AI Shopping for Small Businesses: How Product Discovery Is Changing for Online Shops
Updated: Sep 10
Online shopping used to have a fairly predictable path.
A customer searched Google.
They opened several websites.
Compared products.
Checked prices.
Read reviews.
Then decided what to buy.
AI assistants are beginning to change that journey.
A shopper can increasingly describe what they want conversationally:
“I need a birthday gift under £40 for somebody who loves gardening but already seems to own everything.”
Instead of simply returning ten blue links, an AI assistant may research the request, compare possible products and present a much smaller shortlist.
That creates an important question for online retailers:
Does an AI system have enough accurate information about your products to understand when they are relevant?
For most small online shops, the answer is not to start chasing new “AI optimisation” tricks.
It is to make the fundamentals of the store considerably clearer.
AI Shopping for Small Businesses: The Quick Answer
AI shopping for small businesses means AI assistants are becoming another way customers can discover, compare and potentially choose products.
This does not mean conventional ecommerce, Google Search or the retailer's own website are disappearing.
It means some of the research and comparison that previously happened after a customer reached your website may begin happening before they arrive.
That makes information such as:
product names
descriptions
price
availability
variants
photographs
delivery
returns
reviews
specifications
product identifiers where appropriate
clear product categories
increasingly important.
The objective is simple:
Make it easy for customers, search engines and emerging AI shopping systems to understand exactly what you sell.
Our guide to AI search visibility for small businesses explains the broader shift from conventional search rankings towards recommendations.
This article concentrates specifically on what that means when the customer is buying a product.
What Is AI Shopping?
AI shopping describes the growing use of AI assistants to help customers:
discover products
research options
compare alternatives
narrow a shortlist
understand specifications
compare prices or features
decide which product might suit a particular need
move towards a purchase
Microsoft Advertising describes this as a shift in which AI increasingly sits between the retailer and the shopper.
Its August 2026 guidance argues that complete, structured product information becomes particularly important because an AI system needs factual information it can understand before confidently recommending an item.
That does not mean every AI assistant operates identically.
And it certainly does not mean every purchase will suddenly be made through an AI agent.
But the direction is worth paying attention to.
A customer journey that previously looked like:
Google search
↓
Open several shops
↓
Browse products
↓
Compare
↓
Buy
may increasingly look like:
Ask an AI assistant
↓
AI researches possible products
↓
AI narrows the shortlist
↓
Customer investigates one or two options
↓
Buy
The important change is what happens before the retailer receives the visit.
AI May Narrow the Shortlist Before Somebody Reaches Your Shop
Imagine two businesses selling similar products.
Business A has a beautiful website but its product information consists mostly of:
“The perfect gift for every occasion.”
Price information is unclear.
Delivery times are difficult to find.
Variants are poorly explained.
Availability is uncertain.
There are few useful product details.
Business B clearly explains:
exactly what the product is
who it might suit
available sizes or options
price
availability
delivery times
materials or specifications
returns
genuine reviews
common customer questions
Which business gives an automated system more factual information to work with?
Business B.
More importantly, which business gives a human customer more useful information?
Also Business B.
That overlap matters.
Many sensible improvements for AI shopping are simply good ecommerce practice implemented properly.
Your Product Data Is Becoming Part of Your Marketing
Product data sounds technical.
But in simple terms it is the factual information describing what you sell.
For example:
Product name: Personalised Birthday Cartoon Video
Price: £29.99
Availability: Available to order
Delivery: Digital delivery within the stated turnaround
Options: Different packages or lengths
Description: What the customer actually receives
For a physical retailer it might also include:
colour
size
dimensions
brand
model
material
condition
stock status
delivery cost
product identifiers
Microsoft's current AI-shopping guidance specifically emphasises complete product catalogues and structured product data rather than giving AI systems only a limited subset of the products available.
Google has been making the same underlying requirement important for ecommerce for years.
Its Merchant Center product-data specification uses structured attributes for details including titles, descriptions, price, availability, images, variants, delivery information and identifiers. Google also says accurate product information helps it match products with relevant searches and supports AI-powered advertising formats and experiences.
The practical lesson is not:
“Create a product feed because AI says so.”
It is:
Treat accurate product information as a core business asset rather than an administrative afterthought.
Google Is Starting to Show Retailers How Their Products Perform in AI Shopping
One sign of how quickly product discovery is changing is that Google has started giving some retailers dedicated information about how their products appear within its generative AI shopping experiences.
Google Merchant Center now includes an AI Performance Insights report for eligible retailers.
The report is designed around conversational shopping queries rather than simply conventional keyword searches.
That matters because somebody using an AI shopping experience might ask:
“What are good waterproof walking shoes for somebody with wide feet who walks ten miles at weekends?”
rather than simply searching:
“walking shoes”
Google's reporting is beginning to show retailers how their products perform within those more detailed shopping journeys.
What Does Google's AI Performance Report Measure?
The report includes several useful measurements.
Your share of voice
This measures how much AI visibility your brand or products receive compared with the competitors Google has identified within Merchant Center.
It is effectively asking:
When relevant AI shopping results appear, how much of that visibility are our products receiving?
Google also provides a competitor-average share so retailers can compare their visibility with the wider competitive group.
Frequency
Google uses frequency to indicate how popular particular search types, terms, customer intents or product attributes are.
This can help identify areas where customer interest is high but your product visibility is relatively weak.
Products showing
Retailers can see how many of their products are appearing for particular terms, attributes and search intents.
That creates a useful distinction between:
we sell products relevant to this need
and:
Google is actually surfacing those products for this type of shopper.
Google Is Measuring Three Different Shopping Stages
The report divides conversational shopping searches into three stages.
Discovery
The shopper is exploring possible products or solutions.
For example:
“What kind of running shoes are best for wet winter conditions?”
Evaluation
The shopper is comparing products, specifications or alternatives.
For example:
“Which waterproof trail shoes have the best grip but are still lightweight?”
Ready to buy
The customer is considerably closer to making a transaction.
For example:
“Where can I buy these shoes in size 10?”
This distinction is useful even beyond Google's reporting.
It reinforces something that applies to all ecommerce websites:
different shoppers need different information depending on where they are in the buying journey.
A customer discovering a product category needs different information from somebody who already knows the model, colour and size they want.
The Report Can Reveal Missing Product Information
One particularly practical part of Google's reporting is its optimisation information.
Merchant Center can expose:
top terms
popular product attributes
top search intents
Popular attributes are especially interesting.
Imagine customers repeatedly include:
material
size
colour
water resistance
capacity
or:
compatibility
in their shopping questions.
If those attributes matter to customers but are missing from your product data, that gives you something specific to investigate.
Google recommends keeping high-quality product information up to date and, where appropriate, incorporating relevant terms and missing attributes identified through the report into product titles, descriptions and product data.
The lesson is not:
“Stuff every popular term into the product title.”
It is:
“If customers care about an important product fact, make sure you actually provide it.”
That improves the information available to Google.
More importantly, it improves the information available to the customer.
It Is Not Currently Available to UK Retailers
There is an important limitation for BrightPath readers in the UK.
At the time of writing, Google's AI Performance Insights are currently available for English-language queries for eligible Merchant Center accounts in:
Australia
Canada
India
New Zealand
United States
The UK is not currently included.
That means a UK retailer should not log into Merchant Center, fail to find the report and conclude that something has been configured incorrectly.
Availability may expand in future, but businesses should work from Google's current documentation rather than assuming that a feature announced elsewhere is already available here.
Why Should a UK Online Shop Care Yet?
Because the things Google has chosen to measure tell us something useful about the direction of ecommerce search.
Google is explicitly analysing:
customer intent
↓
shopping stage
↓
product attributes
↓
products surfaced
↓
brand visibility
↓
competitive share of voice
That reinforces the central point of this article.
AI shopping is not simply about putting a product name into another search box.
Customers can describe what they need in increasingly specific, conversational ways.
For a retailer, that makes accurate and complete product information increasingly valuable.
Even without access to Google's report yet, a UK online shop can ask:
Are important product attributes clearly provided?
Does the product description explain genuine use cases?
Can customers compare options?
Are variants properly identified?
Are price and availability accurate?
Does the feed agree with the website?
Can somebody understand who the product is suited to?
Do product categories reflect real customer needs?
Those are useful ecommerce improvements today.
You do not need to wait for another dashboard before making them.
Product Feeds Do Not Replace Good Product Pages
An online shop can have excellent machine-readable product data and still provide a terrible customer experience.
Customers need more than database fields.
A good product page should still answer:
What exactly is this?
Do not make customers decode a clever product name.
Who is it for?
Explain the relevant customer, use case, problem or occasion.
What do I actually receive?
Be specific.
What does it cost?
Make pricing easy to understand.
What options are available?
Size, colour, package, quantity, personalisation or other variants should be clear.
When will I receive it?
Delivery expectations matter.
Can I return it?
Explain relevant returns or cancellation information clearly.
Why should I trust this shop?
Reviews, real photographs, clear policies, genuine business information and evidence all help.
What do I do next?
The buying action should be obvious.
This connects directly with our guide to what pages a small business website actually needs.
For ecommerce businesses, product pages, categories, delivery information, returns, FAQs and contact details all contribute to the buying journey.
Make the Visible Page and Product Data Agree
Consistency becomes particularly important.
Imagine your product feed says:
£29.99 — in stock
while your website says:
£34.99 — currently unavailable
That creates a problem.
Google explicitly warns that inaccurate, missing or conflicting product information can affect eligibility or how products are displayed through Merchant Center.
The same principle is sensible more broadly.
Keep important information aligned across:
product pages
ecommerce platform
product feeds
structured data
Merchant Center
marketplaces
advertising platforms
Do not allow your product catalogue to become several conflicting versions of reality.
Structured Data Can Help Machines Understand Products
Structured data is information embedded within a website in a machine-readable format.
Customers normally do not see it.
Search engines and other systems can use it to interpret information more reliably.
For a product, structured information can communicate things such as:
product name
description
image
price
currency
availability
ratings
offers
Google explains that Product structured data can be used alongside Merchant Center data and can help systems retrieve current product information from website pages.
Do not treat schema as magic.
Adding Product structured data to a weak product page will not suddenly turn it into the world's best ecommerce listing.
Think of it as:
clear customer-facing information
↓
clear machine-readable information
rather than choosing one or the other.
Do Not Write Product Descriptions Only for Algorithms
AI shopping creates an obvious temptation:
“How do I write my product descriptions so ChatGPT recommends them?”
That is the wrong starting point.
A customer still has to want the product.
Avoid filling descriptions with repetitive phrases such as:
“best personalised birthday gift”
“best personalised birthday present”
“unique personalised birthday gift”
“personalised gift for birthdays”
over and over.
Instead, provide useful detail.
Explain:
what makes the item distinctive
materials
dimensions
intended use
who it suits
what's included
limitations
compatibility
how personalisation works
delivery expectations
care instructions
genuine examples
Specific information helps customers make decisions.
It also gives discovery systems more substance to interpret.
Generic Product Copy Is Becoming Easier to Ignore
AI can generate respectable product copy in seconds.
That means generic wording is becoming incredibly cheap.
Consider:
“Our premium-quality product combines exceptional craftsmanship with timeless style.”
What did you actually learn?
Very little.
Compare that with:
“The frame measures 30cm × 20cm, uses solid oak and can be personalised with up to three lines of text.”
That is information.
The second description is useful because it contains facts.
AI can help organise your genuine product knowledge.
It should not remove the details that make the product real.
Images Still Matter
AI systems may increasingly use structured facts.
Humans still buy with their eyes.
For ecommerce, use useful images rather than simply having an image because the platform requires one.
Depending on the product, customers may benefit from:
clear main product image
multiple angles
scale or size reference
close-up details
packaging
product in use
available colours
personalisation examples
genuine customer examples
Poor imagery creates uncertainty.
And uncertainty kills purchases.
Google's Merchant Center guidance treats image information as a fundamental component of product listings, reinforcing the importance of maintaining usable product imagery as part of the overall product dataset.
Delivery and Returns Are Product Information Too
Online shops often concentrate heavily on describing the product and then bury practical questions.
Customers may be equally concerned about:
Will it arrive before Friday?
How much is delivery?
Can I return it?
What happens if it is personalised?
Do you ship to my location?
When will the order be dispatched?
Those questions affect whether somebody buys.
So delivery and returns information should not be treated as irrelevant legal small print.
They are part of the buying decision.
Make them easy to find.
Reviews and Trust May Become Even More Important
AI may help a customer identify products.
It does not remove the need for trust.
A shopper may still ask:
Is this business genuine?
Is the product actually good?
Will it arrive?
What happens if something goes wrong?
Do other customers recommend it?
Are these photographs real?
Does the company provide clear contact information?
Our guide to AI and customer trust explains the broader principle: polished marketing claims are easy to produce, but genuine evidence is harder to fake.
For an online shop, useful evidence can include:
genuine reviews
customer photographs
clear contact details
delivery information
returns information
secure checkout
realistic product photography
real examples
transparent pricing
clear business information
AI might help a product make the shortlist.
Trust still influences whether the customer actually buys.
Unusual Products Need Even More Explanation
This is something we have experienced directly through Send-A-Scene.
BrightPath Digital created, owns and continues to operate the online business.
Send-A-Scene sells personalised cartoon video gifts.
That is not a product every customer already understands.
Somebody discovering it needs answers to questions such as:
What exactly am I buying?
What photographs do I need?
What information do I provide?
What will the finished result look like?
How much does it cost?
How long does it take?
What happens after ordering?
Those questions influenced the structure of the website and buying journey.
We explain the experience in Building an Online Business: How We Built Send-A-Scene.
This becomes especially relevant to AI shopping.
If your product is unusual, personalised, handmade or difficult to categorise, do not assume a discovery system—or customer—will automatically understand it from the product name.
Explain it.
Categories Matter as Well as Individual Products
Suppose you sell 500 products.
Customers do not always know the precise item they want.
They may begin with:
gifts for gardeners
running shoes for wet weather
birthday presents under £30
office chairs for small rooms
vegan skincare gifts
That means clear product categories can help organise the catalogue around genuine customer needs.
Useful categories also help visitors who land somewhere in the middle of the buying journey.
Avoid creating hundreds of near-identical categories purely to capture slightly different keywords.
Create categories because they genuinely help customers navigate and understand the range.
AI Shopping Makes Every Product Page More Like a Landing Page
A customer arriving from AI may not enter through your homepage.
They could land directly on:
one product
one category
a comparison
a guide
a FAQ
a review
a collection page
Our guide to AI referral traffic makes the same point across AI-assisted discovery more generally: AI visitors may arrive halfway through the buying journey.
That means an important product page should make sense independently.
Can somebody arriving directly determine:
who is selling it
what it is
price
important options
delivery
returns
why they should trust the business
what happens next
Do not assume they will visit your homepage first.
What About AI Buying Without Visiting the Website?
This is where ecommerce may change more significantly.
Microsoft is already developing experiences where AI can help shoppers move further through the buying process, including Copilot Checkout for participating merchants. Its current commerce strategy is explicitly built around getting products discovered, helping them get chosen and measuring what happens afterwards.
For a small retailer, there is no reason to panic.
You do not need to rebuild your entire business around hypothetical AI agents.
But you should recognise the direction.
The website may no longer be the only interface through which product information is consumed.
That makes accurate underlying information more important.
Not less.
How Should a Small Online Shop Measure AI Shopping?
Do not limit measurement to:
“How many people clicked ChatGPT?”
AI influence can be less direct.
Merchant Center AI Performance Insights
Where Google's AI Performance Insights report is available, eligible retailers can use Merchant Center to assess product visibility within AI Mode and AI Overviews.
That can include share of voice, products showing, shopping stages, search intents, popular attributes and high-frequency terms.
For UK retailers, the report is not currently available, so do not expect to see it in Merchant Center yet.
And remember what the report measures.
AI visibility is not automatically a sale.
Use this information alongside website visits, product engagement, add-to-cart activity, purchases and customer feedback rather than treating share of voice as the final commercial result.
Useful signals include:
AI referral traffic
Are identifiable visitors arriving from AI assistants?
Product landing pages
Which products are receiving those visits?
Engagement
Do those visitors stay, scroll, view images or interact?
Add-to-cart activity
Are they moving towards a purchase?
Sales
Does identifiable AI traffic actually convert?
Branded search
Are more people searching specifically for the shop or product after discovering it elsewhere?
Customer feedback
Where practical, ask customers how they found you.
Our detailed guide to AI referral traffic and measurement explains why crawling, citations, referrals and conversions should be treated as separate stages.
Microsoft Advertising also now points retailers towards AI visibility and referral measurement rather than simply counting crawler activity.
Do AI-Referred Shoppers Convert Better?
Possibly.
But do not take one industry statistic and apply it automatically to your own shop.
Microsoft cites Adobe data showing that AI-referred visitors converted more strongly than non-AI traffic within the dataset being referenced.
That is interesting.
It is not a guarantee that your AI traffic will behave the same way.
Measure your own outcomes.
Three AI visitors who buy may be more valuable than 300 irrelevant visitors.
The commercial question is not:
“Are we getting AI traffic?”
It is:
“Is this discovery producing useful customers?”
What Should Small Online Shops Do Now?
You do not need a 50-point “AI commerce transformation programme”.
Start with the basics.
1. Review your most important products
Pick the products that matter commercially.
Check whether the information is complete.
2. Improve vague descriptions
Replace generic marketing language with useful facts.
3. Check price and availability consistency
Make sure the visible website and underlying product systems agree.
4. Review product images
Ask whether the imagery answers customer questions rather than simply looking attractive.
5. Make delivery and returns obvious
Reduce uncertainty before checkout.
6. Review categories
Make sure products are organised in ways customers genuinely understand.
7. Check mobile pages
Many shoppers will still complete their research on a phone.
8. Review structured product information
Where your ecommerce platform supports it, make sure Product structured data and feeds are working sensibly.
9. Look for genuine proof
Reviews, examples and real customer evidence matter.
10. Measure outcomes
Track visits, engagement, add-to-cart behaviour and purchases rather than celebrating visibility alone.
A 15-Minute AI Shopping Check
Choose five important products.
For each one, answer:
Is the product name completely clear?
Does the description explain what it actually is?
Is the price obvious?
Is availability accurate?
Are options or variants clear?
Is delivery easy to understand?
Are returns explained?
Are the photographs genuinely useful?
Is there evidence that customers trust the product or business?
Could somebody understand the product without first visiting the homepage?
Does the structured product information match the visible page?
Is the next buying action obvious?
If several answers are no, you have useful work to do regardless of how quickly AI shopping develops.
What Should You Avoid?
Do not create hundreds of thin AI-written product pages
More pages do not automatically create more visibility.
Do not invent specifications
Product information needs to be accurate.
Do not hide important information
Price, availability, delivery and options should not require detective work.
Do not chase every new AI-shopping tool
The underlying platforms will change rapidly.
Do not neglect conventional search
AI shopping is an additional discovery route, not permission to abandon Google Search, Merchant Center, SEO or good ecommerce fundamentals.
Do not forget the actual customer
If an optimisation makes the page easier for machines but worse for people, reconsider it.
Frequently Asked Questions
What Is AI Shopping?
AI shopping is the use of AI assistants and AI-powered search experiences to help customers discover, compare, research or buy products.
The amount of the journey handled by AI varies by platform.
Can ChatGPT or Copilot Recommend Products?
AI assistants can provide product suggestions or comparisons when answering relevant user requests.
Which products appear can vary according to the question, available information, platform and sources being used.
Do I Need an AI Product Feed?
Do not assume every AI platform requires the same feed.
For established commerce ecosystems such as Google Merchant Center and Microsoft Merchant Center, accurate structured product data already plays an important role in product discovery.
Start by keeping the feeds and product information you already control complete and accurate.
Is Product Structured Data Important?
It can help search engines and other systems interpret product information such as price, availability and offers.
It should complement a strong product page rather than compensate for a poor one.
Should I Rewrite All My Product Descriptions for AI?
No.
Rewrite weak descriptions because customers need better information.
The best descriptions clearly explain the real product rather than repeating keywords or trying to predict a secret AI-ranking formula.
Will AI Replace Ecommerce Websites?
There is no good reason for a small retailer to assume that.
What is changing is that more product discovery, comparison and potentially transaction activity may take place through external AI interfaces.
Your underlying website, product information, fulfilment, trust and customer experience remain fundamental.
How Can I Tell Whether AI Is Sending Customers?
Use analytics to identify AI referral traffic where possible, review the pages those visitors reach, measure shopping actions and ask customers how they discovered you where appropriate.
Do not confuse AI crawler activity with human customers.
Final Thought: Make Your Products Easy to Understand
The strongest preparation for AI shopping is not a clever optimisation hack.
It is clarity.
Make products easy to:
find
↓
understand
↓
compare
↓
trust
↓
buy
That works for Google.
It works for AI assistants.
Most importantly, it works for customers.
If your online shop already exists but you are unsure whether the public-facing pages make products easy enough to understand, trust and act on, BrightPath Digital's Website Audit & Improve service reviews website clarity, customer journeys, mobile experience, trust signals, calls to action and visible barriers that may be holding the site back.
Comments