Customer journey: map discovery to returns
Map the customer journey from discovery to returns. Improve ecommerce customer experience with a reusable journey map and clearer channel handoffs.
- Published
- Last updated
A customer journey is the sequence of interactions a customer has with your store, from discovering a product to buying it and getting help after the sale. A customer journey map makes the gaps between those interactions visible, so you can assign a specific fix to the team that controls it.
Start with one purchase scenario. Map the information a customer needs at each stage, then compare that need with what your store actually provides.
Map the ecommerce customer journey
Record the customer’s goal, actions and expectations alongside the evidence for each touchpoint in your ecommerce customer journey map. Customer experience is how the customer experiences those interactions: whether they can find a suitable product, understand the total cost and get a promised remedy.
Customer journey mapping covers the relationship across channels and after purchase. A user journey usually follows one task through an interface, such as changing a delivery address. A successful address form doesn’t establish that the parcel reaches the right place.
Shopify’s guide to mapping the customer journey distinguishes the broader customer journey from a user’s task within a digital interface. Keep that distinction when a website team and a support team map the same purchase.
Reusable customer journey map template
Copy this table into a spreadsheet and add one row for each stage. Split the combined fields into separate columns if your workshop needs more room.
| Stage and customer goal | Customer action and touchpoint | Information needed and expectation | Friction and failure signal | Evidence and emotion | Accountable team and success measure |
|---|---|---|---|---|---|
| Initial need: define the job | Record the trigger and first contact | Record the intended use and constraints | Record what prevents a useful next step | Add an observation reference. Record an emotion only from customer research | Name one owner and the observable outcome |
| Product discovery: find candidates | Record the search or category visit | Record the attributes needed to find products | Record failed searches or missing categories | Cite the search record or customer statement | Name the catalogue owner and discovery measure |
| Comparison: choose a product | Record a product question or store visit | Record specifications, availability and policy needs | Record unanswered questions or inconsistent advice | Cite the enquiry. Keep inferred feelings labelled as assumptions | Name the merchandising owner and choice measure |
| Checkout: place the order | Record basket, payment and confirmation steps | Record total cost and delivery promise | Record payment errors or unexpected charges | Cite checkout records without retaining payment credentials | Name the ecommerce owner and completion measure |
| Delivery: receive the purchase | Record tracking and receipt | Record status, arrival expectations and next action | Record status gaps and delivery contacts | Cite order and carrier records | Name the fulfilment owner and delivery measure |
| Returns: resolve the problem | Record return request and resolution | Record eligibility, instructions and remedy | Record repeated requests or an unresolved return | Cite the service case and customer statement | Name the support owner and resolution measure |
| Repeat purchase: buy again | Record the next need or reorder | Record changed stock, price and preferences | Record a failed reorder | Cite the later order or enquiry | Name the ecommerce owner and repeat-order measure |
Fictional retail purchase example
Consider a fictional customer, Sam, buying a desk lamp from Harbour Home. Every observation below is invented to demonstrate the template. These are sample records, not retailer results.
| Stage and goal | Action and touchpoint | Need and expectation | Friction and signal | Illustrative evidence and emotion | Owner and success measure |
|---|---|---|---|---|---|
| Initial need: light a small desk | Sam notices glare while working | A lamp that fits the desk | Unclear dimensions prevent a shortlist | Scenario note F1. No emotional evidence recorded | Merchandising: dimensions present for eligible lamps |
| Product discovery: find lamps | Searches the website for a narrow base | Search results include base width | Search returns lamps without dimensions | Sample search F2. Emotion unrecorded | Catalogue team: searches reaching a suitable product |
| Comparison: choose between two lamps | Asks online, then visits the shop | Same dimensions and bulb advice in both channels | Staff cannot find the earlier question | Sample enquiry F3. Fictional statement: “I’m unsure which one fits” | Store team: question resolved without repeating measurements |
| Checkout: order the chosen lamp | Pays on the website after the visit | Delivery cost and dispatch timing before payment | Timing appears only after purchase | Sample order F4. Emotion unrecorded | Ecommerce: delivery answer visible before payment |
| Delivery: receive the lamp | Opens tracking, then contacts support | A status explanation and next step | Tracking shows no explanation for a delay | Sample case F5. Emotion unrecorded | Fulfilment: delivery contacts per dispatched order |
| Returns: resolve damage | Reports a damaged shade | Clear return instructions and a remedy | Photos are requested twice | Sample case F6. Fictional statement: “Repeating this is frustrating” | Support: returns resolved within the agreed service window |
| Repeat purchase: buy another lamp | Reopens the product page later | Current availability and product details | The earlier variant is unavailable | Sample visit F7. Emotion unrecorded | Ecommerce: repeat orders completed after starting a reorder |
Use a stage-based customer journey framework when you need a readable view of the purchase and its handoffs. Add a service blueprint when the problem depends on work behind the touchpoint, such as warehouse and support responsibilities for damaged goods.
A circular repeat-purchase model suits consumables and regular ordering because the next need follows the previous purchase. Customers can revisit comparison, abandon a basket or return an item. Draw those branches instead of forcing every interaction into a one-way funnel.
Carry context across an omnichannel journey
An omnichannel journey connects interactions through relevant context, even when the customer switches between the website and a physical shop. In Sam’s fictional journey, the online enquiry records the lamp models and desk measurements. With Sam’s agreement, shop staff retrieve that question when Sam presents its reference.
Later, support needs the purchased variant and the delivery issue. A product-question reference can help staff understand the request, but it mustn’t authorise access to a private order.
End-to-end customer journey diagram
This diagram explains the fictional sequence. The sample references F1 to F7 match the example map. They show how to organise evidence, not observed customer behaviour.
The sequence also works as a compact infographic in text: Initial need (F1), then Product discovery (F2), Comparison (F3), Checkout (F4), Delivery (F5), Returns (F6) and Repeat purchase (F7). Repeat purchase leads back to Product discovery. A customer can also return from Comparison to Product discovery without ordering.
Carry product identifiers, variant choices and the customer’s stated requirements between channels when the customer expects that handoff. Recheck stock and price at the next decision because an earlier answer can become stale.
Keep personal details and order history within authorised staff access. Verify identity again through the retailer’s approved process before disclosing private order information or changing an address. A shared device or an enquiry reference alone doesn’t prove who is asking. Share only the issue summary and records needed for the next team’s task.
Remove uncertainty during product selection
Remove product-selection uncertainty by fixing the missing fact at its source. Add base dimensions to the lamp catalogue, expose stock by variant and compare compatible bulb types consistently. A general reassurance about quality won’t answer whether a lamp fits a narrow desk.
For unavailable variants, show the stock state before the customer reaches payment. Explain a suitable alternative through the attribute the customer needs, such as a smaller base, rather than suggesting any product that happens to be available.
Shopping assistance can use those same catalogue facts and delivery policies in the conversation. Retailers use shoppa, an AI shopping agent, to help customers find, compare and buy products on their site. Its answers use the retailer’s catalogue and documents. Customers can compare lamp dimensions, ask about delivery and pay in the chat. Order support and returns continue there after the sale.
For example, a customer asks for a lamp with a base under 18 cm wide. The agent can narrow the catalogue, compare matching lamps and answer a delivery question from the store’s policy. The example describes the workflow, not a measured improvement in sales.
Join checkout with delivery and returns
Keep the promise made before payment available to the teams handling delivery and returns. Record what dispatch timing the customer saw, which shipping service they selected and what the confirmation says. A dispatch estimate and a delivery estimate describe different events.
When a parcel is delayed, show its latest known status and the next action your team can take. If the carrier has supplied no update, avoid inventing an arrival time. Explain the retailer’s actual escalation route and the remedies that apply to the customer’s issue.
A handover needs the problem, relevant order reference, checks already completed and the promised follow-up. Staff must be able to see that the customer already supplied a damage photo. Transfer that record through approved access instead of asking for it again.
Use checkout optimisation for the payment-stage checks. Decisions about queues and staff escalation belong in the guide to ecommerce customer service.
Prioritise changes by customer outcomes
Customer journey analytics analyses connected events and outcomes across customer stages. A journey map organises the customer’s needs and handoffs. Analytics tests where progress stalls and whether an assigned change improves the outcome.
Choose the measurement unit before calculating a rate. A browser session, a person and an order are different units. Anonymous sessions cannot establish a person’s full cross-device journey. Report unmatched records separately instead of joining them through guessed identities.
Customer journey metrics and KPIs
Use the following operational definitions as a starting point for your key performance indicators (KPIs). These are suggested reporting contracts, not universal platform defaults. Segment each measure by device and channel, then by relevant product category or delivery service where volumes permit.
| Metric and denominator | Data source and identity | Reporting window | Consent and access boundary | Interpretation and assigned action |
|---|---|---|---|---|
| Stage completion: eligible sessions reaching the next stage divided by sessions entering the stage | Storefront events joined by session identifier | Weekly cohorts, allowing 7 days to progress | Include only events permitted for analytics. Keep anonymous sessions separate | Check event coverage first. Ecommerce investigates the largest confirmed break |
| Abandonment: entrants without recorded next-stage completion divided by entrants | Same event stream and session identifier | Same 7-day progression window | Same analytics permission and access controls | A missing event isn’t proof of lost intent. Ecommerce inspects the affected device flow |
| Time to next step: median elapsed time among records completing both steps | Linked event timestamps and session identifier | Weekly cohorts with 7-day follow-up | Permitted analytics records only | Completers exclude unresolved journeys. Report their count separately. Ecommerce checks slow steps |
| Failed searches: zero-result searches divided by all submitted searches | Search service result counts joined by search request identifier | Weekly | Apply the store’s analytics permissions. Redact personal data typed into search | A submitted-search event alone doesn’t show failure. Catalogue team repairs supported query gaps |
| Payment failure: failed attempts divided by all payment attempts | Payment outcome records joined by attempt identifier | Weekly | Restricted transaction reporting. Exclude payment credentials | Separate technical errors from declines. Payments owner diagnoses the relevant category |
| Delivery contacts: distinct dispatched orders with a delivery enquiry divided by dispatched orders | Order and service records joined by order identifier | Dispatch cohorts followed for 14 days | Staff access for service reporting. Separate any optional marketing reuse | Repeat contacts need a separate count. Fulfilment checks the status or promise gap |
| Repeat contacts: cases with another contact about the same issue divided by cases opened | Service records joined by case identifier | Weekly case cohorts followed for 14 days | Restricted service reporting. Redact shared examples | Distinguish an agreed follow-up from a repeated unresolved question. Support repairs the failed handover |
| Unresolved returns: requests still open after 14 days divided by return requests | Return and service records joined by return identifier | Weekly request cohorts with 14-day follow-up | Restricted service records used for the approved reporting purpose | Exclude immature cohorts. Support checks blocked remedies and assigns a case owner |
Shopify’s standard customer events distinguish checkout starting from payment information submission and checkout completion. Payment information submission alone doesn’t prove a successful payment. Join payment outcomes to the transaction record when investigating failures.
For consent-aware collection, Shopify’s privacy permission states exposes analytics and marketing permission states separately. Your reporting must respect the permissions and purpose attached to each source. Service records don’t become marketing permission because they share an order identifier.
Worked delivery-information improvement
Suppose a fictional review finds 8 of 20 sampled lamp enquiries ask when an order will dispatch. In the sample, the product page omits dispatch timing. The customer journey map assigns that information gap to the ecommerce owner, with fulfilment responsible for the underlying promise.
The owner adds the approved dispatch answer beside the product’s delivery information and updates the shopping assistant’s policy source. The proposed change is small enough to test through one product category.
Before editing, record whether a customer can find the dispatch timing before payment without contacting support. After editing, repeat the same task on phone and desktop. Acceptance means the answer is visible, agrees with fulfilment’s policy and reaches the order confirmation without changing meaning.
Then compare the proportion of delivery-timing enquiries across comparable enquiry samples over two equal reporting periods. Keep product mix and channel visible. A lower rate would support fewer information gaps in that sample. It wouldn’t establish a revenue gain or prove the page change caused every difference.
Run a customer journey mapping workshop
Run the workshop around one bounded customer scenario and finish with a named improvement owner. Bring customer statements and operational records that participants are authorised to use. Remove personal details from shared workshop material.
- Choose the scenario and boundary. For the lamp example, start with product discovery and end when a delivery or return issue is resolved.
- Invite the teams that control those touchpoints, including merchandising, ecommerce and support. Include fulfilment when delivery promises are under review.
- Bring an evidence pack. Use redacted enquiries, search outcomes, checkout records and relevant service cases. Record the reporting window and selection method.
- Build the current-state map together. Write customer actions first, then attach the evidence. Mark an unsupported assumption explicitly in the working map.
- Resolve disputed stages through records. If staff disagree about where dispatch information appears, inspect the product page and confirmation from the same order scenario.
- Choose the friction with the clearest customer consequence. Assign the source correction and the handover to named owners, with an acceptance check.
- Draw a separate future state. Review the changed answer through the same customer task and preserve the original map beside the revised version.
If teams can’t reconcile two records, retain both observations with their source references. Assign someone to investigate the difference. Agreement around a tidy diagram doesn’t turn an assumption into customer research.
Map the B2B buying group and repeat order
A business-to-business (B2B) customer journey must show who can request, approve and pay for an order. One trade buyer’s successful search doesn’t establish that the buying organisation completes the purchase.
Consider a fictional café ordering replacement lamps. The buyer chooses products, the manager approves the quote and the accounts contact handles the invoice.
Fictional B2B customer journey map
| Stage | Trade buyer | Approver or accounts contact | Handoff and authoritative record | Success measure |
|---|---|---|---|---|
| Catalogue access | Signs into the authorised trade account | Account administrator confirms access | Customer account record sets the permitted catalogue | Buyer reaches the correct trade range |
| Negotiated terms | Reviews the account’s agreed price and delivery terms | Accounts confirms applicable terms | Approved account terms, not a remembered quote | Order uses the agreed terms |
| Quote or approval | Submits quantities and delivery needs | Manager approves the quote | Quote identifier preserves the exact version | Approval delay measured from submission to decision |
| Ordering | Converts the approved quote into an order | Accounts receives the payment obligation | Order links the approved quote and account | Approved orders complete without re-entry |
| Delivery | Confirms receipt at the café | Accounts reconciles the invoice | Delivery and invoice records link to the order | Quantity or invoice exceptions reach an owner |
| Repeat order | Reorders the same products or approved replacements | Manager reapproves changed cost where required | New order retains the earlier reference and current terms | Completed repeat orders divided by initiated reorders |
Analyse person-level actions using authorised user identifiers and account-level outcomes using the business account identifier. Preserve each person’s role and permission. Two staff members at one café are two people within one account, not two acquired customers.
Measure approval delay for submitted quotes and repeat-order completion for initiated reorders over a stated window, such as 30 days. Separate quotes still awaiting a decision. Restrict reporting to approved account operations, and keep promotional reuse subject to its own permission.
A buyer who selects a lamp quickly can still wait several days for approval. That finding assigns an approval handoff to the account team. It doesn’t justify changing search rankings.
Connect journey strategy with marketing and service
A customer-centric journey strategy starts with the outcome the customer needs and the handoff preventing it. For Harbour Home, choose “understand dispatch timing before payment”. The ecommerce owner controls placement, fulfilment supplies the promise and the acceptance task tests whether a customer can find it.
Match digital marketing to the customer’s stage without treating every service interaction as a promotional opportunity.
| Customer need | Message and trigger | Permission and suppression | Responsible team |
|---|---|---|---|
| Awareness | A guide to lamp dimensions when someone visits the guide | Reading a page doesn’t itself authorise a promotional message | Marketing maintains accurate educational copy |
| Comparison | Send a requested product comparison | Deliver the requested information. Suppress promotional follow-ups without the required permission | Merchandising supplies the facts |
| Post-purchase | Send the requested order or return update when status changes | Limit the message to the service purpose. Suppress duplicate or obsolete status messages | Support and fulfilment own the update |
| Repeat purchase | Send a promotional reminder to an eligible opted-in audience | Apply recorded permissions, withdrawal and frequency rules. Suppress reminders for customers with an open relevant complaint | Marketing owns the audience and suppression checks |
Keep the map focused on the purchase and its service handoffs. A full lifecycle automation programme needs a separate operating design.
Use AI without inventing customer insight
Use AI to summarise approved feedback or propose labels for an existing map, then trace each proposed finding to its source. AI can also help a customer compare catalogue products and complete a purchase. Those shopping actions and a generated summary are different kinds of evidence.
Suppose AI labels the fictional lamp case “customers distrust delivery promises”. Open the underlying enquiries. If they ask only when dispatch occurs, change the finding to “dispatch timing is missing from the product answer”. Keep a claim about distrust only when customer statements support that interpretation.
Remove inferred emotions and invented quotations from the map. Use redacted material in approved tools, with access and retention controls suitable for the data. Keep personal information out of unapproved tools. A generated label cannot supply an observation that your team never collected.
Give the journey a maintainable architecture
Customer journey architecture records how touchpoints connect through identity checks, authoritative records and team handoffs. It gives each map row a source that staff can maintain when the store changes.
Journey ownership register
| Journey element | Authoritative source | Connection and check | Accountable owner |
|---|---|---|---|
| Product advice | Catalogue and approved policy documents | Product and variant identifiers connect answers to current facts | Merchandising |
| Purchase and delivery promise | Order record and confirmation | Order identifier preserves the chosen variant and shipping service | Ecommerce and fulfilment, with one assigned handover owner |
| Private order access | Customer authentication and staff permissions | Verify identity before disclosing or changing private details | Customer access owner |
| Service and return progress | Service case and return record | Case and return identifiers retain attempted remedies and next action | Support |
| Measurement | Documented event and reporting definitions | Session, order or account identifiers match the stated unit | Analytics owner |
| Map version | Maintained map and change record | Evidence references, assumptions and acceptance checks survive each revision | Journey owner |
When the product range changes, check specifications and comparison answers. Checkout changes require a review of payment events and confirmation promises. After revising the returns process, check instructions, case handovers and unresolved-return reporting.
For each change, the journey owner versions the affected map rows and records who confirms the new answer. Retain unresolved assumptions until evidence resolves them. Start with the handoff that currently leaves a customer repeating a question, and keep its fix open until the same customer task succeeds.
Give your store its own shopping agent
Customers ask in their own words. Your agent finds, compares and checks out from your catalogue, right on your site.