Ecommerce customer issue analytics: delivery and returns problems

Short answer

To analyse delivery and returns problems in e-commerce, four groups of fields are enough: order status (status question, change, cancellation), delivery problem (delay, damage, wrong address, courier didn't get in touch), product mismatch (size, colour, not as described, wrong item) and returns (conditions, process, refund). Crossing these fields with courier service and product category shows in which process — and under whose responsibility — the problem sits.

The most valuable result is not the number of complaints but their owner: the warehouse, the courier partner, product content (photos and descriptions) or the returns policy. Each is fixed by a different team.

Why conversations matter in this sector

In an online store the customer cannot touch the product or see the seller, so questions and dissatisfaction almost always arrive in writing. In some stores, "where is my order?" can be a large share of all enquiries. These questions are both a support cost and a repeat-purchase risk: a customer with a bad delivery experience often does not buy again. Conversations show where a problem arises faster than the order system does.

Recommended fields

1. 1Request typeOrder status, order change, cancellation, delivery problem, product mismatch, return, payment, pre-purchase question, other.
2. 2Delivery problemDelay, damaged parcel, wrong address, courier didn't call, incomplete order, none.
3. 3Mismatch typeSize, colour, not as described, quality, wrong item, none.
4. 4Return stageQuestion about conditions, question about process, delayed refund, refused, none.
5. 5CategoryClothing, electronics, home, cosmetics and so on — from the order or the conversation.

The most useful cross-tabs

• Delivery problem × courier partner: which partner has more delays or damage.
• Mismatch type × category: size in clothing, description mismatch in electronics.
• Mismatch × specific product: which product page creates wrong expectations.
• Return stage × payment method: which method's refunds are slow.

The problem's owner

1. 1WarehouseWrong item, incomplete order, late dispatch.
2. 2Courier partnerDelay, damage, no contact — used as evidence in talks with the partner.
3. 3Product contentSize chart, photos, description — most mismatches come from here.
4. 4Returns policy and financeUnclear conditions, delayed refunds.
5. 5Support and botIf there is no automatic tracking answer to "where is my order?".

The "where is my order?" question

This question is often the largest enquiry group and should be tracked on its own. A high share means either the customer is not receiving tracking information or delivery takes longer than promised. The first is fixed with order confirmation and status notifications; the second with a more accurate promise or work with the partner. If the bot is connected to the order system the question can be answered automatically, but that does not solve the problem — it only hides it.

Mismatch: a product page problem

A large share of returns start with "not what I expected". Crossing mismatch type with product and category reveals specific pages: the size chart is wrong, the colour looks different in the photo, the material is not stated. These fixes are cheap and cut return costs directly. Customers' words in conversations — "it looked lighter in the photo", "the L is like an M" — are the clearest task for the content team.

The returns process

Return questions come in two kinds: conditions ("can I return it?") and process ("when will I get my money back?"). The first shows the rule is unclear, the second that the process is slow. If the share of delayed-refund questions rises, that is a signal to look at the finance process — customers are most dissatisfied at this stage and the risk of a negative review is high.

Illustrative example

This is an illustrative example. A clothing store sees its return rate rising. The mismatch × category cross-tab shows the rise is mainly in one brand's jeans, with the type "size": customers write that the size runs small. In parallel, the delivery problem × partner cross-tab shows damaged-parcel complaints with one courier service.

Decisions: the size chart for that brand is corrected and a "choose one size up" note is added to the page; a formal request with the evidence conversations is sent to the courier partner. A month later both indicators are compared.

Seasonal and campaign load

On big sale days and holidays order volume rises, and with it delivery problems. A rising share in these periods is expected; what matters is the comparison with the same campaign last year. Before a campaign, showing realistic delivery times and strengthening status notifications reduces the "where is my order?" wave.

Typical mistakes

• Lumping every complaint under "delivery problem".
• Guessing the courier partner from the conversation instead of taking it from the system.
• Automating the status question with a bot and never looking at its cause.
• Treating a mismatch as the customer's mistake — the problem is usually the product page.

Limits

• Conversations only see customers who complain or ask; silent returns come from system data.
• A complaint about a courier is the customer's perception; use it with system records in talks with the partner.
• Return conditions depend on consumer law — have changes to the policy checked legally.

E-commerce fields in Vexvon

In Vexvon the store sets up fields such as request type, delivery problem, mismatch and return stage itself; AI picks from the customer's words in each conversation and keeps the sentence as evidence. The panel's AI assistant gives cross-tabs, trends and period comparisons. A Shopify store can be connected to Vexvon, and conversations are stored with the customer record in the CRM. More: Shopify integration and analytics.

Next step

Read 50 return-related conversations from last month and write down the mismatch type and product for each. The three most repeated product pages are your first fix list. For the root cause method, see complaint root cause analysis; we can set up the fields together in a demo.

Further reading on this topic: customer pain point prioritization, customer journey pain points.