Cost of Living and Comparison Shopping Trends

The UK has been in an extended cost of living squeeze. Inflation has eased from its peak, but prices remain meaningfully higher than they were three years ago, and household budgets have not recovered. People are not spending less overall - they are spending more carefully. That shift in behaviour has direct implications for how product data services and AI agents can be useful.

The spending picture

The broad trends are consistent across multiple sources. A majority of UK adults report that their cost of living has risen over the past year, with food and energy cited most frequently as the pressure points. Material deprivation has increased. A significant proportion of adults cannot afford an unexpected expense of a few hundred pounds.

But people have not stopped buying things. They are still purchasing electronics, switching energy and broadband providers, and maintaining their homes. What has changed is how they buy. More people are setting budgets. More people are comparing prices before committing. More people are actively looking for the best deal rather than defaulting to a familiar retailer.

This is not a temporary blip. The comparison shopping habit, once established, tends to persist even as financial pressure eases. People who learn to check prices across three retailers before buying a laptop do not stop doing it when their income improves. The behaviour becomes normal.

What this means for agents

The rise of comparison shopping creates a natural opportunity for AI agents. The core task - check multiple sources, compare prices, find the best deal for a specific product - is repetitive, structured, and tedious for humans. It is exactly the kind of work that agents handle well, provided they have access to reliable data.

When someone asks an agent “where is the cheapest place to buy this TV?”, the agent needs to check current prices across multiple retailers, confirm stock availability, and present the options clearly. If the agent can do this reliably, it saves the user time and often money. In a cost-conscious environment, that combination is genuinely valuable.

The demand signal is not speculative. Comparison shopping volumes in the UK have been rising steadily across electronics, insurance, energy, and broadband. Price comparison websites have seen increased traffic. The underlying behaviour - wanting to know you are getting a fair deal before you commit - is broadly distributed across income levels, not limited to those in financial difficulty.

Where the current tools fall short

Traditional price comparison websites were built for this moment, but they are showing their age. They are designed for humans using browsers: visit the site, type a search query, scroll through results, click through to a retailer. The workflow is manual, time-consuming, and does not integrate well with the AI agent layer that is increasingly sitting between people and their purchases.

For agents, these comparison websites present the same data access problems as regular retail sites. The information is rendered for visual consumption, not structured for machine parsing. Prices are embedded in HTML. Availability data is generated client-side. Category structures are optimised for human navigation, not programmatic filtering.

The gap is not in the availability of comparison data - it exists, across multiple services. The gap is in the accessibility of that data to machines. Agents need structured, queryable, timestamped product data delivered through APIs, not web pages.

The agent-mediated future

The convergence is straightforward. People want to compare prices more than they used to. AI agents are increasingly capable of handling that task. But the data infrastructure connecting agents to reliable product information is underdeveloped.

Filling that gap does not require building a better comparison website. It requires building a data layer that agents can query directly - one that normalises product information across merchants, maintains freshness, and delivers structured responses that agents can process without scraping or parsing.

The cost of living environment is not the reason to build this kind of service, but it sharpens the value proposition. When every purchase decision matters a bit more, the accuracy and reliability of the data behind that decision matters a bit more too. An agent that can confidently tell someone “this is the cheapest place to buy this product right now, and the price was checked two hours ago” is providing something genuinely useful - and that usefulness is only increasing.