Ecommerce live chat QA checklist before a major promotion

Test the promise, not only the widget

A launcher opening on desktop proves very little. Before a major promotion, test the exact journeys a shopper will take from advertisement to landing page, product choice, discount, checkout, delivery question, order confirmation, and later support.

The purpose of quality assurance is to find disagreement before customers do. Promotion terms, stock messages, saved replies, return rules, delivery estimates, and agent guidance should point to the same answer.

Name one owner for the checklist and one decision maker for unresolved policy questions. A shared document with no owner becomes a record of risks that everyone saw and nobody closed.

Choose complete test journeys

• New mobile visitor arriving from a promotion link.
• Returning customer with an existing chat history.
• Shopper applying the main discount to eligible and ineligible items.
• Customer choosing delivery near a stated cutoff.
• Visitor asking a question while the team is online and while it is away.
• Shopper whose payment, address, stock, or promotion step fails.

Give each journey a starting URL, device, location assumption, basket, expected promotion result, expected chat state, and expected human answer. Test what the customer sees rather than relying on admin configuration screenshots.

Include a low bandwidth or delayed response scenario. Peak traffic can expose timing issues that do not appear during a fast office test, including duplicate sends, stale stock messages, and a widget that covers the checkout action.

Verify promotion terms word by word

Collect the approved offer, dates, timezone, product exclusions, minimum spend, combination rules, stock limits, gift conditions, delivery terms, and return effects in one source. Link agents to that source instead of distributing copied versions.

• Check the landing page, banner, product page, cart, checkout, email, and chat reply against the approved terms.
• Test the minute before and after the start and end time.
• Test currencies, markets, customer groups, and devices that change eligibility.
• Confirm whether discounts combine and what message appears when they do not.
• Remove old saved replies and campaign shortcuts from the inbox.

If the promotion cannot be explained in a short accurate reply, simplify the customer wording or give agents a clear decision tree. Complexity that customers cannot understand becomes support volume and disputes.

Test the questions that block a purchase

Review likely questions about size, fit, compatibility, ingredients, materials, stock, subscriptions, warranty, installation, and delivery. Confirm which source is authoritative and who can resolve an exception.

Open chat from several product pages and verify that useful page context reaches the inbox. The teammate should know which item the visitor is viewing without asking for information already available.

• Check that links in saved replies open the correct current page.
• Verify that product names and variants match the storefront.
• Do not let agents promise stock that is only visible in an old document.
• State when an answer depends on the customer's measurements or use case.
• Escalate safety or suitability questions that require qualified judgment.

Test discount, payment, and delivery failures

Run successful and failed checkout attempts with the approved test method. Capture the exact customer message and the context visible to the support team. A generic Something went wrong response gives neither side a safe next step.

• Eligible discount applies correctly.
• Ineligible item explains the relevant rule.
• Expired or mistyped code does not imply a site outage.
• Payment failure does not invite the customer to paste card details into chat.
• Delivery methods match the address and promotion promise.
• A duplicate attempt can be distinguished from a completed order.

Define the route for payment concerns before launch. Agents need to know what they can see, what they must never request, when to escalate, and what evidence confirms an order without exposing sensitive information.

Test the configured widget on the real storefront

Test the production theme, brand colors, position, greeting, online state, away state, proactive prompts, consent layer, and mobile keyboard. A staging widget can miss conflicts introduced by advertising, analytics, banners, checkout extensions, and theme code.

Use keyboard navigation, zoom, screen reader checks, and touch testing informed by WCAG 2.2. Make sure the launcher and composer do not cover price, variant, cart, consent, or checkout controls.

• Send text, emoji, links, and an allowed attachment.
• Reconnect after changing page or losing the network.
• Check notification behavior with sound off.
• Confirm the visitor can close and reopen without losing the draft unexpectedly.
• Test every supported browser and a representative small phone.

Prove routing and handoff

Send one test for every promotion topic and verify the correct team, owner, notification, and response guidance. Include a test when the intended owner is absent so the fallback is real.

• Product and fit question.
• Discount or promotion interpretation.
• Payment or account concern.
• Delivery promise question.
• Existing order exception.
• Fraud, safety, privacy, or abusive content escalation.

Complete a handoff while the visitor is active. The second teammate should receive the summary and context without asking the customer to repeat the issue. Confirm who remains accountable after the transfer.

Test the away path separately. Make sure contact details are captured once, notifications reach the real owner, and the stated follow up window matches the staffed rota after the promotion closes for the day.

Check measurement without leaking conversation data

Verify that chat opens, meaningful conversation starts, generated leads, and purchases are not collapsed into one conversion. Test consent choices and confirm that personal information or message text is not sent to analytics.

Add campaign and page context through controlled values rather than full free text. Compare a test conversation in the inbox with the corresponding analytics events, using a safe test marker that can be excluded from reporting.

Document expected event counts for each journey. Duplicate events after route changes, refresh, or widget reopen can make a campaign appear more successful while the actual customer outcome is unchanged.

Run one incident rehearsal

Choose a plausible failure such as a broken discount, payment outage, wrong delivery message, or unavailable best seller. Simulate three customers reporting it within a few minutes.

• Who confirms the pattern?
• Who can pause advertising or proactive prompts?
• Who updates the storefront message?
• Who gives agents the approved explanation?
• How are affected customers recorded and contacted?
• What evidence marks the incident resolved?

The rehearsal should reveal missing permissions, contacts, and decision rights. Fix those before launch. A beautifully written incident plan is not useful if the person on shift cannot change the banner or reach the payment owner.

Shopify's current Black Friday and Cyber Monday strategy guide treats preparation as a whole journey. The support rehearsal is how a small team proves that journey remains coherent when one part fails.

Use a clear release decision

Record each test as passed, failed, accepted with a named limitation, or blocked. Attach the evidence needed to reproduce the result and give every failure an owner and due time.

Do not launch with an unexplained blocked checkout, false delivery promise, unsafe payment instruction, or unstaffed live status. Lower risk wording or presentation issues can be accepted only by the person responsible for the customer outcome.

Repeat the critical path after the final promotion, theme, and routing changes. Then keep the checklist for the campaign review so defects found by customers become regression tests next time.

FAQ

FAQ

Recommended next steps