Building Fynd — A look-back at how our journey evolved

3 years of learnings and unlearning

This month Fynd completes 3 years. On October 20, 2015 we released the first version of our App and started operations of fulfilling orders from fashion stores in Mumbai. We have come a long way since then. This blog is not about what all we have achieved, rather how we achieved it.

Every month we have a State of the Firm, in which the founders talk about the progress made and answer questions that the team has. To write this blog I went through all the decks that we presented over the last 3 years. I was curious to learn if there was a pattern or method in how we were building and evolving the company. I am not sure how the evolution happened but across the decks, we kept talking consistently about 3 themes — Product & Data Centricity, Constant Innovation for Relentless Growth, and Culture — Fynd’s Bedrock.

What were the pivotal things we did in these themes? Read on!

Product & Data Centricity

There are many ways to build a company — sales-centric, operations-centric, marketing-centric or product-centric. We chose Product & Data Centric simply because this is most effective way to build a non-linear growth company. By removing the linear growth constraint of team-size from the equation, we are able to design an exponential organization.

To understand this in more detail, please refer to  Exponential Organisations by Salim Ismail

Data Democratisation, SQL Literacy

Very early in the journey, we realized that if we want our teams to make faster decisions, we will have to make everyone experts on all our internal data systems and SQL. The classic approach of solving this is by having a dedicated data insights team. But this approach only solves for the short term, and eventually this team becomes a choke point in the organization’s decision-making process. Internally, we have a joke that people at Fynd speak English, Hindi and SQL.

To make everyone’s lives easy, the data insights team stream all our daily data into BigQuery which allows the larger team to ask deeper questions as they work across data systems. For visualizations we use BIME and Metabase, live dashboards are on Geckoboard.

Internal Products

From Day 0 we built our products to be both consumer-facing and internal. Our internal product stack touches all aspects of our processes for smooth functioning. The internal products further evolved when we started applying machine learning to automate key processes in content cataloging.

How we Automated Content Cataloging using Deep Learning

Our teams also took the initiative to start building internal tools to ease everyone’s work — help people access data faster and self-automate their work. Workbench by Fynd was born this way.

Introducing Developer-less Data Workbench — Making business analysts, Masters of the data!

Performance Engineering

We had very little experience in building an internet-scale business. This led us to be paranoid about ensuring all our systems were scale ready and well-instrumented. The attention span of consumers is very short and a slow-loading page is a sure way to kill your business. The teams were pushed to ensure everything loads fast and users have a great shopping experience.

Of course, all of this didn’t happen overnight and we had our share of mistakes several times. Eventually, we got our act together when we rebuilt our infrastructure to be auto-scalable and started following stricter reliability standards.

The aim for the team now is to Plan for 1000x → Build for 100x → Deploy for 10x. You never know when scale hits you!

How we rebuilt Fynd’s Infrastructure

Another thing we did was to ensure that the teams had access to every instrumentation, reliability, alerting and logging tool required. We use New Relic, PagerDuty, Logz, Fabric, Sentry, Statuspage, Segment, and Prometheus + Grafana. We ensured the power-use of all tools and integrated them into each other + Slack.

Constant Innovation for Relentless Growth

Our ambitions far exceed our resources

We have had a fairly difficult fundraising history, but that has never stopped us from thinking big and then executing relentlessly. Where is the entrepreneurial courage when you have billions at your disposal? Our limited resources have ensured we are constantly thinking by first principles, listening attentively to the market and our customers, constantly figuring out more effective ways of growth. The hard times have ensured that no challenge looks impossible today, and the team will rise to the occasion one way or another.

Growth ≠ Marketing

Once the App was launched, we hired a small team to kick-start our growth. On paper, the team looked great as they checked the classic marketer skill-set. Traditionally, marketing is a spend allocation problem — given X amount of money how do we get Y amount of a particular metric in the most cost-effective manner. The really tough problem occurs when X = 0.

Initially, we really struggled with building this team but after a few iterations, we now have a kick-ass growth team that is a hybrid of engineers and growth generalists. The team was built to be extremely data-driven, think from first principles about consumer behavior, and to be open to failure.

We did eventually spend money across all mediums and channels to understand classic marketing better. The eventual conclusion was — if we had to grow, classic marketing was not for us.

7 Things We Did to Scale to 5 Million Users

Growth Engineering

For any internet company, one of the toughest tasks is to reach millions of customers with a limited (or sometimes no) budget. After acquiring our first 100k sign-ups we realized, that to scale to millions we will have to engineer user-driven growth. We built features in the App and custom built the growth stack to engineer our way to over 15+ million users today.

How we grew 10x: From 100K to 1 Million in 6 weeks

We have bigger things planned for next year as we embark on the next 10x growth phase — stay tuned.

Platform Evolution

We operate in a hyper-competitive space. For us to make a disproportionate dent we had to think beyond the classic channels of Apps and Web. By listening closely to our Brand Partners, we launched our in-store product Fynd Store and Brand Sites via Open API. By evolving into a platform, we started collaborating with internet giants to enable newer store inventory use cases. In hindsight this seems very logical, but it isn’t priority when you have limited engineering resources.

We are super excited with the possibilities and our initial collaborations are showing great promise.

Open Sharing our Pitch Deck for our investment from Google

Culture — The Bedrock

Great culture builds great companies

Culture is very loosely understood and mostly felt. It’s lazy to define culture simply as a set of common values and shared purpose, because it’s much more than that. Rarely do we on an ordinary day, read the core values. The team finds its own natural rhythm and people collectively figure out the culture. It’s quite difficult for me to explain how the Fynd culture evolved but we are super proud of how it has shaped up.

Flat, Honest, Transparent

As founders, the onus of culture — good or bad lies with us and it flows from the top. One thing we did from Day 0 was have a very flat organisation. For the longest time we didn’t want to have any management. Any policy or process that involved us taking a ‘management decision’ was actively discouraged because we didn't want to waste our precious time. We trusted our teams to do the right thing.

To ensure continuous communication between the front-line and the team leaders we held monthly State of the Firm sessions. We circulated an anonymous questionnaire to ensure every voice was heard and every concern addressed. We were brutally honest about how things were shaping up and ensured we were never in denial of any issue — big or small.

Unlearning: Rot at the Top

One of biggest failures in this journey has been not spotting individuals who were destroying the company especially those whom we had entrusted leadership roles. This is such a messy problem to discover — you trust your leaders to deliver and you also trust your team to tell you when things are not working. Unfortunately teams find it almost impossible to bring this up to you. We introduced a core value Brutal Honesty especially to address such situations.

What does one do when rot is discovered? We move extremely fast to remove it. This takes a massive toll on the team’s rhythm and confidence. How does one bring this back? Quite simply be brutally honest about the situation, one’s failures, and how we will have to do better.

Continuous Improvement: Learning Organisation

Ordinary people achieve extraordinary things

We firmly believe in this. Organisations need to improve continuously and that happens only when the individual is improving, who then improves the team. This is a hands-on process and needs to be done religiously.

At Fynd, we have a dedicated group called Open Learning whose task is to teach the team anything from English speaking to building ML models. We recently implemented OKRs to ensure teams are inter-aligned, and strongly aligned with company goals. We conduct a leadership feedback survey twice a year to ensure everyone is accountable.

Another fun thing we did was ‘Workcations’. Teams would travel to some distant location to work during the day and party the night away. The teams have workcationed in Singapore, Malaysia, Thailand, Turkey among many other locations within and outside India.

Can’t Take a Vacation? Try a Workcation

Open — Do, Learn, Share

One of our challenges when we were building Fynd, was the complete lack of material on how to build internet companies in India. Sure there is a lot of material about building US internet companies, but none of that translates to us — different market, different people, different times.

Our first blog was written to share with a fellow entrepreneur, to explain how we setup internal communication and collaboration. We quickly realised what we were doing must be shared. Let people learn from us — good, bad or ugly. We never shied away from writing on unpopular topics.

On the engineering side, we shared our code repos, models, and data streams on gofynd.io

Introducing gofynd.io — A slice of Engineering at Fynd

For our Brand partners, we started Fynd Voices for them to share their vast knowledge and perspective.

Launching Fynd Voices — Webinar Series

So much more can be achieved by simply sharing. It’s a criminal waste of our startup ecosystem’s resources if everyone kept committing the same mistakes over and over.

State of the Firm — Archive

Continuing our open sharing initiative, today we are opening all our past 20 State of the Firm decks from April 2016 to September 2018 (~500 slides). The decks are shared as-is with limited redaction in a few confidential slides, links and numbers. As you flip through the deck you will understand how we evolved and got more confident over time.

Book Recommendation — All-time favourites

Right now, somewhere out there in the world, is a paragraph, a chapter, or a book that could change your life forever

The above 3 books have had enormous influence on how we are building/evolving Fynd.

If you are building for long and scale, I cannot stress enough on how important it is to first and foremost build your org/team/leadership mental model. One builds only what one knows. If you want to build a rocket ship, you first need to know how rocket ships are built. Without knowing you will be shooting in the dark, hoping something sticks. Please read these books and then read them again!

Thank you

Our journey has just began. We would like to thank everyone who has helped us reach here

• Our Customers — who tried us out, who were patient with us when we goofed up. You are our â„–1 priority and we are working to ensure we don’t let you down
• Our Brand Partners — who gave us a chance and trusted us to represent them publicly. Thank you for your continued support
• Our Service Providers — Special mention to Fuzail Qureshi and Dipti Jena of Netcore
• Our Investors — who believed in us even when sometimes things were down. Sasha Mirchandani and Vidushi Kamani of Kae Capital — our oldest champions. Sailesh Ramakrishnan, Anand Rajaraman and Venky Harinarayan of Rocketship.vc who bet on us without even meeting us. Prashasta Seth and Purvi Parkeria of IIFL. Surojit Chatterjee, Swati Trehan, Shilpa Anand, Amit Jain, Seema Rao of Google and everyone else (it’s quite a long list)
• Our Advisors and Friends — MG Parameswaran, Vikram Limaye, Gagan Goyal, Zishaan Hayath, Deval Shah, Kunal Bahl & Rohit Bansal, Aneesh Reddy, Anand Chandrasekaran
• The Ecosystem — Fellow Entrepreneurs especially the Bombay Hackers & Painters WhatsApp gang. Tech blogs who covered us early on, in particular Inc42 Media
• Our Team and Alums — you make us proud!
• Our Families — we couldn’t have done this without your enormous patience, belief and support
• Lady Luck — Keep smiling on us

The Founders.

#HappyFynding