If you run an online shop, you know three things keep you up at night: answering repetitive enquiries around the clock, watching 70% of carts get abandoned without a purchase, and handling returns or after-sales issues that clog up your inbox. Artificial intelligence won’t solve all three problems at once, but it can automate the mechanical tasks and let your team focus on what generates value.
In 2026, AI applied to ecommerce is neither science fiction nor the preserve of Amazon. It is accessible technology that integrates with Shopify, WooCommerce or PrestaShop, and you pay for it by usage. This article breaks down what to automate in support, cart recovery and after-sales, what results to expect and which mistakes to avoid.
This page is for information only and is not binding advice. Each case is tailored after a diagnóstico.
TL;DR
- AI covers stock, delivery, sizing and tracking enquiries 24/7, freeing up to 60% of support volume.
- Recovering abandoned carts with automated WhatsApp or email can rescue between 10% and 25% of lost sales.
- Automated after-sales handles returns, exchanges and reshipments without human intervention in standard cases.
- ROI is measured in tickets closed, cart conversion and team hours saved, not in impressions.
- The key is defining what gets handed over to a human: complex cases, disputes and consultative sales are not automated.
- A basic project starts in 2-4 weeks; a full one with after-sales can take 4-8 weeks.
24/7 customer support without growing your team
60% of ecommerce enquiries are repetitive: stock availability, delivery times, order status, size guides, payment methods. A chatbot with access to your inventory and order database can answer them in under 10 seconds, in any language and at any time.
The difference from a 2020 chatbot is that today’s language models understand context, rephrase ambiguous questions and stay coherent in long conversations. If someone asks “do you have the 42 in black” after viewing a trainer, the system knows which product they mean.
What to automate
- Catalogue enquiries: availability, variants, features.
- Order status: tracking number, estimated delivery date, carrier issues.
- Policies: return deadlines, shipping costs, accepted payment methods.
- Size guides, compatibility, instructions for use.
What not to automate
- Consultative sales: if the average order value exceeds 200 euros or the product requires technical advice, hand over to a human.
- Complex complaints: disputes over quality, damaged products or lost deliveries.
- Customisations: engraving, made-to-measure adjustments, orders outside the catalogue.
The handover to a human must be seamless: the agent receives the full conversation history, and the customer doesn’t have to repeat information. This is set up by integrating the chatbot with your CRM or ticketing system.
You can deploy automated support on your website, on the WhatsApp Business API or on both channels. WhatsApp has open rates above 90%, compared with 20-30% for email, and lets you send proactive notifications (order confirmation, shipping alerts) without the user having to visit your website.
Abandoned cart recovery that converts
69% of ecommerce carts are abandoned before the purchase is completed. The reasons: unexpected shipping costs, a long checkout process, a lack of payment options, or simple distraction. An AI system can contact the user at the optimal moment, remove friction and close the sale.
How a recovery flow works
- The user adds products to the cart but doesn’t complete payment.
- After 60-90 minutes, the system sends a WhatsApp message or email reminding them about the cart, with a direct link to checkout.
- If there is no response within 24 hours, it sends a second message with an incentive: free shipping, a 5% discount, or an extended return period.
- If the user replies with an objection (“shipping is expensive”, “I don’t have PayPal”), the chatbot can offer alternatives or hand over to sales.
Industry data indicates that the first window (1-4 hours after abandonment) converts 3 to 5 times better than reminders sent the following day. The key is to act quickly and personalise: “Hi Ana, you still have the payment pending for the black Nike Air Max trainers in size 42. Do you need any help?” works better than “You’ve left items in your cart”.
Incentives that work
- Free shipping if the cart exceeds a certain amount.
- A 5-10% discount valid for 48 hours.
- A limited stock reminder (if it’s true).
- Making payment easier: a direct link to Bizum, a saved card or cash on delivery.
What doesn’t work: bombarding people with daily emails, offering 30% discounts (you train the customer to abandon carts to get a reduction), or sending generic messages with no context.
You can integrate cart recovery with your ecommerce platform in under a week if you use Shopify or WooCommerce. ROI is measured in net conversion: recovered sales minus the cost of incentives and messaging.
Automated after-sales: returns, exchanges and loyalty
The sale doesn’t end when the customer pays. Then come shipment tracking, delivery confirmation, the request for a review and, in 10-15% of cases, handling returns or exchanges. All of this can be automated without losing quality.
Typical after-sales flows
Confirmation and tracking: when an order is placed, the system sends a WhatsApp confirmation with a summary, the amount and the estimated time. When the carrier updates the status, the customer receives an automatic notification. When it is delivered, a thank-you message and a review request.
Returns and exchanges: the customer starts the process from WhatsApp or the website. The chatbot validates the order number, checks that it is within the legal deadline (14 days in Spain for online sales), offers options (return with refund, size exchange, voucher for the next purchase), generates a return label and updates the status in your system.
Issues: damaged product, wrong size, a missing item. The system collects evidence (a photo of the product, a description of the problem), creates a ticket in your issue tracker and hands over to a human if the case requires manual assessment.
What the system must not do
It must not decide in disputes: if the customer says the product arrived damaged but the shop’s policy is “no signs of use”, a human must review the photos and use judgement. Nor should it offer unauthorised compensation: discounts, express reshipments or exchanges outside policy.
The advantage of automating after-sales is twofold: you reduce the operational load (a size exchange can be resolved in 2 minutes without intervention) and you speed up the response (the customer doesn’t wait 24 hours for you to read their email).
Integration with your ecommerce platform
For an AI system to operate on your shop, it needs access to three data sources: product catalogue, orders and customers. This is done through APIs or webhooks, without migrating platform.
Compatible platforms
- Shopify: REST and GraphQL API, real-time webhooks, apps in the marketplace.
- WooCommerce: REST API, native plugins, integration with Zapier or n8n.
- PrestaShop: webservice API, third-party modules.
- Magento: REST API, real-time events.
A typical integration takes between 3 and 10 working days, depending on the complexity of your catalogue and the number of flows to automate. If you use custom payment gateways or external ERPs, the timeline goes up.
Data the system needs
- Up-to-date inventory: stock by variant, price, availability.
- Orders: number, status, items, delivery address, tracking.
- Customers: name, phone number or contact email, purchase history.
None of this data leaves your infrastructure except for processing (sending a message, validating a cart). If you work with a reputable provider, the architecture must comply with the RGPD (Spain’s name for the GDPR): data encrypted in transit and at rest, role-based access limits, log auditing.
If you want to explore integrations beyond ecommerce, take a look at the options for AI automation applied to other sectors.
How much it costs and how to measure return
There is no standard price because every shop has a different volume, catalogue and set of flows. A shop with 500 orders a month and 2,000 SKUs doesn’t have the same needs as one with 50 orders and 50 products.
What you can estimate is the cost per WhatsApp conversation (between 0.01 and 0.05 euros depending on country and message type), the cost of the AI platform (many providers charge by volume of messages processed) and the initial development time.
How to measure ROI
- Tickets closed without human intervention: if the chatbot resolves 50% of enquiries, multiply that volume by your team’s hourly cost.
- Conversion of recovered carts: net sales generated by the recovery flow minus the cost of incentives and messaging.
- Reduction in after-sales handling time: how many hours a month you save by automating standard exchanges and returns.
- Increase in satisfaction: NPS, Trustpilot reviews, repeat purchase rate.
Typical return on ecommerce projects shows between the second and fourth month. If the implementation is poor (badly configured handovers, robotic messages, no integration with inventory), the system creates frustration and ROI is negative.
Common mistakes when automating ecommerce with AI
Automating without cleaning your data: if your catalogue has incomplete descriptions, badly labelled variants or out-of-sync stock, the chatbot will give wrong information. Audit your data before automating.
Not defining handover criteria: if you don’t specify which cases escalate to a human, the system will try to resolve everything, including complex disputes, and the customer will end up frustrated.
Generic messages: “Hi, you have a cart pending” converts less than “Hi Marta, the size 39 Dr. Martens boots you looked at yesterday are still available. Do you want them?”. Personalisation matters.
Ignoring the customer’s preferred channel: if your audience is over 45, they may prefer email to WhatsApp. If they’re under 30, the other way round. Ask or test before deciding.
Not measuring: if you don’t track cart conversion, response time or handover rate, you don’t know whether the system works or is a dead cost.
If you want to avoid these mistakes from the design stage, a diagnóstico previo lets you identify what to automate first and with what return expectations.
When it makes sense to invest in AI for your ecommerce
Not every shop needs automation. If your turnover is under 5,000 euros a month and you handle enquiries in 10 minutes a day, the return probably doesn’t justify the investment. But if you are in any of these scenarios, it does:
- You receive more than 50 enquiries a week and can’t answer them during working hours.
- Your cart abandonment rate exceeds 65% and you have no recovery system in place.
- You handle returns manually and it takes you more than 2 hours a week.
- You want to scale sales without hiring more support staff.
- Your customers ask for support on WhatsApp and you can’t respond in under 4 hours.
AI doesn’t replace your team, it amplifies it. A salesperson who used to close 10 sales a month because they spent time answering basic enquiries can close 15 or 20 if a chatbot filters out the mechanical work.
If you sell in sectors such as fashion, electronics, sport or food, where customers research before buying and value a quick response, automation is especially profitable. For high-ticket products (jewellery, made-to-measure furniture, professional equipment), selling remains consultative, but after-sales support can be automated.
Take a look at other use cases in the soluciones section if your business isn’t purely ecommerce but has an online sales component.
Frequently asked questions
How many abandoned carts can be recovered with AI?
Industry data indicates that between 10% and 25% of abandoned carts are recovered with automated follow-up via WhatsApp or email, depending on the sector and average order value. The key is timing: contact them between 1 and 4 hours after abandonment, with a personalised message that includes the cart contents and removes purchase friction.
Can a chatbot handle returns and exchanges?
Yes, it can take the request, validate the order, generate the return label and update the status on your platform. What it must not do is decide complex cases: if there is a dispute over the condition of the product or a request outside policy, the system hands over to a human with all the context gathered.
Which ecommerce platforms integrate with these systems?
Shopify, WooCommerce, PrestaShop and Magento are the most common. Integration is done via API or webhook: when someone abandons a cart, places an order or requests support, your platform notifies the AI system, which acts accordingly. There is no need to change platform.
How long does it take to get a system like this up and running?
A basic support and cart recovery chatbot is usually operational in 2 to 4 weeks. If you add automated after-sales, issue management and complex flows, the range goes up to 4-8 weeks. The factor that most extends the project is the quality of your product data and the clarity of your shipping and returns policies.
Do I need the WhatsApp Business API or is the regular app enough?
For automation you need the official API. The WhatsApp Business app doesn’t allow integration with external systems or scheduled sending at scale. The API does, and it also gives you account verification, templates approved by Meta and delivery metrics. The cost per conversation starts in cents, not euros.
Can AI recommend products the way Amazon does?
It can suggest products based on purchase history, cart contents or customer questions, but don’t expect Amazon’s level without massive data. In a shop with a medium-sized catalogue (fewer than 5,000 SKUs), the most effective recommendation is usually a business rule plus conversation context: if someone buys trail running shoes, offer technical socks, not a watch.
Next step
If you run an ecommerce business and want to automate support, recover carts or streamline after-sales, the first step is to map out which enquiries you receive, how many carts you lose each month and how much time you spend handling returns.
You can explore how to implement a WhatsApp chatbot integrated with your shop, review automation options beyond ecommerce, or book a free diagnosis via contacto to design a system tailored to your volume and sector.