 85–87% of spend still occurs in-store. InternetRetailing. As a result, the quarter can be described as a “two-speed” season: AI‑amplified digital channels expanding in value and efficiency, versus physical formats posting marginal or negative real growth.
Why this Golden Quarter matters for ecommerce infrastructure
The 2025 Golden Quarter does not just confirm the strength of online demand; it crystallises a structural change in how demand is generated and mediated. Traffic and revenue are increasingly routed through AI agents, conversational interfaces and intent-driven discovery rather than traditional search, paid media or social. This reconfiguration has direct implications for product data structures, catalog operations and content pipelines.
AI-driven demand and product feeds
If 20% of global retail sales are already influenced by AI agents, product feeds are no longer just inputs for ad platforms and marketplaces; they are the core “language” that AI systems use to understand, rank and recommend inventory. In this environment:
Feed completeness becomes a commercial risk factor. AI systems rely on structured attributes (size, material, colour, usage context, compatibility, sustainability flags, price history) to infer relevance and personalise rankings. Missing or inconsistent data reduces the probability of a product being surfaced in high‑intent AI interactions, even if it performs well in traditional search.
Real‑time and event-aware feeds gain priority. Golden Quarter trading is characterised by rapid price changes, flash promotions and fast‑moving stock. For an AI agent that is expected to negotiate constraints (budget, delivery deadlines, brand preferences) on behalf of the shopper, stale inventory or pricing data degrades results and trust. That shifts investment from batch feed exports toward low-latency APIs, event‑driven updates and tighter integration between merchandising systems and AI layers.
Multi-channel feed governance becomes more complex. AI referrals increasingly arrive from sources that sit outside classic “walled gardens”: independent agents, general-purpose conversational tools, and retailer-owned assistants. All of these consume product and offer data in slightly different ways. Standardising taxonomy, attribute sets and update schedules across channels becomes a prerequisite for consistent performance. Product Feeds becomes a prerequisite for consistent performance.
Catalog standards and semantic structure
The strong outperformance of AI-influenced traffic underscores the importance of machine-readable catalog standards. The more sales move through agentic experiences, the less tolerant the ecosystem becomes of noisy or unstructured catalogs.
Several trends are reinforced by this Golden Quarter:
From keyword orientation to semantic orientation. AI agents interpret user intents that are naturally expressed in goals (“find a winter coat that’s warm but under £150 and delivered by Friday”) rather than product names. Catalogs must expose attributes and relationships that map to these intents: warmth ratings, insulation type, delivery promise by postcode, care requirements and so on. Where such data is unstructured or buried in marketing copy, agents have to infer it, raising the risk of irrelevant or non-compliant recommendations.
Normalisation across brands and categories. As agents compare items from multiple retailers in a single conversation, inconsistent attribute naming (e.g. “navy” vs “midnight blue”, mixed sizing conventions) makes cross‑retailer comparison harder. This creates an incentive for sector-wide harmonisation of attribute definitions, unit standards and value lists, and for more rigorous internal data dictionaries at retailer level.
Lifecycle and policy metadata. With agents now handling a sharply higher volume of service tasks such as returns, exchanges and shipping updates, catalogs need to incorporate structured policy and lifecycle data: return windows by product type, restocking fees, refurbishment status, warranty terms. When these attributes are explicit and standardised, agents can answer operational questions and pre‑empt friction before checkout.
Product detail pages: quality, completeness and machine-readability
The shift toward AI-mediated discovery changes the function of the product detail page (PDP). Human-facing storytelling remains essential, but PDPs increasingly operate as “source of truth” for AI models that parse content at scale.
In the context of this Golden Quarter:
High-converting AI referrals accentuate the value of complete PDPs. Since AI tends to direct already-qualified intent, the bottleneck often lies in resolving the last uncertainties: fit, compatibility, care, bundled items, or in confirming return and delivery conditions. Retailers that expose this information clearly and consistently – both for humans and machines – are better positioned to capitalise on that intent.
Rich media becomes structured input. Images, video and user-generated content have traditionally boosted conversion through human persuasion. As computer vision and multimodal models are incorporated into shopping agents, these assets also become data sources. Clean tagging of imagery (angles, context of use, model measurements) and consistent metadata around videos or guides enables agents to answer visual or stylistic queries with more precision.
Review and Q&A content is a training signal. User reviews and Q&A sections now inform not only human perception but model understanding of product strengths, weaknesses and real-world usage. Retailers gain leverage by moderating, tagging and structuring this content – for example, summarising recurring themes, surfacing frequently asked questions as explicit attributes, and ensuring that key clarifications propagate back into base product data.
Speed of assortment deployment and seasonal agility
With UK online sales growing faster than total retail and non‑food ecommerce up by double digits over Christmas, time-to-market for new SKUs becomes even more critical. AI amplifies this dynamic rather than relaxing it.
The Golden Quarter data points suggest several operational pressures:
Shorter content lead times. To capitalise on peak demand windows, retailers must be able to ingest supplier data, enrich it, and deploy live PDPs in days rather than weeks. Manual copywriting and studio-heavy workflows struggle to keep up with this cadence, especially across long-tail assortments and seasonal capsules.
Dynamic assortment curation. AI agents that understand basket-level context and stated constraints can steer shoppers toward alternative SKUs when primary items are out of stock or fail on a constraint such as delivery time. For this to work, range architecture, substitution rules and compatibility metadata must be codified in systems rather than left to ad hoc merchandising decisions.
Pricing and promo elasticity experiments. Higher average selling prices and selective promotional response during the Golden Quarter point to shoppers trading off value more carefully. Embedding experimentation frameworks into pricing and content layers – for example, testing different bundles, benefit framings or threshold offers – requires close coupling between pricing engines, content management and AI-driven personalisation.
No-code, AI and the industrialisation of content operations
The efficiency narrative emerging from the Salesforce data – agents handling a large increase in service tasks and supporting above-average revenue growth – highlights a deeper shift: the industrialisation of ecommerce content and operations through no-code and AI tooling.
Several patterns stand out:
Automation of repetitive content tasks. AI is increasingly used to generate first drafts of titles, descriptions and SEO text, to localise content, to fill missing attributes from supplier PDFs or images, and to standardise tone of voice. No-code interfaces allow merchandisers to specify rules and workflows (e.g. which attributes to prioritise by category, how to handle regulatory phrases) without developer intervention, compressing cycle times while retaining editorial oversight.
Rule-based quality assurance. As catalogs grow and AI mediates more journeys, automated checks on completeness, consistency and compliance become necessary. No-code orchestration and AI validation models can flag products that lack critical attributes for AI ranking, that deviate from taxonomy rules, or that contain contradictory policy information, pushing only “AI-ready” items into high-visibility feeds.
Operational agents as a backbone. With agents already handling a surge in returns and shipping queries, the next step is their integration deeper into back-office systems: inventory, order management, customer communication and content repositories. This enables, for example, automatic creation or update of PDP content based on return reasons, or dynamic adjustment of onsite messaging in response to logistics constraints.
Democratisation of experimentation. No-code tools make it easier for ecommerce and content teams to configure and test variations in product presentation, navigation, filters and editorial storytelling without waiting for development sprints. When combined with AI‑driven segmentation, this allows for continuous optimisation aligned with the behavioural shifts observed in the Golden Quarter.
Strategic implications for ecommerce and content infrastructure
The 2025 Golden Quarter positions AI not as a peripheral optimisation tool but as a central layer mediating demand, discovery and operations. For ecommerce players, the data points from this period translate into a set of infrastructure priorities:
- Treat product data as a first-class asset, structured for machines as carefully as it is designed for humans.
- Invest in catalog standards and governance capable of supporting cross-channel, multi-agent consumption.
- Rebuild PDPs and auxiliary content with dual audiences in mind: human shoppers and AI systems that will increasingly pre‑filter and pre‑explain options.
- Shorten and automate the content supply chain from supplier feed to live PDP, using AI and no-code to maintain quality at scale.
- Embed AI agents into both customer-facing and operational workflows, viewing them as a connective tissue between content, commerce and logistics rather than as isolated chat widgets.
In that sense, the Golden Quarter serves as a live test of an emerging retail model in which growth is increasingly contingent on how well a retailer’s content and catalog infrastructure “speaks” to AI systems that now influence a substantial share of spend. Understanding the impact on your product data management is key to success.
The insights from the Golden Quarter underscore the critical need for robust product data management. As AI becomes the primary driver of discovery and sales, the quality, completeness, and structure of product information are paramount. At NotPIM, we recognize this shift and offer a comprehensive solution that allows e-commerce businesses to streamline feed management, enrich product data, and adapt to the evolving demands of AI-driven commerce. Companies that prioritize their data infrastructure are poised to gain a significant competitive advantage.