Ecommerce Market Research: Methods & Examples

Ecommerce market research studies what online buyers need, how they understand an offer and what happens when they try to order. Use it to identify missing facts, unclear terms, navigation problems and rejected offers. Test the relevant page change, then measure purchases separately to learn whether the business result improves.

Example: checking a shelving product page

In this hypothetical EC-v1 example, assembled shelf measurements cannot answer whether the package fits a doorway. Another buyer expects upstairs service from entrance-only delivery. One issue needs verified package facts; the other needs clearer service terms.

EC-v1: the page cannot yet answer package fit

Flat-pack shelf · EC-v1

USD 89 before tax

Assembled size cannot determine whether the delivery package fits through a doorway.

Assembly is required. Keep assembled dimensions separate from the package dimensions you still need to verify.

The earlier page said “delivery included” and showed assembled size, image and price, but not package dimensions or delivery scope. EC-v1 now states entrance-only delivery. Package dimensions are still absent. Distinguish misunderstanding a stated term from needing a fact the page does not supply.

Exact page EC-v1: “Flat-pack shelf, assembled dimensions 80 × 30 × 120 cm. Proposed price USD 89 before tax, delivery included to the building entrance within the listed service area. Assembly required. Room-of-choice delivery is not included.” Package dimensions are absent, so the doorway task remains unanswerable from this page.

Answer EC01 says “They will carry it upstairs”; code delivery-scope misunderstanding because the stated offer ends at the entrance. EC02 says “I need the box dimensions before I buy”; code missing package information, not dissatisfaction with shelf size. Add verified package measurements and place the entrance-only condition near the offer before testing the revised page. If the unresolved decision is the price itself, use our pricing research guide with realistic alternatives.

Save the exact image, price, dimensions, delivery terms and page version. An answer about the earlier “delivery included” line cannot describe a later entrance-only page. Tie each response to what that buyer actually saw.

Ask what the buyer expects before explaining the page

Ask a supplier to show a useful output: the buyer’s unprompted explanation, recorded task path or assigned purchase outcome. Keep it connected to the exact page and opportunity to buy. Each answers a different question.

Show the actual product page at a defined viewport. Ask without directing attention to the missing information:

1. What would you expect to receive if you placed this order?
2. What is included in the displayed price?
3. Where would you expect delivery to end?
4. What would you need to check before buying for your home?
5. What is unclear or difficult to evaluate?
6. Which alternative, including not buying, would you consider?

Record the exact page version. If images, promotions or inventory change during collection, note the exposure difference.

Ask what arrives, what the price includes, where delivery ends and which fact is needed to decide. Someone may understand the assembled shelf perfectly but need box measurements. Do not code that as low appeal or failed understanding.

A listing-image test can help choose the version to investigate. PickFu’s Angry Orange case describes Thrasio testing product images and packaging with pet owners. For the shelf page, ask what each image tells buyers about contents, size and delivery. Carry the selected version into the fit task; feedback on creative does not establish that the task can be completed or that more orders will follow.

The task becomes testable when the necessary fact exists

Verify package measurements and delivery terms first. Label assembled and box dimensions separately. Put entrance-only delivery beside the offer. If delivery itself changes to room of choice, test it as a different complete offer.

Then observe a buyer finding the information and determining fit through a specified doorway. Do not tell them which measurement matters. Record the path and conclusion. Before the package facts exist, “cannot determine” is the correct answer to the original EC-v1 doorway task.

A fact present but not found calls for navigation or placement work. A stated condition misunderstood calls for communication work. A complete offer understood but rejected calls for offer or audience work. Each has a different owner and next comparison. Our message testing guide develops the communication question.

To connect those findings with earlier expectations, support and the eventual purchase, follow our customer journey research guide.

Price rejection and payment execution are different diagnoses

Once verified package information lets buyers check fit, follow them into checkout. This is a new task, separate from the unanswered EC-v1 doorway question. A checkout problem may concern price, its explanation or payment execution.

For the checkout example at one eligible address, assume the shelf is USD 89, tax is USD 7.12 and entrance-only delivery has no additional charge. The one-time total is USD 96.12. Assembly is required and room-of-choice delivery remains excluded. These are worked assumptions; a real study must use the actual address-specific bill and service terms. Place the total beside a clear payment action, such as “Pay USD 96.12,” rather than ask buyers to evaluate a price that changes after the task.

Before pointing to the summary, ask what the buyer expects to pay and receive. Then observe the path to the agreed task endpoint with neutral instructions, following GOV.UK's moderated usability guidance. A prototype can test finding and interpreting the bill. It cannot demonstrate that the live payment service completes a charge.

The diagnosis changes the next move. A buyer who correctly understands USD 96.12 and rejects it needs a different offer or a different reason to buy. Someone who expects USD 89 to be the final charge needs the total and tax treatment communicated earlier. A buyer who expects upstairs delivery still has the service-scope misunderstanding from EC-v1. None of those answers demonstrates a broken payment form. Our consumer behavior guide connects the account to the buying decision.

For a reported execution failure, connect the observed session with the order, payment attempt and provider response where the records allow. Do not infer the reason from a missing purchase event alone. For example, Stripe's payment-confirmation documentation distinguishes an additional authentication step, represented by requires_action, from a failed or successful payment. An unfinished authentication step is not evidence that the buyer rejected the shelf's price. A retry is also another attempt, rather than another buyer.

Fix the specific problem. A calculation error needs corrected billing logic. An unclear delivery term needs a clearer explanation. A payment failure needs a working technical path. Bigger text will not fix a complete, understood offer that buyers reject.

For a purchase experiment, define eligible checkout entrants and assign them before exposure. Keep assigned buyers in the comparison if they leave before payment or hit a failure. Record their actual bill and checkout version. Report attempts and technical success separately; link retries to the same order opportunity. Paid orders, refunds and complaints are different outcomes from understanding the button.

The revised page is a candidate for a commercial test

A correctly interpreted page and completed doorway task give the team a candidate for a purchase experiment. Specify the change, assignment, purchase opportunity, observation period and checks such as returns or delivery complaints. Record availability, promotions and device conditions that alter exposure. Our quantitative research guide develops the comparison and its base.

More orders with more delivery complaints may mean stronger persuasion without clearer expectations. Task and understanding results help explain the problem. A behavioral comparison measures the business effect. Do not label clearer wording as measured conversion lift.

More online orders can also reflect a move from another channel. Numerator’s Nestlé Coffee Partners case abstract pairs observed online and in-store shopper behavior with targeted surveys about the reasons for change. If you sell through several channels, distinguish a first purchase, a repeat purchase and an existing purchase moving online where your records allow. Purchase records establish the movement; answers supply reported explanations. Assess the page change through your own behavioral comparison.

The finished next move is verified package information and conspicuous entrance-only delivery, followed by the task check and a defined purchase comparison. Marketing cannot use a survey to establish what service the courier actually provides.

Sapien can compare simulated reactions to the page’s offer and message, then explore the reasons and audience differences behind them. For the shelf, that means investigating how buyers interpret verified package information, entrance-only delivery and the complete bill. Use the report to develop a page candidate; observed navigation, package fit and purchase outcomes remain separate evidence.

Once the box dimensions and address-specific bill are verified, plan the page comparison with Sapien. The original EC-v1 exposure and the candidate page should stay attached to their answers, so your team can identify whether the next change belongs in product information, the delivery offer or its explanation.