Skip to main content

Should a shopping chatbot offer customers a game? What a Psychology & Marketing study found

Retailers are adding games to their chatbots in the hope that fun turns into sales. A survey and two experiments with working chatbots suggest the type of game decides the result: a quick spin of a wheel raised interest in buying, while a product quiz gave about the same result as a plain chatbot.

Elmashhara et al. find a spin-the-wheel game in a shopping chatbot raised purchase intention and shop visits, but a quiz matched a chatbot with no game.
Date

14 November 2026

Marketing in 3 · 2023 in research · No. 41 · Paper published 12 October 2023

Blue, yellow, green and red wooden pawns on a brightly printed race board with red and pale blue squares, two white dice beside a red pawn in sharp focus
Dice leave the outcome to luck, like the spinning wheel that worked best in the chatbot tests. Photo: Nik Korba on Unsplash.

The short answer

A quick game of luck can help; a quiz did not. In an experiment with 275 adults, a shopping chatbot that let people spin a wheel for their discount produced higher purchase intention than a plain chatbot offering the same reward, while a chatbot running a multiple-choice quiz only matched the plain one. In a follow-up experiment, the wheel also led more people to click through to the online shop.

Evidence level
  1. One study
  2. Several studies
  3. Meta-analysis

Key takeaways

  • After spinning a wheel in the chatbot, people rated their purchase intention at 5.07 of a possible 7, against 3.66 when the chatbot simply offered a discount with no game.
  • 78.1% of people who used the wheel chatbot chose to visit the online shop, against 50.0% after the plain chatbot and 49.3% after the quiz.
  • Quiz players felt less free to decide whether to play, scoring 2.95 on a two-question measure of choice against 4.01 for wheel users.
  • The quiz lifted emotional engagement to 5.06 on the same scale, the highest of the three chatbots, while the wheel scored 3.64; this is the kind of engagement linked to weaker buying intentions.

What the researchers did

Maher Georges Elmashhara, Susana C. Silva and Maria Levi Silva of Universidade Católica Portuguesa in Porto, Roberta De Cicco of the University of Urbino Carlo Bo in Italy and Maik Hammerschmidt of the University of Goettingen in Germany studied a growing habit in online retail: putting games inside shopping chatbots. The authors mention AliExpress, whose chatbot lets shoppers gamble for discounts or digital coins, and the lingerie brand Aerie, whose chat game hands out points and badges. Retailers hope the fun keeps people in the conversation and leads to a sale. The team asked whether it does, and which kind of game helps.

They built working chatbots in Facebook Messenger with the tool Chatfuel, all for an invented online shop selling suitcases under its own brand. People tapped through set options to browse products and play. In Study 1, a survey, participants watched a video of the chatbot, said what they expected from it, then used it for a shopping task and described how they had taken part. Studies 2 and 3 were experiments that set a plain chatbot against two gamified ones.

Participants
177 European consumers who often use Facebook Messenger, reached through referrals (Study 1); 275 and 218 adults recruited on Prolific (Studies 2 and 3)
Setting
Functioning Messenger chatbots for a fictitious online suitcase brand
Games
Study 1: a four-question product quiz, a spot-the-differences game and badges. Studies 2 and 3: a wheel to spin or a multiple-choice quiz, compared with a chatbot that had no game
Reward
In the experiments every participant won the same 30% discount, to use on a later purchase
Measured
Utilitarian and hedonic motivation, three kinds of engagement and purchase intention; also a sense of free choice (Study 2) and a click to the online shop (Study 3)

Not every kind of engagement helps

The authors split motivation in two. Utilitarian motivation is about getting a job done: finding what you are after and getting good value. Hedonic motivation is about fun and enjoyment. They split engagement in three. Cognitive engagement is concentration and absorption. Emotional engagement is enthusiasm, dedication and attachment. Behavioural engagement is the time and effort put into the chat, such as asking the bot for ideas and help, or thinking of recommending it to others.

Study 1 traced these links with structural equation modelling. People who expected the chatbot to be fun reported more behavioural engagement, and behavioural engagement went with higher purchase intention. People who expected it to be useful also reported more behavioural engagement, but more emotional engagement as well, and emotional engagement went with lower purchase intention. Usefulness therefore worked through two links with opposite effects. Cognitive engagement carried neither motivation through to buying.

Study 2 ran the same analysis on new participants and confirmed the main pattern. One link did not hold the second time: usefulness no longer raised purchase intention through behavioural engagement.

Wheel against quiz

In Study 2, 275 people were randomly given one of three chatbots. The plain one simply offered a discount. The second let people spin an animated wheel for their reward. The third asked several multiple-choice questions. The games were set up so that nobody lost.

The quiz drew more utilitarian motivation and more emotional engagement than both other chatbots, and less hedonic motivation than the wheel. The wheel had the lowest utilitarian motivation, cognitive engagement and emotional engagement of the three chatbots. It also produced the highest purchase intention. Two comparisons were not significant: the wheel was not rated reliably more fun than the plain chatbot (p = .17), and the quiz did not raise purchase intention above it (p = .34).

A serial mediation analysis traced both routes. Compared with the quiz, the wheel raised enjoyment, enjoyment raised behavioural engagement and that raised purchase intention. The quiz raised the sense of usefulness, which raised emotional engagement, which in turn lowered purchase intention.

Emotional engagement and purchase intention, by chatbot (1 to 7 scale)
Emotional engagement and purchase intention, by chatbot (1 to 7 scale)1234567Plain: emotional engagement: 4.14 Plain: emotional engagement4.14Wheel: emotional engagement: 3.64 Wheel: emotional engagement3.64Quiz: emotional engagement: 5.06 Quiz: emotional engagement5.06Plain: purchase intention: 3.66 Plain: purchase intention3.66Wheel: purchase intention: 5.07 Wheel: purchase intention5.07Quiz: purchase intention: 3.89 Quiz: purchase intention3.89
Emotional engagement and purchase intention, by chatbot (1 to 7 scale)
Group Value ()
Plain: emotional engagement 4.14
Wheel: emotional engagement 3.64
Quiz: emotional engagement 5.06
Plain: purchase intention 3.66
Wheel: purchase intention 5.07
Quiz: purchase intention 3.89

Study 2, 275 adults on Prolific, means from Table 4. All three chatbots ended with an identical discount.

Feeling free to play

Games can leave people feeling they have to play. Study 2 therefore asked two questions about choice: whether people felt forced to play the game, and whether they felt they could choose to play it. Wheel users felt more free than quiz users, and that sense of choice explained part of the wheel's advantage in purchase intention (indirect effect 0.43, 95% confidence interval 0.15 to 0.86).

The authors' explanation is effort. The quiz asked people to read questions and think about answers before the discount arrived, so taking part felt like an obligation. Spinning a wheel asked for one action and left the outcome to luck.

From intention to a click

Study 3 used the same three chatbots with 218 new participants on Prolific. After the chat, each person saw the brand in an online shop setting and chose between two buttons: visit the shop where the discount could be used, or not. The wheel led to more visits than either other chatbot (both p < .001). The plain chatbot and the quiz produced almost identical shares of visits (p = .93).

A click is closer to buying than a rating, and the authors added this study for that reason. It is still not a purchase: nobody spent money, and those who clicked were told afterwards that the chatbot was part of an academic project.

Share of people who chose to visit the online shop after the chat
Share of people who chose to visit the online shop after the chat020406080100Plain chatbot: 50.0%Plain chatbot50.0%Spin-the-wheel chatbot: 78.1%Spin-the-wheel chatbot78.1%Multiple-choice quiz chatbot: 49.3%Multiple-choice quiz chatbot49.3%
Share of people who chose to visit the online shop after the chat
Group Value (%)
Plain chatbot 50.0
Spin-the-wheel chatbot 78.1
Multiple-choice quiz chatbot 49.3

Study 3, 218 adults on Prolific, binary choice after the chat; percentages as reported in the text.

The authors' recommendations

Their main warning is against assuming that any rise in engagement becomes spending. The kind to build is behavioural: people asking the bot for information and help with their choice. The kind to handle with care is emotional. The authors suggest that strong emotional involvement can turn a shopper with a goal into someone browsing for the experience, who then loses sight of the purchase. They also see little return in making a chat more absorbing.

On the choice of game, they favour quick games of chance over games that test knowledge, and they ask designers to check that players do not feel obliged to take part. Both points rest on one product and two game formats, so they are best read as a strong starting point rather than a rule for every category.

Marketing in 3 steps

Ready. Set. Market.

ReadyDecide and prepare
  • Name the action the game should lead to. Write down the one thing a customer should do right after playing, such as opening a product page or putting something in the basket. If a game idea does not lead there, drop it.
  • Choose a game of luck. Pick a game that takes one tap and less than ten seconds, such as a wheel, a scratch card or a mystery box. Keep quizzes and puzzles out of the buying journey.
SetBuild it
  • Keep playing optional. Show the game as an offer with two buttons, Play and No thanks. Give product help and the standard offer to people who skip it, so nobody feels they have to play.
  • Put products next to the prize. On the screen that shows the prize, add two or three matching products and a button to the shop. Have the bot ready to answer questions on size, price and delivery.
MarketLaunch and measure
  • Compare it with a plain offer. Run the game for 14 days with every second chat user, and give everyone else the same reward without a game. Count clicks to the shop and orders for every 100 chats in each group.
  • Check who plays without buying. Each week, count the people who play more than once without opening a product page. If that number grows, allow one spin per visit or shorten the game.

Get it by email

New issues every Monday and Thursday, in your inbox. Free, no ads, unsubscribe anytime.

Your address is used only to send Marketing in 3. You will get one email to confirm.

What to expect

  • More interest, not proven orders. The wheel raised stated buying intentions and clicks to the shop. Completed purchases were not measured.
  • Taking part is not enough. Behavioural engagement was about as high with the quiz as with the wheel (no significant difference, p = .13), yet only the wheel lifted purchase intention.
  • A game on top of an offer. Every chatbot in the experiments handed over the same reward, so the evidence concerns a game layered onto a discount, not a game used instead of one.
  • Emotional pull has a cost. People who become absorbed in the game may linger without buying. This is the authors' explanation for the results, not something they observed directly.

Where it does not work

  • A practice purchase. Participants used a real chatbot, but no money changed hands; purchase intention and a click were the outcomes.
  • One product, one invented brand. Suitcases from a fictitious retailer were the only offer, so other categories may respond differently.
  • Two games compared. Only a wheel and a multiple-choice quiz were tested against each other; points, levels or leaderboards on their own were not.
  • Every player won. The games were set so that nobody lost, so how people react to losing a spin is unknown.
  • A survey for the paths. Study 1 measured motivation, engagement and intention in one questionnaire, so its paths are associations; the experiments tested the chatbot types directly.

Questions marketers ask

Does adding a game to a chatbot increase sales?

It depends on the game. A chatbot where people spun a wheel for their reward raised purchase intention and clicks to the online shop compared with a plain chatbot, while a multiple-choice quiz did not beat the plain version (Elmashhara et al., 2024). Real orders were not tracked.

How do utilitarian and hedonic motivation differ?

Utilitarian motivation is practical: reaching a goal, finding the right product and getting economic value. Hedonic motivation concerns enjoyment: whether the interaction seems fun, entertaining and immersive. Here both were measured as what people expected from using the chatbot.

Can customer engagement with a chatbot lower purchase intention?

One kind did. Emotional engagement, meaning enthusiasm, dedication and attachment during the chat, was linked to lower purchase intention in both the survey and Study 2. The authors suggest an absorbing experience can satisfy people enough that buying feels less pressing.

Why did a quiz work worse than a game of chance?

The quiz made the chat feel more useful and more emotionally involving, a combination that predicted weaker purchase intention. It also made people feel more obliged to play, because they had to read questions and pick answers before getting their reward.

Which kind of engagement should a shopping chatbot aim for?

Behavioural engagement: customers using the bot to look for ideas, information and assistance, or considering telling others about it. It was the dimension linked to higher purchase intention, while concentration and absorption showed no benefit.

Also new at the time

International Journal of Consumer Studies · 2024

The interplay of app characteristics and smartphone addiction in mobile shopping behaviour

Prodanova & Chopdar

A survey of 302 people in India who shop through mobile apps, analysed with PLS-SEM, asked which app features feed compulsive smartphone use and what that means for buying. Perceived privacy and security, personalization, offers tailored to the context and the retailer's reputation were all linked to stronger smartphone addiction. Combined with app incentives, that addiction went with a stronger intention to shop on mobile and more frequent purchases. Teams that design app rewards may want to look at how their heaviest users respond, not only the average shopper.

Electronic Markets · 2023

“Sorry, too much information”—Designing online review systems that support information search and processing

Seutter et al.

Shoppers facing hundreds of reviews can easily feel overloaded, and review systems now add features meant to make scanning and reading easier. Drawing on 65 review systems from a range of online platforms, the authors built and evaluated a taxonomy of these design features, with 50 distinct characteristics. It also sums up what is already known about how such features affect information overload. Anyone redesigning a review section now has a structured list of options to set against their own.

Source

Elmashhara, M. G., De Cicco, R., Silva, S. C., Hammerschmidt, M., & Silva, M. L. (2024). How gamifying AI shapes customer motivation, engagement, and purchase behavior. Psychology & Marketing, 41(1), 134-150. https://doi.org/10.1002/mar.21912

This brief is written for practitioners from the published article. Open access (CC BY 4.0): free to read at the publisher. Numbers are as reported by the authors; read the article for full methods.

Keep reading

More issues on AI · Retail

Bruno Schivinski

Bruno Schivinski is Associate Professor of Marketing at Gdańsk University of Technology. He studies how consumers engage with brands in digital and social media.

About Bruno · All Marketing in 3 issues · Subscribe