Why we stopped supporting Cash On Delivery and are loving it!
Myth-buster: COD support isn’t necessary for business!
Introduction:
COD (Cash on Delivery) is the lifeline of an e-commerce business. Or wait, at least that’s what the e-commerce giants and analysts claim it to be. In this article, we will take you through our operational cost structure, why we shut down COD and how we have been performing after 4 months of ZERO COD orders. Stay tight!
COD Fueling E-Commerce Growth:
There is absolutely no denying of the fact that COD kind of fueled the e-commerce growth. Back in 2010, when e-commerce was still at a nascent stage for the majority of the country and people had serious trust issues, the introduction of COD as a payment mode by Flipkart significantly helped them increase their conversion rates. Logically thinking, even if COD had a fulfillment rate of (say) 80%, it is still better to have 100 extra orders out of which 80 gets accepted and cash is paid at the time of delivery rather than not getting any of these 100 orders at all.
Operational cost at that time was not a concern and the company was okay with asking their logistics partner to handle the cash at the time of delivery for some extra fee. Since the pockets were deep, cash flow too wasn’t a concern for them. Other e-commerce companies also soon followed the suit. COD soon became a standard (rather expected) service to be provided by all e-commerce players. The only thing which an e-commerce company was judged on was GMV which does not adjust for cancellations, returns nor discounts. So anything to unsettle this top-line metric, even at the expense of significantly worsening bottom-line, was a big no-no.
And yes, at Fynd, we too followed suit. Since its inception in November 2015, Fynd always had COD as one of the payment modes available to its customers. This helped us get our initial customers who heard of us for the first time and were skeptical about placing an order on a new platform. Appreciative of the gesture, we were delivering these orders with full sincerity. Our RTO numbers in 2016 were as low as 0.12%. But then during this 2-year journey from 0 to ~10 Million customers, was a roller-coaster ride with a lot of experience to learn from. We soon noticed some alarming patterns in our COD orders.
What went wrong?
No matter what your business model is,
Scaling isn’t an easy task!
While we were working day and night to scale up the order volumes and efficiently deliver them, we attracted some unwanted fishes to the party. These were major of 2 types:
- Affiliate Networks
- Customers you don’t want
Affiliate Networks
Being a data driven company, we were always looking for ROIs on whatever growth activity we do. Having said that, performance marketing was one of the major pillars of our growth strategy in the early phase of the company. What this means is, we had some major affiliate partners on board who were running performance driven campaigns on our behalf on the inventory available with their publisher network. All the payouts were tied to purchase conversion rates so that we not only spend on digital ads but also derive ROI out of each publisher network.
All sounds good and intelligent. Then what could have possibly gone wrong?
We were working on the (fad) term ROI and optimizing it as well. There could not have been anything wrong in the system. Right?
Well absolutely wrong! We made mistakes at 2 levels in this entire approach.
- Measuring ROI on top line numbers
- Ignoring the impact on operational costs
Measuring ROI on top line numbers
This was our first biggest mistake. While we were looking at the fancy purchase conversion numbers on the third party attribution tools, we were not connecting the data with the post-purchase customer journey. This resulted in the publisher networks exploiting us (and many others with a similar approach).
Their concept was simple — Make somebody download the app (using their link) and place x orders on COD every day and cancel it immediately (or refuse to accept it when it is brought to their doorstep). So, what used to happen is, when a user places an order, it would get attributed to the publisher. Soon after the user will cancel the order which would never get attributed in the third party attribution platform. Conversion rates would become healthy and marketing managers would start dancing.
While our overall order fulfillment rates were ~90%, this network had an 11.44% fulfillment rate. A clear flag that somebody is trying to mess around. And when you confront, the obvious reply is a denial. Have a look:
How It Works: You tell a publisher that you will pay him Rs. 10 per app install only if the conversion rate is 5% or above. The publisher will run an incent campaign (there are 100s of websites/apps doing this), where they will incentivize the users to download your app for Rs. 1 or 2. And then will make one person sit and place 8 orders on your website and cancel them. When you look at the numbers it is an 8% conversion rate, whereas in the bottom line,
YOU MADE NO MONEY!
In fact, you lost double the money on the delivery that was rejected and lost credibility in the eyes of the seller as well.
In any typical e-commerce company, the job of growth team ends at getting an order! Now it is operations job to fulfill the orders and take it from there. But we can’t afford that. When we connected the user acquisition data to the post-purchase journey, we noticed that there was blatant fraud by a couple of publisher networks using the process mentioned above. Surprisingly, one of those networks (the one with 11.44% fulfillment rate) is currently ranked among top 3 publisher networks in India. Their fraud was to a level that one person based in a remote location placed 72 orders in a day, and obviously, all of those either got canceled or RTOed. When we encountered them with post-purchase data, they initially denied any wrong doing but since we had strong data points to prove our case, they kind of agreed to the fraud (terming it as a publisher fraud and not the network fraud) and claimed to have blacklisted the publisher and waived off our marketing fee for all fraud orders.
Since then, all ROI numbers for us and for our all performance channels (including FB, Google, etc.) are measured on bottom line. It doesn’t matter how many orders we get via a channel; how many we deliver is what defines our ROI.
When we asked other e-commerce companies for their numbers, everyone said that these (30–45% RTO) are industry standard numbers and we were shocked at how they can brush this off as BAU.
There was so much money being lost and no one cared as it propped up their GMV, and hence their lofty valuations!
How does this impact cost
As I mentioned, our marketing cost on these publisher networks was waived off after we proved them that it was fraud. It would then mean that we were smart enough to save ourselves a lot of money from this fraud. Right?
Sorry to break it, but we (and you) are wrong again! While we did save ourselves from false bleeding money into wrong marketing channels, we still lose money on operational costs and were continuing to do so.
Here is a simple math of how operational cost works (numbers are indicative in nature):
Assume that forward delivery i.e. delivery to the customer, costs around Rs.100 per order. In case it is a COD order, there is an additional cash collection fee of Rs. 40 levied against the order. This jumps up the net delivery cost of an order to Rs.140 (for the orders which get delivered).
What about the orders which didn’t get delivered? Well, the delivery partner charges Rs.100 and takes it to the customer’s doorstep, but since the customer refused to accept the order they had to bring it back to us which will cost us additional Rs.100. This brings up the total operational cost of non-delivered orders to be Rs.200.
For each RTO order, you earned ZERO and lose Rs.200!
What that means is, more the number of RTO share in the total orders, the more money we lose on operational cost and make absolutely zero money on revenue. It’s a complete loss-loss situation, where, the top line numbers look good, CAC (Customer Acquisition Cost) and Cost Per Sale (CPS) also look good, but at the bottom line, the company would start bleeding because of operational costs.
Now, have a look at our delivery rates of COD vs. Online Payment mode:
The leakage is where we are also spending hefty money in operational cost (as explained above). A failure rate of 95% from any particular payment mode can not be acceptable at all.
It is worse than doing business!
The Customers you don’t Want
In this entire conversation, we have just mentioned bad channels for promotions.
But, are there only bad channels, or there are bad customers too?
Conventional wisdom dictates that every customer is a good customer. But that’s not always the case. Sometimes, there are some customers who try to mess around with your company and for no justified reason.
Though these customers are lesser in number, once they start causing trouble for you, it can turn into a real mess in no time.
Now let us look at the RTO by date pattern for the gentleman on the top:
Now, we have tried reaching out to all of these top defaulters, and believe me, most of them were just “enjoying ” playing with the system. This makes us re-think, whether COD is a service which we should offer to everyone, or should it be an elite service (like credit line) which is offered only to genuine customers.
Anyways, customers gamed the system, and there was no common rule which we could apply to all except for blocking these customers only once they have done “enough fraud” in the system.
Now, let us look at the corrective measures we took to stop both notorious affiliate partners and notorious customers:
Initial Corrective Measures
One obvious corrective measure was to keep a tight check on performance networks and stop the publishers which have a high cancellation rate.
The second corrective measure which we took was to discourage people to opt for COD orders by levying a COD Fee of Rs. 50 against each order. This should discourage people to opt for a COD order and pay upfront via any of the online payment modes. This also had an additional 10% cashback on all online orders (without using any coupon code).
Even though this reduced COD orders by a good 25%, it didn’t curb the problem completely.
We also implemented a third party tool, to detect the probability of an order not getting fulfilled even before shipping it. But who are we kidding, if 95% of the orders are failing how much improvement can a third party tool do anyways.
We had to take a bold decision!
Being Bold: The Ultimate Correction
Until this time, we were studying the data in more detail and trying to identify patterns of COD orders which resulted in failures. We even tried blocking COD according to PIN Codes (which had high failure rates), but the impact of the same in improving the net delivery of COD orders was negligible.
At the same time, we also noticed that our users coming in via Word of Mouth, referral program, Google and Facebook typically opted for an online payment which has a delivery rate of as high as 95%. These are also new customers for us, but they don’t hesitate in paying online.
Since we were not able to identify any other possible reason for COD order failures and we observed that the genuine customers were paying online, we took the final call. In January 2018, we shut down COD completely. We trusted our genuine customers to still transact with us and this would completely shut off the noise of unfulfilled orders which were mere numbers for us.
And yes, in addition to this, we added every payment mode possible to our system, so that none of the customers trying to pay online are left out. We optimized our payment flow so that customers can checkout fast using our “Flashpay” payment service.
In short, we did our homework and prepared well before getting into this decision.
Did it Go well?
Completely shutting down COD definitely impacted our top line numbers. Our conversion rates suddenly collapsed and marketing ROIs shook up. Overall order volumes saw an impact as well but we were well prepared for it. In the end, we were okay doing less business with genuine orders than more business with false orders. This was a conscious call and we held onto it.
Slowly, our COD line diminished and more and more people started accepting the same, in turn switching to the prepaid system. How did this effect the bottom line? Have a look:
You would notice that while the overall order volumes decreased, the blue bars (i.e. order getting fulfilled) kept on increasing. So, the efforts started paying off. We were not only saving Rs.210 per order on orders not getting delivered but also more and more of our customers were getting comfortable in making online payments and hence improving our top line conversion numbers again.
Our leakage numbers which soared up to as high as 65% were now down to a minimal 3.21% which we can bet is the best in the e-commerce industry (any vertical).
So, at the end of the day, we did suffer some initial downfall in the top-line numbers with this decision; but the numbers say it all and it broke the biggest myth in our mind:
COD is a must have for any e-commerce network!
We are now not only having a controlled operation cost but are increasing in top-line month after month. Also, our targets are to bring down this leakage to a 1.5% mark. Yes, that means 98.5 of every 100 orders placed will be successfully delivered to our customers.
Additional Note: But how do we get those customers who still prefer COD?
We constantly have this discussion of yes this worked, but there is still a good number of customers who we are losing just because we don’t have COD payment mode. An answer for this is going to come from the new vertical getting established in e-commerce space, i.e., B2C money lending apps, for ex., Simpl, ePayLater, EarlySalary, etc. These platforms have the infrastructure in place to determine, on the fly, whether a person will be able to pay back the amount or not. A customer, at the time of checkout, can select (say) Simpl as a payment mode and complete the process. They don’t pay money to us immediately and at the same time, we have taken an assurance from a third party for the same. This seems to be the middle ground, though we are yet to conceptually test it out (starting August 2018).
Glossary:
- COD: Cash on Delivery (Cash collected at the time of delivery)
- Leakage: Percentage share of total orders which are not delivered to a customer due to any reason. Reasons primarily being an RTO or cancellation of the order by a customer as soon as he places it
- RTO: Return to Origin (The order is taken to customer doorstep but the customer refuses to accept it and hence is returned to its origin)
- GMV: Gross Merchandise Value i.e. the sum of the selling price of all the products being sold
Special thanks to Farooq Adam and Rajni Kant Sinha who worked on this problem closely. Shutting COD wasn’t an easy decision, but a very hard data driven one!
Do share your experiences/questions with us at growth[at]fynd.com. We would be more than happy to hear from you.
