Chatbots, or virtual assistants conducting conversations with customers, have become a permanent feature in the world of Polish e-commerce.

Their development from the mid-2010s to the mid-2020s has undergone a significant evolution – from simple, script-based programs to advanced assistants powered by artificial intelligence. Below we present a report divided into three key periods: before 2022, the years 2022–2023, and the years 2024–2025. For each of these stages, we discuss the chatbot solutions in use, dominant features, expectations of retailers and customers, as well as the most important changes over time. At the end, we summarize the key trends and highlight the e-commerce sectors that stand out in terms of chatbot adoption.

Period I: Before 2022 – the beginning of chatbots in Polish e-commerce

The first chatbots in Polish online stores began to appear in the second half of the 2010s. This was a time of experimentation and learning – both for companies implementing them and for customers learning to use a new form of communication. In the initial phase, simple, script-based solutions dominated (often referred to as "first-generation chatbots"). These chatbots most commonly operated within messengers such as Facebook Messenger – according to a 2018 report, as many as 16 out of 18 examined Polish bots were integrated with Messenger (Polskie Chatboty 2018 – report on the most interesting Polish implementations | K2). The popularity of Messenger stemmed from its widespread use and the platform’s openness to bots; many companies chose this channel to reach customers where they spent their time. Gradually, bots embedded directly on websites (in live chat windows) and the first integrations with voice assistants on helplines also emerged (Polskie Chatboty 2018), although the latter were less common in e-commerce.

Solutions in use: In the early days, chatbots were mostly based on predefined scenarios and basic keyword recognition. Many did not use advanced NLP (Natural Language Processing); users often had to choose from provided options or enter specific phrases for the bot to understand them (Polskie Chatboty 2018). These were so-called block or process chatbots – guiding customers step by step through a fixed flow (e.g., selecting a product, providing address details) without allowing free-form conversation. For example, a chatbot might help place an order for a simple product: the customer would choose a category, product, and provide delivery information through a series of selectable prompts (Polskie Chatboty 2018). Another common use was notification bots – a counterpart to a newsletter delivered via messenger. Such bots could inform customers about shipment status or send daily deals or tips, often with high engagement thanks to the less crowded Messenger channel (Polskie Chatboty 2018).

Dominant features: In the first development phase, the most common chatbot functionalities in e-commerce focused on answering frequently asked questions (FAQ) and simple advisory roles. Bots acted as 24/7 customer assistants – providing information about product availability, payment methods, delivery options, return policies, etc. This relieved support teams from repetitive questions and reduced customer service costs (Chatbots in e-commerce – how do they perform in online stores? – Komerso.pl). Some were also used to support sales in a limited capacity – for example, as virtual advisors helping choose products or gift guides suggesting presents based on criteria (such a scenario was implemented by the gift shop Wyjątkowy Prezent) (Polskie Chatboty 2018). Notably, even at this stage, chatbots were used for collecting marketing leads (e.g., Orange Polska’s bot gathered contacts for telecom services) and initiating contact with customers (sending promos and alerts). However, most of these early bots focused on narrow tasks – each “learned” a specific domain. For instance, one bot might only handle order status inquiries, while another assisted with product selection. Personalization was still in its infancy – advanced product recommendations were not yet widespread, although some bots could remember a user's name or previous choices to tailor future interactions. As early as 2018, it was emphasized that a well-programmed chatbot could remember a customer's shopping preferences, allowing it to offer personalized suggestions on return visits (e.g., size, style, color) – though in practice, this required integration with user accounts or databases, which was not always present in the earliest implementations.

Expectations of stores and users: E-commerce stores that decided to implement a chatbot before 2022 primarily hoped to improve the availability of customer support. Traditionally, customer service was only available during business hours on weekdays, while most online shopping happened in the evenings and on weekends. Chatbots were meant to fill this gap, providing 24/7 support – so customers were not left without help outside of working hours, reducing the risk of cart abandonment due to unanswered questions. There was also an expectation that bots would speed up support – offering instant answers to simple queries instead of making users wait for email or call center responses. For e-shops, automation and cost reduction were strong incentives: a well-designed bot could handle up to several dozen percent of inquiries, translating into reduced call center staffing at scale. At first, users approached chatbots with curiosity but also caution. Many found them a convenient source of quick information – appreciating the simplicity of asking a question and getting an immediate answer without browsing through the FAQ. On the other hand, the limitations of early bots could lead to frustration – if the bot didn’t understand a question outside the programmed script, the conversation stalled. That’s why it was crucial for bots to hand off smoothly to a human agent when needed. Many companies designed their solutions this way: the chatbot acted as the first line of support handling simple matters, and in case of a more complex issue, transferred the conversation to a human consultant (the so-called hybrid model). Users quickly adapted to this role division and accepted it as long as the handover was seamless.

Example implementations in Poland (before 2022): Even before 2020, several pioneering chatbot implementations appeared in various Polish e-commerce sectors. For instance, the sporting goods retailer Decathlon introduced a simple bot responding to questions about product availability (using prewritten answers – a block-based solution) (Chatbots in e-commerce – Komerso.pl). The luxury fashion store Moliera2 launched a chatbot on its website that greeted customers and offered help with topics ranging from product questions to loyalty programs and discount codes. Shoe retailer CCC deployed a Messenger-based bot integrated with its product catalog. Users could type what they were looking for (e.g., “red heels” or even “shoelaces”), and the bot could find relevant items, showing pictures, prices, and buy-now buttons. This was an early preview of more advanced sales functions – quite innovative for its time.

Another interesting case was the virtual assistant from Travelplanet.pl (tourism sector), which helped customers select vacation offers – the user interacted with the bot by specifying preferences (destination, dates, type of trip), and the bot suggested matching packages with images and booking links. In the gift sector, the mentioned Wyjątkowy Prezent used a bot as a gift idea finder – asking users about the occasion or the recipient’s interests and suggesting appropriate products/services.

Beyond strictly online stores, customer service bots also appeared in services related to e-commerce: for example, logistics company InPost launched a bot (on Messenger and its website) for tracking parcels and answering locker-related questions, while furniture retailer IKEA experimented with a chatbot offering product advice and FAQ support. These examples showed that fashion, footwear, and sporting goods quickly adopted chatbots (often as interactive product catalogs with FAQ support). The tourism and gift sectors also stood out for their creative use of bots to inspire customer interest.

However, these pre-2022 implementations were largely pilot projects – used mainly by larger players and market innovators testing how the technology would impact sales and customer satisfaction.

Example: Chatbot from CCC shoe chain operating on Messenger (circa 2019). The bot greets the customer and suggests a help topic, and can also search products – when asked for “shoelaces,” it displayed relevant accessories along with the option to purchase online.

Period II: 2022–2023 – the mainstreaming and AI shift

The years 2022–2023 marked the transition of chatbots in e-commerce from technological novelty to mainstream tool. Increasingly, not only large but also medium and smaller retailers began implementing virtual assistants, drawing on the experiences of early adopters and emerging “chatbot as a service” platforms. According to expert estimates, by the end of 2022, most Polish online stores were already using some form of chatbot on their websites or in messaging apps (E-handel w 2023 r.: cross-border, AI, omnichannel, chatboty... – Omnichannel News). Chatbots were no longer seen as an experimental feature but began to be treated as an almost obligatory element of a professional e-store—much like social media or mobile apps once were. Retailers recognized that the growing scale of e-commerce (further accelerated by the pandemic in 2020–2021) generated such volumes of customer inquiries that automating service became a necessity.

Solutions used: This period saw a clear qualitative leap in chatbot technology. A wave of AI-powered tools emerged—sometimes referred to as “second-generation chatbots” (Polskie Chatboty 2018 – a report on the most interesting Polish implementations | K2). In practice, this meant that e-commerce bots began using improved NLP engines to understand natural language and conversational context. As a result, users were no longer limited to rigid commands or clickable options—they could increasingly phrase their questions freely, and the chatbot was able to interpret them correctly. Platforms became available in Poland that enabled training a chatbot based on company documentation, FAQs, and product catalogs, meaning the bot could answer hundreds of product-related questions without manually programming every variant.

Integration with internal store systems also increased: chatbots connected to databases could, for example, check order statuses by number or provide a customer’s current loyalty points balance—tasks that previously required calling a helpline. At the same time, ready-to-use SaaS solutions offered by specialized companies gained traction—rather than building bots from scratch, e-shops could use Polish platforms such as SentiOne or K2 (PerfectBot), or international ones like LiveChat/ChatBot, or messenger-integrated solutions (e.g., Meta offered an improved Messenger API for businesses). These tools increasingly featured user-friendly bot training interfaces, conversation analytics, and ready-made integrations (e.g., plugging a bot into a Shopify or IdoSell store). In 2022, messaging apps such as WhatsApp gained significant popularity—many stores began offering chatbots there as well, especially since WhatsApp opened its Business API. Overall, chatbot touchpoints became more omnichannel: the same virtual assistant could serve a customer on the website, Messenger, WhatsApp, and even in the store’s mobile app.

A major technical breakthrough came at the end of 2022 with the public release of ChatGPT—this advanced AI demonstrated the possibility of truly natural conversation with a machine. Already in 2023, first attempts appeared to use GPT models in e-commerce customer service (e.g., some Polish stores began pilot-testing ChatGPT to generate answers to complex questions or to provide translations). This trend was still in its infancy in 2023, but it clearly pointed to the direction of future development.

**Dominant functions: ** In 2022–2023, the range of chatbot capabilities expanded significantly. In addition to their ongoing core role—instantly responding to FAQs—bots increasingly acted as active shopping advisors. The assistant could greet a visiting customer and ask if help was needed, then suggest product categories to browse—personalizing the welcome message based on purchase history or previously viewed items (E-handel w 2023 r.: cross-border, AI, omnichannel, chatboty... – Omnichannel News). Personalization became a key concept: chatbots began using customer data to offer products tailored to individual needs (E-handel w 2023 r.: cross-border, AI, omnichannel, chatboty... – Omnichannel News). For example, returning customers could immediately be shown new items from their favorite brand or prompted to repurchase complementary products.

Bots also became more effective salespeople—many stores used them not just reactively (answering questions), but proactively for upselling and cross-selling, i.e., suggesting higher-end products or accessories that matched the customer’s basket (E-handel w 2023 r.: cross-border, AI, omnichannel, chatboty... – Omnichannel News). For instance, if a customer inquired about a phone model, the bot might immediately ask whether to add a case or headphones to the order. A visible 2023 trend was integrating chatbots with other sales ecosystem elements: payments and checkout within the chat window. Although full transaction processing in chat was not yet widespread, integrations emerged that allowed sending a quick payment link or generating an order in the system upon the customer’s confirmation.

Another increasingly common feature was post-sale support: chatbots facilitated returns or complaints (e.g., guiding the customer through the required steps, providing downloadable return labels, etc.). Naturally, all these new capabilities were enabled by the growing “intelligence” of bots—thanks to machine learning, they could learn from conversations, better recognize customer intent, and understand more complex queries.

In summary, the e-commerce chatbot in 2023 could perform a multifaceted role: acting simultaneously as a customer service agent (FAQs, order statuses, technical issues), sales advisor (product suggestions, promotions), and customer care assistant (help with returns, collecting post-purchase feedback).

Expectations of stores and users: During this period, expectations for chatbot quality increased. As retailers saw their competitors implementing similar solutions, they sought to stand out by offering a better conversational experience. It was expected that a modern chatbot would sound like a “real person”—making the conversation more natural and less formulaic (E-handel w 2023 r.: cross-border, AI, omnichannel, chatboty... – Omnichannel News). Businesses paid attention to refining the bot’s language and tone to match their brand image and foster positive relationships.

Large online stores also began demanding multilingual support from their bots—driven by the cross-border trend (expanding into foreign markets). The ideal bot would respond fluently in Polish, English, or Czech—depending on the customer’s location—and thereby support international sales (E-handel w 2023 r.: cross-border, AI, omnichannel, chatboty... – Omnichannel News). Another growing expectation was measurable performance: companies wanted to see tangible results, such as increased conversion rates or a reduced number of conversations requiring human intervention. As a result, greater attention was paid to conversation analytics, metrics like FCR (first contact resolution), and average handling time.

By 2022–2023, users were also far more familiar with chatbots. Many customers expected to be greeted by a bot upon opening a site chat—and often preferred this for simple matters, as it was associated with quick resolution. Thanks to improved AI, the frustration caused by misunderstanding decreased—bots less frequently replied “I don’t understand the question,” which improved user trust in the channel. Naturally, the option to talk to a human remained key—in 2023, it became standard for users to request “connect me with a human” at any moment, with the bot respecting this request, transferring the conversation or submitting a ticket for follow-up.

User expectations were also shaped by the growing popularity of voice assistants (Google Assistant, Siri)—consumers increasingly desired similar convenience in e-commerce, e.g., dictating questions instead of typing, or receiving spoken responses. Not all stores were ready for this, but the direction was clear: increasingly natural, “human” interaction.

In summary, in 2022–2023, chatbots ceased to be seen as gimmicks and became full-fledged elements of customer service and sales—facing specific business expectations (higher sales, customer satisfaction) and needing to keep pace with a dynamic environment (multilingualism, personalized communication, etc.).

Examples of implementations (2022–2023):

During this period, virtually every major e-commerce platform in Poland had implemented some form of chatbot. Allegro, the largest marketplace, launched an improved assistant to help users in the Help Center (automatically answering questions about orders, payments, returns). Empik—the leader in books and multimedia—implemented a bot that assisted with store navigation and recommended products from various categories based on customer preferences. InPost significantly expanded its chatbot for parcel handling—by 2022, it was one of the best-known bots, serving hundreds of thousands of users with 24/7 parcel tracking and pick-up code retrieval. IKEA Poland also used a virtual assistant to answer questions about products and in-store availability—although it primarily sells offline, many online users sought information through the chatbot (Chatbot w sklepie internetowym).

Industry-specific solutions also appeared. For example, in 2023 the e-commerce platform IdoSell announced the development of its own AI “e-seller” for stores using its system (E-handel w 2023 r.: cross-border, AI, omnichannel, chatboty... – Omnichannel News). Notably, in 2023, Polish digital agencies and software houses began building GPT-3/ChatGPT-based bots for the first bold retailers—chatbots capable of generating unique responses even to unusual questions, using a vast language knowledge base. Though still pilot projects, they indicated growing interest in generative AI in e-commerce service.

Among the standout sectors—besides fashion and electronics, which continued to intensively use bots—the financial and insurance sectors (offering online products) also invested in assistants for things like insurance or credit offers. However, in the core e-commerce field, fashion and footwear proved particularly successful in using chatbots for shopping inspiration (e.g., suggesting outfits, collections), while the electronics and home appliances sector used them to help filter a wide range of products (e.g., “which 55-inch TV do you recommend?”—the bot could suggest several matching models). Wherever the assortment was extensive and involved many technical questions, chatbots proved highly useful for quickly delivering specific information.