Core Tenets 1: Trust Your Users
Earlier, I have talked about patterns and anti-patterns for problem solving.
But before we look at the specific patterns, there is another layer to effectively solve problems. And that is your design principles – what are the fundamental tenets of your business and your org? What are the basic principles underlying your approach to solving problems?
Often different parts of the org are running in different directions, because they don’t have a common set of principles they agree on.
These principles don’t have to be necessarily abstract. They can be specific to the business as well. But they can guide in decision making and finding the right solutions.
Amazon’s leadership principles are one set of examples. Externally, they can sound generic and vague. But within the firm, they have been converted to institutional mechanisms and practices. The famous Amazon memos often begin with a list of the core tenets that the team believes in, relevant to the problem being discussed.
One such principle – that has helped me solve large problems, and avoid making mistakes, is
Trust your users.
In general, people are honest.
Our customers are honest. Our suppliers are honest. Our partners are honest. Our employees are honest.
When we’re looking to solve any business problem, start with this assumption. And stay true to it.
It’s not that no one will ever cheat. There will always be people who try to game the system, but that will be at the margins. It will usually not be mainstream.
What we typically end up doing is to give inordinate importance to the margins, and create bad solutions because we’re trying too hard to prevent misuse. So we end up creating solutions that don’t really solve the real problem.
Sometimes, we might end up creating layers upon layers of complex solutions because our hypothesis about root causes are coloured. This creates more friction for everyone involved, and ends up making things worse, rather than better.
So, design your solutions assuming people are honest. Keep things simple. Design for the majority, not for the minority.
But then, how do you take care of misuse?
- Make sure you can measure and discover misuse when it happens
- Make sure you have mitigation steps you can take.
- If the misuse is minimal, then there’s no need to do anything about it.
Essentially, build a layer on top of the core experience that can check for and intervene when misuse is occurring. Don’t put constraints on the basic experience.
A few examples from my own experience:
DVD returns at Madhouse
At Madhouse, we used to provide subscription based movie rentals via DVDs. Consumers could keep a DVD till they were ready to order the next one. One of our biggest challenges was reverse logistics. We needed to pick the previous DVDs at the same time, or before, delivering the next one. But there were no readymade solutions available.
One of the solutions we proposed was to allow customers to return DVDs via Skypak dropboxes. (Skypak had a large network of small drop boxes, primarily used for dropping in your checks for bill payments. They would collect from this network every day.) Customers only needed to tell us they had dropped a movie in the box, and we would immediately ship the next one to them.
There were worries around this – customers could claim that they had dropped DVDs in a box, without actually doing it. Could lead to a lot of losses. We almost didn’t even try the solution.
But we went ahead. We relied on our fundamental belief that our customers were honest.
However, we did make sure we had the ability to measure and track the returns via Skypak. And we looked at them very closely.
Recommended by LinkedIn
As it turned out, over a few thousand instances of returns, we had only one (Yes, one!) instance of a set of DVDs missing for a customer. We politely informed the customer that this had happened. We were ready to switch the option off for him in case this happened again.
But at an aggregate level, this was a success, and a clean, simple approach to solving the reverse logistics problem.
Refunds for product Returns at Snapdeal
At Snapdeal, when customers returned items, we used to process refunds only after the items reached our warehouse and we had inspected them. At some point, we decided to trust the users and issue refunds as soon as the product was picked up. As it turned out, our refund rates remained the same. Customers were a lot happier and we saved a lot of customer support costs.
MMT Assured Hotels
At MakeMyTrip, when we proposed the idea of MMT Assured Hotels guarantee, there were worries about fraud and misuse. Folks were jittery about an open ended satisfaction guarantee.
In this case, we couldn’t really experiment beyond a point – because this was going to be a major brand campaign. Once we went live with the campaign, we couldn’t shut the program down. Because of the risk, we needed, at launch, a mechanism to control misuse. But without impacting the core experience.
So we thought about building a solution that allowed us to prevent (or at least, reduce) misuse. The “we will fix it in 60 minutes, or your stay is free” gave us that power. Most potential problems a guest would have at a hotel are fixable in 60 minutes. (The simplest solution often is just changing the room, for instance.)
We went with this construct, and the results were striking. There were very few complaints to begin with (even though we had made it super easy to raise a claim, and published that front and center.).
And there were hardly any where we were not able to fix the issue to the customer’s satisfaction. Our trust in customer behavior at scale had turned out to be well founded.
Cancellations and Fraud at Ola
At Ola, a commonly held belief was that drivers were trying to game the system all the time, and therefore there was a lot of focus on monitoring/policing. And behavior control through penalties.
High cancellation rates, low ride acceptance rates, and fraudulent behavior were all considered to be due to bad behavior by drivers.
When we brought in this fundamental tenet of trusting the user, and approached these problems with that firmly in place, we started finding fundamental issues that we’d never have found otherwise. Simply by trying to find legitimate reasons from the driver’s point of view.
A few examples of root causes we discovered:
- In cancellations, there was behavior that was legitimate behavior- we’d just not built our product to take that into account. (One example: Drivers wouldn’t log-out of the app during the day, and so, when they had to take short breaks during the day, rides they got during these breaks would get cancelled.)
- Another common cancellation reason (cancelling after asking for the destination) turned out to be linked to low/negative earnings on certain routes. That needed to be fixed through the commercial model itself. It wasn't bad behavior - it was bad economics.
- We used to believe that for all cancellations, either the customer or the driver is at fault. Once we started questioning this, we were able to find scenarios where the Ola system or the traffic environment could be the cause instead.
- Low ride acceptance – it turned out that in quite a few cases, the different tech systems were integrated incorrectly. Some ride requests sent to drivers got lost due to network issues. But the system interpreted these as denials by drivers and penalized them. Drivers who needed our help the most were getting penalised the most!
Another example from Ola was our fraud control system. At one point, the system was penalizing a fifth of our drivers as committing fraud every month. To make matters worse, we were too afraid to actually penalize this behaviour, because we were afraid of losing supply. (All we used to do was to reverse the fraudulent earning.) Which also meant that the system was not able to reduce fraud- it stayed at the same level month after month.
With our trust the user tenet, this made no sense at all.
There were other signals that should have alerted us. But we didn’t see them, because we were wearing the ‘drivers are out to game the system’ glasses. Once we removed the glasses, the problems became obvious.
Once we applied our tenet – we dramatically reduced the scenarios in which penalties would apply. (We also introduced actual penalties in the highest confidence cases.) If drivers called the support team to question a penalty, we now assumed that they were genuinely feeling wronged. And we started telling them why it had been applied - so that they had a way to refute it. (Earlier, we would not tell them any details at all).
Within a few weeks, the fraud detection rate went down dramatically – the new system was having a big impact. CSAT on fraud related support calls went up, and the feedback from drivers helped improve the system.
A few footnotes
There are similar business tenets that have always existed. Like ‘the customer is always right’. I don’t know what’s the origin of that one, but it’s mostly come to mean something else. Rather than it meaning that the customer is actually right, it has come to mean something a bit different. That we should give the customer what they want, because else they will create more trouble than it is worth. It’s a cynical approach, which doesn’t help you solve any fundamental problems for the future. In addition, it’s focused more on individual transactions than on fundamental problem solving.
Gospel Sharing Enthusiast | Social Media & Brand Content Strategy & Creation 📱| Branding | Communications | Leadership | Digital Storytelling
2yHi Mr. Agarwal, This is Tanya Chandran. I wanted to let you know about the latest event and wanted to check if you as a company would like to participate in the Branding of your company, through sponsorship and it comes along with extraordinary benefits, that will help your company immensely. I've attached the pamplet (on my profile) for the event which will be conducted in August 2022. We will be ready to present the PPT to the sponsors with all sponsorship benefits, explaining in depth. Kindly let me know at the earliest. Thanks! Regards, Tanya Chandran Email: contentbusinessacc@gmail.com Contact: +919820176002 Mumbai, India.
Chief Technology Officer at Leadingdots Solutions Pvt Ltd
2yDefinitely ankur..u said it like u do and in such an impactful way. I faced this a lot when checking consumer issues/bugs..the general feeling is customer is raising a fuss because they don't want to pay, or simply they aren't smart enough like us techies...but yes, when I started researching them, found out genuine issues which could be easily tackled resulting in much greater satisfaction of users.
GM & Country Head @ Computomic | Databricks | Data Engineering
2yAwesome read.. great insights and case studies!