purpose driven growth | product-market resonance | brituals | leading from within


Role of Hidden Assumptions in Start-up failures 

Case Study – Mango DVM (2007-2011)

Ramkumar R S


Mango DVM was my own company. We were two founders. Deepak Ramesh was the CTO and was responsible for technology (hardware and software and servers) and delivered flawlessly on these aspects. I was the CEO and responsible for Marketing, Sales, Content Licensing and Aggregation. The Angel investors supported us with the money. All mistakes were entirely mine.

Role of Hidden Assumptions in the Mango DVM Story.

There are many dimensions to the Mango DVM story. There are many learnings from its failure. 

I did not pause to ask critical questions like “Who is our customer?”, “What business are we in?” and “What unfullfilled emotional need are we fulfilling?”. 

Even if I had asked these questions, my answers would have been different at different points in my journey.

There are many other mistakes I did as a promoter, during the course of the journey, which I will probably analyze and share in subsequent writings. 

In this piece I want to focus only on the role of “Hidden Assumptions”.

It is important to remember here that these Hidden Assumptions became visible to me, only after they were proved wrong. If only I had uncovered these hidden assumptions much earlier, I could have taken mitigating actions to steer the business away from its eventual failure.

The Original Idea: Music Vending Machines

It was 2006, in India. Long before 2G, 3G, 4G, Broadband, Fiber-net and Smart phones. Audio Cassettes were on their way out and CDs have been around for a while in India. Apple iPOD (launched in 2001) was making waves in the US, and the first iPhone was one year away.

Mango DVM was a Finalist in the Intel India Business Plan Competition. We got some of the best mentors assigned to us. They helped us concretize our idea and formulate a solid business plan and go to market. Without this help, we would not have been able to raise angel funds in 2007.

We did not win the finals at Intel but we went ahead and incorporated Mango DVM in 2007 June. We started the product development with our own money. 

The product: (2007)

Wall mounted touch screen based Music Vending Machine that occupied 18” X 36” of wall space. It had a local storage capacity (on encrypted hard disk) for 200,000 songs. It had a multi-lingual meta database and multi-lingual keyboard for searching.

It was designed to write music on a blank CD,  or write music files on to Pen drives and memory cards through a USB port. Latest songs released every day could be updated to the DVMs in the market using a background process from a remote server. Every transaction was authorized and recorded in a remote server using a Reliance Modem.

Why Mango in Mango DVM?

My long-term vision was to build a brand that connects emotionally with consumers. I thought giving them easy access to music of their choice is a great place to start. I said “What Apple is to America, Mango will be for India”

My plan was to get to at-least a 1000 DVMs in each state. Generate Revenues and Profits. Raise more funds. And then parallelly extend the brand to other modes of delivery such as home internet and mobile internet. At that time the cost per GB of data in India was prohibitively expensive. I was hoping that I will get enough traction in the DVM distribution model and become an established brand with accrued profits, by the time the market is ready for other modes of delivery.


In September 2007 Mango DVM raised Rs 62 Lacs Angel Funding from a group of 8 Angel Investors. They were some of the best minds in the Chennai Entrepreneurial eco system.

My business model was to sell the DVM to shop keepers for Rs 30,000 and also give them a 25% revenue share on each song sold. The royalty share to the music company was 50%. The  gross margin left was 25%. The numbers and financial model looked viable and scalable.

Over the next 6 months (While our product was getting ready), I successfully convinced 100s of music labels including the Big Boys like Sony and Saregama, to license their content on a revenue sharing basis, without any upfront payment.

We aggregated and digitized 50,000 songs in 6 months and another 100,000 in the following 6 months.

The Launch

In April 2008, exactly 6 months after raising the first round, we deployed 5 Music Vending Machines in 5 retail Locations across Chennai. 

Immediately after, Mango DVM raised another follow-up round of Rs 40 Lacs.

The first month sales from all the 5 machines put together was less than Rs 1000. 

Over the next 3 months, this did not improve much. I stood there in each shop for hours together, cajoling customers to try out our product and buy some music.

Odyssey Chain of Book Stores and Pyramid Group that managed over 700 movie theatres (two different customers) were negotiating with us for deploying Mango DVMs across their shops / theatres. But they did not go ahead with their decision, because of the poor sales record from the existing DVMs that Mango DVM already had.

Hidden Assumption 1, that became visible after it was proved wrong.

Consumers will be willing to stand before a self-service touch screen machine, to search and select and download songs from a database of 100,000 songs.

Realization 1:

We are not in the business of “Self-Service Music Vending Machines”. We are in the business of distributing and selling music. Technology is just an enabler. We have to adapt to what customers are comfortable with.

Pivot 1

By this time, I knew there were 1000s of shops in each state, that were recording music in cassettes and CDs. Customers were indeed spending Rs 100 to Rs 200 to get a collection of songs of their choice. 100% of these shops were illegal. Because till then, there was no licensing mechanism for such shops.

So, I abandoned the idea of Self – Service Touch Screen DVM. We converted our software and music database in to a package that can be deployed on any existing PC. Encryption and DRM were used to ensure that only licensed dealers can use our software and database to sell music.

I decided to sign up dealers who are already in the business of music recording. Provide them, with a large inventory of legally licensed music and an easy to use software. Our software was so user friendly and multi-lingual, even small-town shop keepers could search, select and record music, even if they did not know any English.

I personally went and talked to many potential dealers. But after 3 months of effort, I had signed up just 2 dealers.

Hidden Assumption 2 that became visible after it was proved wrong

The no. of customers in each town, who are willing to pay Rs 120 for a set of 10 songs will be large enough to build a scalable business.

Realization 2: - Price per song is a major concern 

At this point in time, I came across an opportunity from Samsung, who were willing to deploy our software platform in over 1000+ mobile phone show rooms (Exclusive Samsung Showrooms) to deliver music to their mobile customers. However, they refused to move ahead, because our price per song was very high.

I did some hard negotiations with all the music labels and asked them to lower the price per song. They partially agreed for a 30% reduction. They also agreed that I could sell at any price I want, but I have to pay a royalty of Rs 4 per song.

To prove my point, one fine day, I dropped the price per song from Rs 12 to Rs 4 per song. I decided to pay the entire amount as royalty. (zero margin). I had some money left in the bank and I decided to subsidize the dealer commission of Rs 1 per song from the investor’s funds.

After this price revision, we gained some very good traction. I hired a sales guy to travel all over Tamilnadu. We quickly signed up close to 50 dealers / shops many of whom were from Tier 2 Tier 3 towns. Dealers even paid a signup fee of Rs 15000 to set up the software and database on their existing PC.

At this time, I reached out to a senior person of a popular venture fund and he said, “come back to us once you reach 100 shops”

The dealer experience with the software was great. There were stories of customers dancing in joy after finding and recording songs of their choice (we had latest releases, old songs, classical music and devotional songs) They felt Rs 40 for 10 songs was reasonable.

But the average download volume per shop was only 1000 songs per month. (Approx 3 customers per day)


The biggest competitor for our Dealers, were other shops. (who were still illegal) Their price point was Rs 1 per song. And they easily sold close to 10,000 songs per month. And 100% of that revenue belonged to the shop keeper. 

Whereas for Mango DVM dealers, at Rs 4 per song and a dealer commission of Rs 1 per song, they just made Rs 1000 per month.

After a few months, many of our dealers started becoming in-active. They were actually switching back to the illegal mode using another PC. Or formatted our hard disk and went back to their old ways. Our active dealers went down from a peak of 50 dealers to 15 dealers within a span of 1 year.

I was already burning Rs 1 per song sold. And I did not have the money to reduce the price further and compete with illegal shops. 

Pivot 2

After a lot of discussions and negotiations with the music labels and their associations, I got the permission to pay a fixed royalty of Rs 25000 per year per shop and the permission to sell unlimited no of downloads at any price.

I was very excited about this plan.

The music industry at that time, estimated that around 50,000 illegal shops existed in each state. 

They hoped that even if 10% of them convert to become legally licensed shops, the license revenue will be substantial. They were so excited about this model, that they did not want to wait for Mango DVM to scale. So, through their associations, they deployed their own sales teams to sell their yearly license with a certificate. And since we were operational only in Tamilnadu, they recommended Mango DVM Software and Song Database to their licensees in Tamilnadu, even though it was not compulsory for the licensee to use our software.

We started selling our software and database for Rs 5000 per year.  The license fee of Rs 25000 per year was in addition to that.

The no. of Mango DVM dealers started going up once again. Now the dealer was free to charge whatever he deemed fit. He could sell 100 songs for Rs 50 (for a pre-bundled package) and charge extra if his services and time are required for customized bundles. We allowed them to create and save custom bundles which they could sell with less effort.

We also helped the dealers by printing catalogues of pre-prepared bundles (Example : Hits of Ilayaraja). Once a customer chooses the catalogue, they could download instantly, by just entering the catalogue code.

In order to encourage faster adoption, we started offering a monthly payment scheme of Rs 2500 per month instead of an upfront payment of Rs 30,000 per year. I hoped that this scheme will rope in at-least 1000 shops in Tamilnadu, (2% of the estimated 50,000 illegal shops). If we get there, our gross revenue will be Rs 300 lacs per year with a net margin (after royalties) of Rs 50 lacs per year.

This will allow Mango DVM to raise more funds and expand to other states.

Hidden Assumption 3.1 that became visible after it was proved wrong

Once we have a licensing regime of Rs 25000 per year per shop supported by the entire Indian Music Industry, the government and police authorities will crack down on all the illegal shops. This will speed up the process of enrolling dealers for our plan.

Hidden Assumption 3.2 that became visible after it was proved wrong

Dealers will be willing to share 20-25% of their music download revenue to remain a legally licensed dealer in a market where there are 10 more shops in the same town, who still remain illegal and pocket the entire revenue.

Hidden Assumption 3.3 that became visible after it was proved wrong

In a market road of a town or city, if there are 10 illegal shops doing a business of 10,000 each, and if one of them converts himself in to a legally licensed shop, the business volume should skew around the legal shop (Say Rs 40,000 PM) and the business of illegal shops on the same road would reduce. (Rs 60000 shared by 9 shops). With a fixed fee of Rs 2500 per month and a potential to increase revenue from Rs 10,000 PM (illegal) to Rs 40,000 PM (Legal), more dealers will come forward to become legal.

The Moment of Truth:

Even though the price of the song / service was same, the business volume of legal shops did not increase. They continued to do, the same volume of business that they did while they were illegal. Similarly, there was no reduction in the business volume of illegal shops located on the same road.

Realization 3

Customers don’t care whether the music they are hearing is legal or illegal. They are willing to pay for convenience, variety, access, service and delivery but not for the legal rights or copyrights of those who created it.

Pivot 3  

I was also hearing that the music industry’s own effort to sell their annual license to 1000s of shops in each state wasn’t going as expected. The police and government machinery did not have the bandwidth to support them. Customers were not rewarding legal shops over illegal shops. And the whole scheme was moving towards imminent failure.

The sales team at Mango DVM gave me an interesting pivot option. To sell the Mango DVM software and Music Database for Rs 5000 per year. We can obtain a written declaration from the shop keeper that he will purchase the annual license from the music association directly. The risks involved in not having the license is entirely on the dealer.

The sales team assured me that they can easily scale to 1000 shops in Tamilnadu alone, if we turn a blind eye to whether the dealer really bought the license or not and let the music industry worry about enforcement and collection of their license fee.

During these discussions I also found out, that even some of the existing Mango DVM dealers had forged their license certificate, to show their own name on a certificate bought by some other shop. Their intention was to save on that Rs 25,000. They were more than willing to pay Rs 5000 for the software and song database alone, because they saw value in it.

Legally we might have been right to claim that we are only selling software and the enforcement of the yearly license fee per shop is not our concern. Particularly when the music industry had their own sales team for this purpose. But morally I wasn’t sure whether it was right.

The Shut down

After a long period of dilemma, I decided to shut down operations. We returned money to some of the active dealers whose annual subscription for the software was still alive. 


After Rs 112 lacs of funding over 3 rounds, Rs 15 Lacs of revenue, Half a Million Downloads and 4 years of effort, Mango DVM became a failure.

Primarily because, I failed to foresee and mitigate some of the key hidden assumptions behind the business.