How to discover the product comparisons shoppers make before purchase

Ecommerce teams can discover the product comparisons shoppers make before purchase by analyzing the questions they ask in search, chat, support, reviews, and sales conversations—not just clicks and conversion rates. Group those questions by alternative, use case, and tradeoff, then use the patterns to improve product data, comparison content, recommendations, navigation, and merchandising tests.

Start with the questions shoppers ask

The fastest way to find comparison behavior is to collect the language shoppers use while they are deciding. A question such as “Which jacket is warmer?” exposes a decision criterion that a pageview cannot.

Useful sources include:

• On-site search queries and refinements
• Questions asked in a conversational product assistant
• Customer support tickets, chat transcripts, and contact forms
• Product reviews and questions-and-answers
• Sales-call notes for higher-consideration products
• Queries that lead to repeated visits without a purchase

Treat each question as evidence of a decision, not automatically as a request for a new product. “Is this suitable for wet weather?” may indicate a missing attribute, while “Model A or Model B for wet weather?” indicates a comparison between known options.

Anagram is built around turning customer questions into a growth loop. Its Site Agent is designed to support shoppers while they compare options and decide what to buy, while its insights product helps brands learn from those interactions. See Anagram’s approach to customer questions.

Separate explicit comparisons from hidden comparisons

Explicit comparisons name two or more products, brands, or categories. Hidden comparisons ask about the tradeoff that determines which option wins.

Question pattern	What the shopper may be comparing	Merchandising implication
“A or B?”	Two specific products or variants	Add a direct comparison or improve cross-links between the products
“Which is better for hiking?”	Products by use case	Organize recommendations around the use case, not only the category
“What is the difference between regular and wide?”	Fit, size, or variant attributes	Make the differentiating attribute prominent in the selector and product copy
“Is the cheaper one good enough?”	Price against performance or durability	Explain the value tradeoff and show the lower-cost alternative beside the premium option
“What can replace this discontinued item?”	A substitute for an unavailable product	Create a substitution path based on shared requirements

Do not collapse every question into a generic “product interest” bucket. Preserve the comparison object, the shopper’s use case, and the attribute they are weighing. Those three fields make the resulting insight actionable.

Build a comparison taxonomy your team can use

A useful taxonomy records what is being compared and why. Start with a small set of categories, then add terms only when they lead to a different merchandising decision.

Capture these fields for each question:

1. Products or alternatives: the named SKUs, brands, categories, or substitute products.
2. Use case: the job the shopper wants to accomplish, such as travel, gifting, training, or everyday use.
3. Decision attribute: the tradeoff, such as price, size, fit, performance, ingredients, compatibility, care, or availability.
4. Shopper constraint: a requirement that can rule an option out, such as a budget, location, deadline, or physical limitation.
5. Stage of decision: exploration, shortlist, validation, or final selection.
6. Outcome: recommendation click, product click, add to cart, purchase, support escalation, or abandonment.

Keep the shopper’s original wording alongside the normalized category. “Will this fit under an airline seat?” and “carry-on compatible” may belong to one attribute, but the original phrase can reveal the content language shoppers understand.

Prioritize comparisons by commercial value

The most common comparison is not always the most valuable one. Prioritize questions that combine purchase intent with clear friction or meaningful revenue exposure.

A practical prioritization model uses four signals:

• Frequency: How often does the comparison appear?
• Friction: Does it lead to repeated questions, backtracking, support contact, or abandonment?
• Commercial importance: Does it involve a strategic category, high-margin product, or large traffic source?
• Fixability: Can the team address it through content, product data, navigation, recommendations, or an experiment?

For example, a frequent question about two shoe widths may deserve attention before a rarer brand-versus-brand question if the width information is absent from every product page. The first problem has a clearer site-level fix.

Measure comparison patterns against outcomes rather than publishing a volume leaderboard. A comparison that generates many conversations but few recommendation clicks may signal weak answers, poor product data, or a mismatch between the catalog and shopper needs.

Turn comparison patterns into merchandising changes

Each comparison pattern should produce a specific change to the buying experience. Do not stop at a report that says shoppers are confused.

Use the pattern to decide among these actions:

• Missing attribute: Add the attribute to structured product data, filters, cards, and product-page copy.
• Important tradeoff: Explain the tradeoff in a comparison module, buying guide, FAQ, or recommendation response.
• Repeated substitute request: Link unavailable or weaker-fit products to the closest alternative and explain the difference.
• Use-case uncertainty: Create a guided path that asks about the use case and recommends products with reasons.
• Variant confusion: Clarify the relationship between sizes, materials, bundles, generations, or feature tiers.
• Navigation failure: Add a collection, filter, or landing page using the way shoppers describe the need.
• Confidence gap: Surface reviews, proof points, care information, compatibility details, or delivery guidance that resolve the objection.

A comparison module should help a shopper choose, not merely place two products in adjacent columns. Show the attributes that matter for the stated use case and make the differences legible.

Use shopper questions to improve recommendations and content

Comparison questions reveal the inputs a recommendation experience needs. If shoppers repeatedly ask which product works for a particular activity, the recommendation flow should collect that activity and map it to reliable product attributes.

The same evidence can improve content outside the product page. Use recurring language to shape buying guides, category introductions, FAQs, email education, and product titles or bullets—provided the underlying product information supports the claim.

Anagram’s Site Agent can answer product questions and guide recommendations on a branded ecommerce site. That makes the interaction useful in two directions: the shopper gets help during the decision, and the team gets a clearer view of the questions that need better merchandising. The company also describes its AI Visibility product as showing how a brand appears in ChatGPT, including competitive comparisons, topic gaps, and citation sources; that is a separate signal from questions asked on the site, so keep the two datasets distinct. Read more about Anagram AI Visibility.

Validate the change with a testable hypothesis

A comparison insight becomes valuable when the team can test whether the merchandising change reduces uncertainty or improves progression toward purchase.

Write the hypothesis in a concrete form:

Shoppers comparing A and B are blocked by missing information about attribute X. Adding X to the comparison module and product cards will increase product selection from those sessions.

Choose a primary measure that matches the intervention, such as recommendation selection, product-detail engagement, add-to-cart rate, or purchase rate for comparison sessions. Pair it with a quality check: did the change reduce repeated questions, support escalations, or unanswered conversations?

Avoid crediting every improvement to the comparison change. Account for seasonality, promotions, inventory, traffic mix, and price changes. If a test is not practical, compare the same question cohort before and after the change while documenting what else changed.

Create a repeatable comparison-insight workflow

Make comparison discovery part of the merchandising cadence rather than a one-time research project. A lean ecommerce team can run this workflow each week or reporting cycle:

1. Collect new shopper questions and remove personally identifying information.
2. Tag explicit alternatives, use cases, tradeoffs, constraints, and outcomes.
3. Group near-duplicate questions without losing the original wording.
4. Rank the groups by frequency, friction, commercial importance, and fixability.
5. Assign one owner and one merchandising action to the highest-priority group.
6. Update the relevant product data, content, navigation, or recommendation logic.
7. Test the change and review both commercial and question-level outcomes.
8. Feed unresolved questions back into the next taxonomy review.

This approach gives an ecommerce team something aggregate funnel data cannot provide on its own: the shopper’s stated reason for comparing alternatives. The result is merchandising based on the decisions customers are actually trying to make, not assumptions inferred from where they happened to click.