Online valuation systems are overrated. Most systems are flawed and become less and less important as trust in the brands of platforms like Airbnb and Uber grows.
At the time, in 2007, when Brian Chesky and Joe Gebbia came up with the idea of creating an online platform on which anyone could rent a room to a completely unknown tourist, they drew many odd looks. How could you know an overnight guest can be trusted? In the mean time, over 100 million nights have been booked through the platform in 191 countries. I’m talking about Airbnb. The secret: online trust through valuation systems.
Sharing rooms and houses has been existent since time began, but thanks to Airbnb the threshold has been lowered and the activity grows exponentially. There is a growing number of online platforms that link up individuals using the same principles. Uber links up drivers with people who want to travel from A to B and the Dutch SnappCar links up car owners to people who need a car for just a short time. In the Netherlands, already over 140 platforms actively mediate between individuals. Every platform’s promise: Carefree pleasure. Trust is their key.
Trust through online platforms is created in two different ways:
- Interpersonal Trust: The trust among users. This is created by personal profiles filled with information about yourself and through reviews of others about you. When you rent a room through Airbnb, you will evaluate the host and he evaluates you afterwards. The more transactions, the better the view on the trustworthiness of users;
- Institutional Trust: The trust users have in the system. Platform builders do all they can to ensure problems are avoided and solved as soon as possible. Preventive measures are, for example, profile scans and automatic processes like credit checks to keep people with wrong intentions out. Reactive measures are a valid insurance or a well-reachable customer service. These together create trust in the brand of the platform.
It has to do with both trust in the platform, as well as trust in other users. Valuation systems (e.g. reviews, stars) serve several purposes: they cause bad apples to be quickly removed from the system, they ensure that people will be chosen according to their proven quality, and they enforce certain desired behavior. At the time you misbehave (or you do not behave as per the applicable or desired standard), you’ll be slowly excommunicated and you’ll end up at the sideline.
Online evaluation in practice
How does online evaluation practically work? Most evaluations are based on a 5 star system. Besides this, most platforms offer a possibility to add a certain explanation to your evaluation in a text box.
When you look at the average number of stars in Uber, it appears that the differences between good, medium, and bad aren’t that big at all:
The impact of only one low grade can be big in case you don’t have many evaluations yet. And one low grade lowers the chance that someone else will choose you the next time. Is this current system the ideal?
Let’s take a ride in an Uber taxi as an example. That particular day, you woke up at the wrong side of the bed, the taxi ends up in heavy traffic and you miss out on an important business meeting. Chances are that you will not give this driver a great evaluation. In this example the context is independent of the evaluation.
Another example: A carpenter offers his service on Werkspot.nl. In his first year, as a beginning carpenter, he does not have the same experience as an old hand in the trade. Over the years he develops himself as an expert, though negative reviews of his first year still weigh in his reputation forever. Such a system doesn’t account for any learning curve.
Reputation and valuation 2.0
A party that, in my opinion, has thought things through really well, is the Dutch Meeting Review: a platform to evaluate event venues. They improved online evaluation in four different ways:
- Linear depreciation of reviews: a review is depreciated over a time of 4 years. After 1 year it’s weighed by a factor of only 75%, after 2 years 50%, 3 years 25% and the ratio becomes 0% after 4 years. This ensures that your mistake won’t be counted against you forever and your most recent performances contribute the most to your overall evaluation;
- Possibility to follow up on a negative review: In the Uber-example, you may imagine that a client calms down and realizes that he has given a too low valuation. At MeetingReview you are allowed to change you evaluation in hindsight.
- The feedback loop: As receiver you are allowed to enter into a conversation with your feedback supplier to find a reasonable solution. So, not only a future client will profit, but also the current user of the system. This feedback loop is based on the good intentions of the evaluated person and the insight that he wants to learn from his mistakes;
- Manual checks: With a scoring system ranging from 1 to 10, all deviating scores below 5 and higher than 8 are being checked manually. This prevents good friends or competition to affect the average score unjustly.
Such a smart valuation system is an exception at the moment. Most platforms still work with the simplest ratings. Though choices that are made, based on this output, may be drastic.
Added value of interpersonal trust in the future
First of all, more and more often algorithms make the first selection of the supply. With Uber, the algorithm makes a match between you and the taxi. Only afterwards you’re able to see the valuation of the driver. Due to the fact that this car is your fastest option, you’re not likely to decide to cancel this ride. Also on other platforms we see automated linking appear. In Airbnb you’re already able to rent many houses by ‘direct booking’, i.e. without the explicit confirmation of the landlord.
What we see happening is a movement from interpersonal to institutional trust. Ultimately, you trust the platform to have all their scans and checks up and running. And, moreover, that the platform will fix things in case it would go wrong. The role of valuation systems will be moved to the background more and more, and will end up as no more than a control mechanism.
Interpersonal trust played a big role when the institutional trust wasn’t yet developed. In 2007, nobody had heard about Airbnb, there was no trust in the brand, and besides people were asked to do something they had never done before. The interpersonal trust through online profiles played an important, though temporary, role. The ofter we use these kind of platforms, the greater the trust in the institutions behind these sites.
Be honest: What was the last time you asked your favorite airline for the online valuation of the pilot of your airplane? Exactly.
Question @RenseC on Twitter: “Doesn’t the fact that eBay reputations are still influential contradict with your theory?”
Yes and no. I think that a certain balance between interpersonal and institutional trust will always remain. The question is about the proportion of each of these, and is determined by several factors:
1. The trust and reputation the platforms has developed. Included are elements as trust in the brand, position compared to competition, how screening of users works, if quality checks are in place, what measures are taken when things go wrong, and which securities (e.g. insurances) the platform offers. For example, in case you know that Airbnb excludes every one with a score under 90% from their platform and profiles are filled well, and these elements are evaluated and checked, the necessity to look beyond to the reviews is smaller;
2. The core of the product or the service. Is the product or service standardized? Take a taxi ride through Uber: there are two certain quality standards that provide security (car type, certificate of good conduct, etc.). Besides, the desired behavior (response time, accepted score, etc.) is smartly directed through de app. Therefore it doesn’t matter if driver A or driver B provides you the ride. You want to go from A to B and that’s your reason for choosing the (brand) promises of the platform. Interesting note on the side: maybe the reputation of the demand side will be more important in the future: I’ve seen several Uber forum discussions about the question if you should pick up a client with a score of 4.3, cause it is guaranteed trouble;
3. The transaction taking place with physical contact or not; With a physical meeting (especially outside the personal living space) you’ve got more input on trust and the other party has a harder time not keeping his promises;
4. If the transaction is about a regular thing (taxi ride) or a once in a lifetime transaction (purchase of some product on eBay);
5. If an algorithm provides the match. In other words: are you able to make a choice based on the full supply, or does a algorithm provide you a match and does the profile only serve you to possibly refuse the choice made for you;
6. Choice: if there is (like traveling with BlaBlaCar from Utrecht to Brussels about noon tomorrow) not much to choose from, you’ll be less picky;
7. Urgency: do you need something fast, or may it wait for a bit;
8. What are you about to lose, when something goes wrong?
9. Probably, I’ll think of some more reasons later on 😉
Conclusion: I think the proportion of interpersonal <> institutional differs per platform and per service/product. I’m of the opinion that a strong institutional trust, trust in / familiarity with the service, small uncertainty, automated or preselected match and the urgency of the matter influence the impact and lower the requested the interpersonal trust .
To concretely answer Rense’s question regarding eBay: At eBay you may trust the institution in the way they handle their processes, but there are many uncertainties –e.g. not being able to check the quality of the product or not know the product to well yourself. With anUber ride things are organized much tidier, for their focus is on only 1 thing: a car driving from A to B while providing a good user experience. Actually, I’m curious if there is a difference in results of people that have picked up their products personally or when it had been shipped.