Do Recommander Systems Manipulate Consumers Preferences?

Recommender systems are important in the decision making. Those systems provide the customers suggestions. As a result of this, firms can better serve their customers. This will also lead to an increase in sales. Research has focused on the development and the improvement of recommender systems. But there is not quite much studied about the behavioral implications of using recommender systems. Are those recommender systems manipulating the customers?

Most recommender systems use the consumers ratings items as input. Those recommendations the system provide present an expectation of how well a customer will like an item. There is also a feedback loop, those are the actual ratings of the customer after the purchase has been made. The recommendations are based on the actual ratings of a customer who already has tried this product or item.

There is a possibility that people are influenced by elements in the environment when they make a decision. The first one is the anchoring issue. People are maybe consumed drawn, because the system is presenting an item to them and they choose it. It is important to know if a customer really likes it or just chooses it because it is presented in a advantageous way. It is difficult for people to see if an item is reasonable for them. It could be presented as a advantageous choice, but it will maybe be the opposite!

When there is uncertainty, a customers seeks for the most plausible item. The suggested item is viewed as the ‘correct’ answer, therefore a lot of people will choose this. The users belief that the recommender system will choose the right option for them, therefore they choose what the systems presents to them.

Users that will receive a high recommendation from the recommender system, will also give higher rating after they bought/used it and vice versa. What they saw as a rating, has influence on their own rating. They are biased. This is although not symmetric.

There is an significant effect when the recommendation is raised, but not when the recommendation is lowered. It is notable that this effect is not only taking place when the uncertainty is high, it also operates at the point of consumption.

Also is the reliability of the system important. When the system is known as reliable, the customers’ ratings will be more close to the recommendation the customer has seen before. When the system is thus less reliable, the ratings given by the customer will be less close to the recommendation.


Adomavicius, G., Bockstedt, J. C., Curley, S. P., & Zhang, J. (2013). Do recommender systems manipulate consumer preferences? A study of anchoring effects. Information Systems Research, 24(4), 956-975.

Cosley D, Lam S, Albert I, Konstan JA, Riedl J (2003) Is seeing believing? How recommender interfaces affect users’ opinions. Cockton G, Korhonen P, eds., CHI 2003 Conference, Fort Lauderdale FL (ACM, New York), 585–592.

The number of Facebook friends on crowdfunding success

(*This entry is based on the research article ‘The Dynamics of Crowdfunding – An Exploratory Study’ by Ethan Mollick)

In order to make something work, one aims to find a recipe for success. This principle holds for crowdfunding, too, in which founders of all sorts of projects request funding from many individuals, often in return for future products or equity (Mollick, 2014). Many crowdfunding projects, however, fail. Therefore, it is of importance to find out the underlying dynamics of success and failure among crowdfunding ventures. This is exactly what Ethan Mollick, Professor of Management at Wharton University of Pennsylvania and author of The Dynamics of Crowdfunding – An Exploratory Study, has done. By analysing a dataset containing 48,500 crowdfunding projects with a combined funding over $237 M, Mollick researches the effects of a fund seeker’s personal network, underlying project quality and geography on successful fundraising. In this post, I will focus attention on the effect of a fund seeker’s personal network through the notion of one’s number of Facebook friends. Then, I’ll show how this looks like in practise on Finally, I suggest a way in which the power of an entrepreneur’s personal network could be even better put to use.

Ethan Mollick

(Professor Ethan Mollick)

Social capital

Social networks have long played an important role in the funding of new ventures (Hsu, 2007; Shane and Cable, 2002). An entrepreneur’s social network influences the succes of raising capital, as it provides (1) connections to funders and resources as well as (2) endorsements of project, its product or service, and the initiator (Shane and Cable, 2002; Sorensen and Fassiotto, 2011; Stam and Elfring, 2008). Actually, an entrepreneur’s social network is the initial source of funding, called friends and family money (Agrawal et al, 2010). As Mollick found, about one in three accounts are linked to social network Facebook. Hence, the author looked at Facebook friends of founders (FBF) for the project initiator, as this number is less likely to increase as the project progresses. Here, FBF is a measure of the size of a founder’s social network. Models 2 and 5 in Mollick’s results (see table below) show that social network size predict success. According to the author, the link between social network size and crowdfunding succes could be compared to the following. Having just 10 Facebook friends leads to 9% chance of succes, whereas a 100 Facebook friends lead to 20% of success. With 1000 Facebook friends denoted on, a fund seeker has 40% change of success. However, Model 6 in Mollick’s results (see table below) shows that having no Facebook account coupled to is yet better than just having few online connections. This suggests that, although larger networks generally lead to more success in fundraising, entrepreneurs yet need to strategize on whether or not linking their social network to their fundraising, based on the number of friends they have on Facebook.

Results Mollick

If you are interested in how the number of Facebook friends is depicted on, visit:

Here you see the SnapRays project on by entrepreneur Jeremy Smith. Click on Jeremy’s photo on the right to view his profile. The number of Facebook friends is depicted on the right.

A suggestion

Could an entrepreneur’s social network be leveraged more as to earn more trust among investors and hence raise more capital? In addition to stating the number of the fund seeker’s Facebook friends, the crowdfunding platform could enable the fund seeker to show the number of steps and the actual relations between him or her and a particular potential funder the way LinkedIn depicts the relations between you and someone else. To get an idea, see the mock-up I made below. To my belief, this would give a potential funder a feeling of being ‘more connected’ to the fund seeker, hence it would raise trust and it might lead to more funding.*

*Note: I e-mailed Professor Ethan Mollick about this suggestion. I’ll update this post if he replies.


Your turn

Now, could you think of other reasons why the number of Facebook friends is a quality signal to potential investors? And could you imagine different ways in which a fund seeker’s personal network could be leveraged more on crowdfunding platforms? Let me know you thoughts in the comments below.


  • Agrawal, A., Catalini, C., Goldfarb, A., 2010. The geography of crowdfunding. SSRN Electronic Journal.
  • Hsu, D., 2007. Experienced entrepreneurial founders, organizational capital, and venture capital funding. Research Policy 36.
  • Mollick, E., 2014. The Dynamics of Crowdfunding – An Exploratory Study. Journal of Business Venturing 29, 1-16.
  • Shane, S., Cable, D., 2002. Network ties, reputation, and the financing of new ventures. Management Science 48, 364.
  • Sørensen, J., Fassiotto, M., 2011. Organizations as fonts of entrepreneurship. Organization Science 22, 1322–1331.
  • Stam, W., Elfring, T., 2008. Entrepreneurial orientation and new venture performance: the moderating role of intra-and extra industry social capital. Academy of Management Journal 51, 97–111.

Can co-creation help in service recovery?

This post is about understanding the co-creation effect: when does collaborating with customers provide a lift to service recovery? By Roggeveen, A., Tsiros, M. and Grewal, D. (2011) 

Co-creation is the process of jointly producing a mutually valued outcome. This means that consumers can help shape or personalize the content of their experience, in other words, it can affect satisfaction of a recovery. A recovery is required when equity of a customer-company relationship is damaged. The research investigates the impact on equity for service delays outside the company’s control in the airline industry.

Study 1:

The first study demonstrates that using co-creation during severe and less severe delays yields different results. When consumers face a sever delay, co-creation improves the evaluation of the company, whereas less severe delay evaluations are not affected. Furthermore, the study demonstrates that co-creation works as good as compensating customers for severe delays. When faced with less severe delays, co-creation performs just as well or even better than compensating a customer.

Study 2:

The second study expands on the first study by expanding the knowledge on repurchase intentions, after co-creation recovery. The study demonstrates that recovery extends beyond the increased evaluations, it increases consumers repurchase intentions as well. This relationship is however mediated by equity, which is the difference between what consumers have received and what they expected to receive.

Study 3:

The third study looks at the perception of consumers of the recovery task. When consumers view the recovery strategy negatively, hence if the view it as akin to work, co-creation can harm evaluation. This is especially the case when consumers are asked to co-create during less severe delays.

Study 4:

The fourth and final study is the most interesting. The researches explore what happens when a company does not meet or exceeds the customers’ expectations after a recovery or compensation. As expected, companies that do not meet a customer request achieve lower evaluations. However, when a company exceeds customer expectations, the customer does not evaluate the company higher than when meeting customer requirements.

Especially the last study has large managerial implications. Consumers evaluate a business the same as when a company meets or exceeds the requirements set by the customer, which means that it is far more economical to only meet the necessary requirements. From this study we can therefore learn that using a co-created recovery for severe problems achieves good results for both customer evaluations and repurchase intentions. As noted by a manager: “if a guest perceives a failure to be just horrible, then they want to be more involved in rectifying the problem. If they perceive that it’s a minor inconvenience, then they don’t want so much input on it.”


Roggeveen, A., Tsiros, M. and Grewal, D. (2011). Understanding the co-creation effect: when does collaborating with customers provide a lift to service recovery?. Journal of the Academy of Marketing Science, 40(6), pp.771-790.


Heading picture: [Accessed on 02-04-15]

Why do people fill in reviews on online platforms?

With the new Internet technologies, traditional word-of mouth communication has been extended to electronic media, such as online discussion forums, electronic bulletin board systems, newsgroups, blogs, review sites, and social networking sites. Everyone can share their opinion and experience related to products with complete strangers who are socially and geographically dispersed this new form of word of mouth, known as electronic word of mouth (eWOM). This research is about eWOM which has become an important factor in shaping consumer purchase behavior. In early research is found that information provided on consumer opinion sites is much more influential among consumers nowadays.

For instance, eMarketer revealed that 61% of consumers consulted online reviews, blogs and other kinds of online customer feedback before purchasing a new product or service. In addition, 80% of those who plan to make a purchase online will seek out online consumer reviews before making their purchase decision (Infogroup Inc, 2009). Some consumers even reported that they are willing to pay at least 20% more for services receiving an “Excellent”, or 5-star, rating than for the same service receiving a “Good”, or 4-star rating (Comscore Inc, 2007).

Cheung et al. (2012) stated that we do not fully understand why consumers spread positive eWOM in online consumer-opinion platforms. Among the few existing publications, eWOM behavior is primarily explained from individual rational perspective with the emphasis on cost and benefit. Consumer participation in online consumer-opinion platforms depends a lot on interactions with other consumers. But why do people participate and are what stimulates consumers eWOM intentions?

The following variables were defined in this research as influencers of consumers’ eWOM intentions: Reputation, Reciprocity, Sense of Belonging, Enjoyment of helping, Moral Obligation and Knowledge Self-Efficacy. To test their theoretical framework they conducted a research using a sample of online consumer-opinion platform users from In total they collected 203 usable questionnaires.


After this study three variables were found significant: Reputation, Sense of belonging and Enjoyment of Helping. Sense of belonging had relatively the most impact on consumers’ eWOM intention. The result is consistent with previous eWOM marketing literature, where sense of belonging is an essential ingredient that creates loyalty and citizenship in a group. Also enjoyment of helping others is crucial in affecting consumers’ eWOM intention. Intentions to write about dining experiences in demonstrate enjoyment of helping others. Consumers can benefit other community members through helping them with their purchasing decisions. Reputation is a small factor affecting consumers’ eWOM intention. This can be explained by some consumers want to be viewed as an expert by a large group of consumers.

The results of this research can be practical relevant in different ways. Online consumer-opinion platform could allow consumers to create their own personal profile to create a stronger sense of belonging to the group. Also platforms could apply reputation tracking mechanisms, so ‘’experts’’ can be found more easily. And last, the platform could provide a mechanism for contributors so readers can show their appreciation for the received reviews, like a chat.

– Cheung, C. M., & Lee, M. K. (2012). What drives consumers to spread electronic word of mouth in online consumer-opinion platforms. Decision Support Systems, 53(1), 218-225.

– ComScore Inc., Online consumer-generated reviews have significant impact on offline purchase,, Online review sway shoppers, accessed.


Individual Self-Design vs Community Self-Design

The customer as a co-creator is becoming more important. Self-design is a new trend. Nowadays customers can customize anything; from self-designed skis (e.g. Edelwiser), to suggesting preferable food flavours (e.g. Lays). Many companies offer their customers a so-called Mass Customization (MC) toolkit to design their own products online. But isn’t it extremely inefficient and difficult to create all these self-designs separately? Isn’t it extremely costly in terms of time and money, for company and customer, to make use of this isolated, dyadic interaction process between an individual customer and the Mass Customization toolkit?

When you are less experienced in designing your own product, you will cope with a lot of difficulties. If you lack experience and/or creativity, you have to create ideas by brainstorming with other people and get inspiration from existing designs on the Internet. If someone asks an inexperienced person to design its own product, and they have to start from scratch they would definitely have a hard time figuring out where to start and what their actual preferences are. This is mainly because there are a lot of possibilities. Many people would start with designing different alternatives but this trial-and-error method is really time-consuming and not effective.

Above-mentioned statements and arguments are reasons for Franke, Keinz and Schreier (2008) to do some research about the Mass Customization toolkit. They thought about how to improve these toolkits, and they did some research whether it would be useful to include user communities.
They found out that peer-generated design solutions and peer-based feedback should be included in the existing MC toolkits because it would influence self-designs positively. More customers are able to design their own products by either adapting or getting inspired by other designs. Other users’ designs can be a great starting point for the less experienced designer. Think about the customized shoes from Nike or Vans. Before you create your own design, you will see a few designs that already have been created. There is an option to adapt these models or adapt a professional design. In other words, Nike/Vans creates a starting point for the less experienced designers and this will help customers to get a better outcome.

The peer-input can also be used as an external feedback channel. Through this way customers are able to show their preliminary product to others, who can help to improve the product. An example is SoundCloud. Anyone can upload their sounds and music and at the same time people are able to criticize and comment each other’s music. One of the community guidelines of SoundCloud is ‘Criticize, but do it constructively’. Because of the user community feedback, anyone can improve their songs. Schermafbeelding 2015-03-27 om 17.20.36 Figure 1. ‘Write your comment on SoundCloud’ (SoundCloud, 2015)

By integrating user communities in the MC toolkit, it is proved that customers are better able to create a more systematic problem-solving behaviour and it leads to self-designed products that meet the preferences of the customer more effectively.

Finally, there is one perfect example of a company that is making use of the integration of a user community: Threadless. When a potential customer starts to create their self-design, he has access to all the designs that have been created in the past. He is allowed to use and adapt other designs. After that the self-designer can ask the ‘Threadless community’ for help: ‘Do they like your design?’ and ‘Do they have any tips for improvements?’ Threadless is integrating both existing solution chunks and external feedback in the problem-solving behaviour. In the end this will lead to a more satisfied customer and the customer will value the product higher based on the perceived preference fit, purchase intention, and willingness to pay.


Customer Empowerment at the University-Spar!

One way to involve active consumer participation is to let the consumers vote.

The supermarket Spar uses this consumer empowerment strategy at our very own university. Hereby, the Spar selects a product and visitors of the Spar can “like” or “dislike” the product by means of pressing a button at a touch screen. When the product receives over 50 likes, the Spar will add it to their assortment. When looking at the customer empowerment matrix by Fuchs & Schreier (2011), this strategy falls in the lower right corner.

Since the Spar gathers information and opinions about their customers, this strategy could be considered as a way of data gathering by means of crowdsourcing. However, I think this tool is mainly used as a marketing tool instead. To investigate whether consumers would like this product or not, Spar could also do a pilot by adding the product for a limited time to their assortment and after that looking at the sales figures whether this product is desired by the customers or not. Besides that, there is a possibility that consumers “like” the product without actually buying it.

The marketing aspect of this tool unfolds in two ways. Firstly, it functions as a promotion campaign for that specific product, since the product is given extra attention in the store. Customers of the Spar may buy it now, because the product is put in a spotlight, whereas they may not have bought it if it was, just like the other products, regularly in the shelves. This is a typical example whereby a store pushes the product towards the consumer (Balugly & Uysal, 1996). Secondly, this marketing tool can have a favorable impact on the image of the Spar as a supermarket itself. According to Fuch & Scheier (2011), (even passive) consumers perceive a higher level of customer orientation, more favorable corporate attitudes, and stronger behavioral intentions when firms empower their customers in such way. Even if consumers do not like the product or do not use the tool, it can still have a positive impact on the brand image of Spar, since consumers feel that Spar integrates consumer’s opinion in assortment selection. Interesting to see is, that there is no marketing found online about this tool. The only place where one can know about this campaign, is by being in the store physically. This is something that Spar could improve, by making this campaign more visible, as Claire Gilby (2012) E.ON executive of stresses: The biggest mistake E.ON could make was to not be visible.

To conclude, with having such a tool with already existing products, I would think that its aim is more for marketing purposes, instead of gathering data about the product or consumers. However, purposes and outcomes could be different with new product development.



Baloglu, S., & Uysal, M. (1996). Market segments of push and pull motivations: A canonical correlation approach. International Journal of Contemporary Hospitality Management8(3), 32-38.

Fuchs, C., & Schreier, M. (2011). Customer empowerment in new product development. Journal of Product Innovation Management, 28(1), 17-32

Electronic Word of Mouth and

Online star ratings

Figure 1: ‘Understanding’ online star ratings

Amblee and Bui (2011) have researched the effect of electronic word of mouth (eWOM) on the sales of digital microproducts. They studied ‘shorts’: short stories (e-books) that are made available for a set price of 49 cents. They classify these shorts as digital microproducts.

This article focusses on their study on de effect of social commerce on product reputation and sales (hypothesis 1). Amblee and Bui (2011) studied this effect from three different perspectives: Valence (how positive or negative a rating is), the presence of a rating versus no rating and volume (thus the amount of ratings). Interestingly, they do not find a significant correlation between valence of the rating and sales. Amblee and Bui (2011) suggest that this might be due to the generally positive ratings, and thus a low variance between positive and negative ratings. Second, they also find that the presence of a rating is a good predictor of higher sales, as compared to no ratings. Furthermore, they also find a significant correlation between the volume of ratings and the volume of sales.

In their discussion, Amblee and Bui (2011) propose a better scoring system which allows users to score the e-book on different dimensions such as content, writing style and so on. While this suggestion might lead to a bigger variance between ratings, I wonder if it would have a positive effect on sales in the end. Amblee and Bui (2011) point out that the majority of past research on valence suggests that valence is not a reliable predictor of sales. However, by including more ratings to fill in like Amblee and Bui (2011) suggest, you might achieve a negative effect: that customers are no longer willing to fill in the rating. And the importance of the presence of the rating, and moreover the volume is exactly what is found to be so important to spark sales by Amblee and Bui (2011).

Fast forward to 2015. Did change the rating system? I went to and I filtered on short stories and on kindle editions. This is what I found:

Amazon ratings of Shorts 2015

Figure 2: Average customer review of shorts on, 2015

Amazon blog2

Figure 3: Layout of customer reviews of shorts on, 2015.

This shows that the great majority of the ratings is still very high (roughly 84% of the ratings are 4 stars or more).  However, it seems like Amazon has added some space for customers to motivate their rating. Furthermore customers can also identify reviews as helpful, and Amazon shows me the most helpful reviews first. By using this system, leaves it up to its customers if they want to motivate their rating (and spend more time rating a book) or not. What do you think, has the system improved? Do you think it will lead to more sales? Please comment below!


  • Amblee, N. and Bui, T. (2011) ‘Harnessing the influence of social proof in online shopping: The effect of electronic word of mouth on sales of digital microproducts’, International Journal of Electronic Commerce, Vol. 16, No. 2, pages 91-113, DOI 10.2753
  • Featured image: GTP Headlines, accessed 31-03-2015,
  • Figure 1:, accessed 01-04-2015
  • Figure 2:, accessed 31-03-2015,
  • Figure 3:, accessed 31-03-2015,