7 Poor Customer Service Examples and How to Fix Them

Ninety-one percent of customers who are unhappy with a brand leave without complaining.

They do not send a strongly worded email. They do not leave a one-star review. They simply stop buying, and you never find out why. By the time declining repeat purchase rates or rising churn show up in your data, those customers have already made their decision.

This is the fundamental problem with poor customer service in ecommerce. The damage is largely invisible until it accumulates into a revenue problem. A single bad experience is enough — fifty-nine percent of customers abandon a brand permanently after just one poor interaction, according to PwC research.

The seven examples in this guide are the most common poor customer service failures in ecommerce, each described through what the customer actually experiences rather than what the business assumes they experience. Every example includes the specific fix, not general advice.

All 7 poor customer service examples at a glance

1. Slow or no response - the failure that drives more churn than any other

The most common form of poor customer service in ecommerce is also the most preventable.

A customer sends a WhatsApp message asking about their order at 8pm. The reply arrives the next morning. The customer has already contacted another store about a similar product. They will not come back.

Eighty-two percent of consumers expect an immediate response when contacting a brand through live chat. The average ecommerce store delivers a response in four to six hours. That gap is where customers make their decision to return or to leave, and most stores have no idea it is happening because the customer who did not receive a fast response does not complain. They just stop.

The financial cost is specific. Research shows that responding to a customer contact within one hour retains seventy-one percent of complaining customers. Responding after one hour retains forty-eight percent. A two-hour response gap costs you nearly one in four customers who were recoverable.

The scenario plays out the same way across every channel. A question asked on Instagram DM at lunchtime, left unread until the next morning. A website chat message that sits in the queue while an agent is handling a phone call. A WhatsApp message that the team does not see because they check the account twice daily.

The specific fix:

AI chatbot automation resolves the response speed problem completely. AeroChat responds to incoming messages on WhatsApp, Instagram, and website chat within seconds, using live Shopify order data to give accurate, specific answers rather than generic holding messages.

A customer who asks about their order at 8pm gets the live tracking status at 8pm. They do not wait until the next morning. The need to contact you again disappears.

For stores not yet using automation, the minimum fix is a genuine acknowledgement message on every channel within five minutes, with a specific time commitment for the full response — not "as soon as possible," which is vague enough to feel like dismissal.

The customer communication strategies guide covers how to configure response time standards across each channel and what automation setup looks like in practice.

2. Ignoring social media messages - the silent brand reputation risk

Brands respond to roughly half of the customer messages they receive on social media.

The other half go unanswered. A customer who messaged on Instagram about a product question, received no response, and found the same product available elsewhere did not complain publicly. They simply bought from someone who responded.

The customers who do not receive a response do not stay silent permanently. Forty-five percent of people who have a bad experience are more likely to post a negative review. Thirteen percent of dissatisfied customers tell fifteen or more people about their experience. A single unanswered Instagram DM about a genuinely legitimate query can become a negative story shared across that customer's network.

For ecommerce brands where Instagram and WhatsApp drive meaningful purchase volume — which describes most DTC brands in 2026 — ignoring messages on those channels is equivalent to leaving your shop door locked during opening hours. The customers arrive, find no one to help them, and leave.

The scenario is specific: a customer comments on a product post asking about size availability. No response. They purchase from a competitor who responded in the comments within the hour.

A second customer sends an Instagram DM asking about return policy before committing to a purchase. No response for 36 hours. They make no purchase.

The specific fix:

Connect every customer contact channel — Instagram DMs, WhatsApp, website chat, email — to a single inbox where every message is visible and where AI handles the initial response automatically.

AeroChat's comment-to-DM automation detects comments on Instagram posts with product questions and sends an automated DM that opens a direct conversation, answers the question using your live catalogue, and provides a purchase link — without any manual agent involvement.

For WhatsApp, the same AI handles every incoming message within seconds. Every channel is covered, every message gets a response, regardless of when it arrives.

For the specific setup of Instagram DM automation for ecommerce, that guide covers the comment trigger configuration and DM flow structure in detail.

3. Making customers repeat themselves across channels - the failure that signals the brand does not know them

A customer messages on WhatsApp about a damaged product. No resolution. They email the support address explaining the same issue from scratch. They are told to contact via website chat for faster service. On website chat, they explain the issue for the third time.

At no point in this journey does any agent acknowledge that the customer has already been in contact. At no point does the context of the original WhatsApp conversation appear in the email reply or the chat conversation. From the brand's perspective, these are three separate tickets. From the customer's perspective, this is one unresolved problem they have had to explain three times to people who did not communicate with each other.

Eighty-five percent of customers expect consistent interactions across all channels with a business. When that consistency breaks, it is not just frustrating. It signals to the customer that they are an order number, not a person. The emotional effect of that signal is disproportionate to the operational cause.

It takes twelve positive experiences to make up for one unresolved negative experience. A customer who explained their problem three times and remained unresolved will require an extraordinary sustained recovery effort from your brand — if they give you the opportunity to make that attempt at all.

The specific fix:

An omnichannel customer service platform with shared customer history across all channels resolves this failure at the infrastructure level.

When a customer messages on WhatsApp and then follows up on the website chat, the agent on chat sees the full WhatsApp conversation linked to the same customer profile, alongside the customer's order history and purchase dates. They do not ask the customer to repeat anything. They continue from where the WhatsApp conversation ended.

AeroChat provides this unified inbox across WhatsApp, Instagram, website chat, and email with full context continuity. The customer's conversation history follows them across every channel they use.

For a full breakdown of how multichannel customer service software provides this context continuity, that guide covers the platforms that do it well and those that only appear to.

4. Scripted, robotic responses that miss the actual question

A customer messages: "My order says delivered but I have not received anything. What do I do?"

The agent responds: "Thank you for contacting us. Your order has been dispatched and you should receive it within 3 to 5 working days. Please check your confirmation email for tracking information."

The customer already has the tracking. The tracking says delivered. The agent did not read the message carefully enough to notice that distinction. The customer is now more frustrated than before they contacted support, because they have received a response that proves no one actually engaged with their specific problem.

This type of response failure is most common in high-volume support operations where agents are managing too many conversations simultaneously, where response templates are used without customisation, or where chatbots are configured with static FAQ answers rather than live order data access.

Scripted responses that miss the point are more damaging than slow responses. A slow but accurate response shows the customer you eventually engaged with their problem. A fast but wrong response shows you did not engage with it at all. The customer's perception of being ignored is stronger from the latter.

The specific fix:

The fix for agent-side scripted responses is training and reduced simultaneous conversation load. An agent handling eight conversations at once cannot read each one carefully. Reducing that load through AI automation of routine queries gives agents the attention capacity to engage properly with the complex ones.

The fix for chatbot-side scripted responses is data integration. A chatbot that answers from a static knowledge base gives the same generic answer to every variation of a question. A chatbot connected to your live Shopify order data gives a specific answer based on the actual order status for this specific customer.

When a customer says their order shows delivered but has not arrived, a chatbot with live carrier data access checks the carrier's actual delivery confirmation, asks the customer to confirm their delivery address, and either identifies a potential safe location or flags for agent investigation — rather than sending a response about 3 to 5 working days.

The AI chatbot for customer service guide covers how data integration changes the quality of AI responses from generic to genuinely accurate.

5. Chatbots that trap customers in loops - the failure that makes AI feel worse than nothing

A customer contacts a store's chatbot asking to return an item. The bot responds with three menu options: Track Order, FAQ, Contact Us. None of them is Return an Item.

The customer selects Contact Us. The bot presents business hours and asks the customer to call during those hours. The customer is on their phone at 9pm. They cannot call.

They return the next day during business hours and click Contact Us again. The bot presents the same three menu options.

This experience — which happens daily across thousands of ecommerce stores — does not just fail to solve the problem. It actively makes the customer feel more frustrated and more disrespected than if no chatbot existed at all. The chatbot created a barrier between the customer and a resolution rather than removing one.

Nearly one in five consumers say they receive zero benefit from AI-powered customer support, according to Qualtrics research. The failures driving that perception are almost always chatbots that are poorly configured — missing data integrations, no escalation paths, and menu-driven flows that cannot handle natural language queries.

The specific fix:

A well-configured chatbot has three non-negotiable elements: it understands natural language queries rather than requiring menu selection, it is connected to live order data so it can give accurate specific answers, and it has a clear escalation path to a human agent when it reaches the limits of what it can resolve.

The escalation path is the most commonly missing element. A chatbot that cannot resolve a query and does not escalate to a human is a dead end. A chatbot that cannot resolve a query but immediately connects the customer to an agent with the full conversation context already transferred — so the customer does not start again — is a positive service experience even though the chatbot itself did not resolve the issue.

AeroChat handles this by detecting when a query exceeds its resolution capability and escalating to the human inbox with full conversation context and relevant order data already loaded. The agent receives a complete picture rather than a blank conversation.

The chatbot vs live chat guide covers the specific query types that benefit from each approach and how to configure the handoff correctly.

6. Flat refusals without a path forward - the failure that converts neutral customers into active detractors

A customer contacts support requesting a return on an item purchased 45 days ago. The store policy is 30 days.

The agent responds: "Our return policy covers purchases within 30 days. Unfortunately, this order falls outside that window and we are unable to accept a return."

Full stop. No empathy. No alternative. No acknowledgement that the customer's circumstances may have been difficult. No next step.

The customer's experience of this interaction is not that the policy was enforced fairly. Their experience is that the brand chose the policy over the relationship. That distinction matters enormously for their future buying behaviour.

A flat refusal without empathy or a path forward is one of the most reliable ways to convert a neutral customer into an active one who shares their negative experience. Thirteen percent of dissatisfied customers tell fifteen or more people. A review that says "they refused my return with no flexibility and no interest in my situation" is significantly more damaging than a review that says "the return process took a while."

The specific fix:

The fix has two components. First, agents should never make unilateral final decisions on exception requests. The appropriate response to a return request outside policy is an acknowledgement, an escalation to someone with decision-making authority, and a specific timeline for that person to respond — not a flat refusal from a front-line agent.

Second, every policy enforcement response should include something. An alternative resolution. An acknowledgement that the customer's situation is understood even if the outcome cannot change. A goodwill gesture where appropriate. The goal is for the customer to end the interaction feeling that the brand engaged with their situation honestly, even if the policy boundary held.

For the specific response language for policy-adjacent complaints, the Shopify complaint handling guide covers the exact phrasing for the escalation approach that maintains goodwill without making unauthorised promises.

7. Resolving the ticket, not the customer - the failure that hides in your CSAT scores

A customer contacts support about a wrong item received. The agent apologises, arranges a replacement, and marks the ticket as resolved.

The replacement is dispatched but arrives with the same wrong item due to a warehouse error.

The customer contacts again. By this time, they have spent four days without the correct product, contacted twice, and explained the situation twice. Their patience is gone. The original ticket was resolved correctly. The customer was not.

This failure is distinct from the others because it often does not appear in standard customer service metrics. The first ticket was resolved. CSAT may have been positive immediately after the first interaction. But the underlying problem was not actually fixed, and the customer's experience of the brand over the week has been materially negative despite the operational resolution.

The failure to follow up after a complaint resolution is one of the most underused opportunities in ecommerce customer service. Almost no stores do it. Which is exactly why doing it creates a disproportionately strong impression when a customer receives it.

A message sent twenty-four hours after a complaint resolution that simply asks whether everything arrived correctly and whether the customer is satisfied costs almost nothing to send and signals that the brand genuinely cares about the outcome rather than the ticket closure.

The specific fix:

Configure an automated follow-up message triggered by complaint resolution events. The message should be brief and genuine.

"Hi, just checking in — your replacement should have arrived by now. Everything come through as expected? If anything is not right, please reply here and we will sort it immediately."

This message costs nothing to send at scale once configured. It catches the cases where the operational resolution did not translate to a customer satisfaction outcome. And for the customers whose replacements did arrive correctly, it creates a strong final impression that significantly increases the likelihood of repeat purchase.

For the specific follow-up sequence and timing for different complaint types, the post-purchase ecommerce strategy guide covers the full automation setup including complaint follow-ups.

How to identify which failures are happening in your store

Most ecommerce stores are running at least two or three of these failures simultaneously without knowing it. Four data sources reveal where your gaps are.

Your reviews. Read your two-star and three-star reviews specifically. Not the one-stars, which are often venting without specifics, and not the five-stars. The mid-range reviews are the most informationally valuable because they describe a real experience that was mostly acceptable but specifically fell short. Each specific complaint in a two-star review maps to one of the failures in this article.

Your support contact volume per order. If customers contact you at a high rate relative to order volume, you have either a fulfilment problem generating contacts or a self-service failure driving avoidable contacts. Both are addressable. Neither is visible without tracking the metric deliberately.

Your repeat purchase rate by support contact history. Customers who contacted support and did not repurchase reveal the specific support failures that damage loyalty. Customers who contacted support and did repurchase show you what your service does well. The gap between those two groups tells you the commercial impact of your support quality.

Your channel response times. Pull the average first response time across every channel where customers contact you. Email, WhatsApp, Instagram, website chat. If any channel exceeds ten minutes during business hours, you have a speed failure on that channel that is costing you customers daily.

For more on using support data to identify specific store weaknesses, the customer needs identification guide covers the full analysis approach with specific metrics for each failure type.

Frequently asked questions

What is the real cost of poor customer service?

Fifty-nine percent of customers abandon a brand permanently after a single poor experience. Ninety-one percent of unhappy customers leave without telling the brand why, making poor service largely invisible until it shows up as a revenue problem. It takes twelve positive experiences to recover from one unresolved negative one. And thirteen percent of dissatisfied customers tell fifteen or more people about their bad experience, actively damaging future customer acquisition. The total commercial cost of poor service is almost always higher than what appears in support metrics.

How do you fix slow response times without hiring more staff?

AI chatbot automation on WhatsApp, Instagram, and website chat resolves the response speed problem without additional headcount. A chatbot connected to your Shopify store data responds in seconds to incoming messages on every channel simultaneously, handling order status queries, product questions, return requests, and FAQ responses automatically. Human agents handle the complex and high-value conversations that genuinely require human judgment. The result is faster responses at lower cost.

Why are scripted chatbot responses so damaging for customer service?

Scripted chatbot responses that do not address the specific question asked signal to the customer that the brand engaged with their contact but not their problem. This is perceived as more dismissive than a slow response. A slow but accurate response eventually demonstrates engagement. A fast but irrelevant response demonstrates the opposite. Chatbots generate scripted failures when they are configured with static FAQ content rather than live data access, or when they use menu-driven flows rather than natural language understanding.

How do you identify poor customer service in your own ecommerce store?

Four data sources reveal specific failures: two and three star reviews that describe specific experience failures, support contact volume per order that reveals avoidable contacts, repeat purchase rate by customers who contacted support versus those who did not, and average first response time across every contact channel. Each of these metrics maps to a specific failure type and points toward a specific operational fix rather than a general "improve service quality" conclusion.