A few months back Sreeraman wrote about how we at Fynd scaled from 10K downloads to 100K in four weeks. When we achieved that milestone, we had amassed a lot of understanding about what works with our customers. After the Festival of Fashion, we spent the next three months on our product to improve the user experience & product discoverability, migrating the entire backend infrastructure to AWS Mumbai, and streamlining our operations to service over 20K+ pin codes across the country.
Our next goal was to scale this 10x again. A very ambitious target for a six-member growth team!
Not only did we scale it 10x, but we also outranked our biggest competition on a shoe string budget!
What Did We Achieve?
- One million customers on the Fynd App
- 200K visitors on Fynd web in 20 days
- 100K+ conversations on Fify
- Top 10 ranks on App Store & Playstore Shopping Lists
- ~90% reduction in Ad spends
How did we achieve this?
No needle movement happens by itself. What looks like overnight growth was meticulously planned several months in advance. The growth team was in lock-down mode for the entire sale period. There is no sure shot way to growth and we had to put some method to this madness. It was all about identifying opportunities that can be scaled massively against hacks that may give us uplift in metrics.
Everyone has a view, trust only data.
Plan, Execute & Repeat
We used to start our day with a team huddle to plan strategies for the day, reflect on the experiments from the previous day and course correct, if necessary. This is the most important part of our process.
NEVER do incentivised installs. This tactic will jack up your downloads and ranks but the users acquired from such channels do not bring any value. We focussed our efforts on a referral programme. The campaign was designed such that you and your friend could both earn INR 200 giving us a much-neededed boost in installs. These clubbed together with organic installs accounted for over 70% of our total installs.
Additionally, identify channels that can give you a positive ROI and kill the ones that do not. For us these were Facebook & Google Ads. For example — we optimised our spends based on conversions from each sub-campaign.
Next, focus on the campaigns within these channels that give you higher retention. Monitor this on a weekly basis and adjust your ad spends accordingly.
Example — A key takeaway from the sample dashboard below could be to increase spends on the campaigns where seven-day retention is higher than 40%. This differs from product to product for instance for a music app — you’d want that number to be somewhere around 50–55%.
Learnings from Facebook Ad Optimisations
- Running the ad sets with decent ad relevance score for longer duration and more budget than the low performing ones.
- Keeping the campaign communication simple and to the point (no fancy words or creatives).
- Creating the right target audiences through ‘Audience Insights’ tool.
- Creating separate campaigns with different creatives for Men and Women (ad-copy can remain the same).
- Optimising ad bids further basis the gender.
- Creating a sense of urgency (at the same time also avoid forceful CTAs like BUY NOW, HURRY etc.) in all our communications.
- Building our Custom Growth Stack
From a product perspective, thankfully, we made some really wise decisions early on in our product development lifecycle. The wisest of all was to build our growth stack in-house from day zero! Primarily because we’ve had a bitter-sweet relationship with third party analytics/marketing automation tools as most of them just fall out of place at scale and/or do not integrate well with each other and you’re left with not having a single unified view of your customer.
A lot of folks may disagree but to put this into perspective — we have captured over 100M customer data points in the last 4 weeks alone and, at that scale, the requirements can get really complex. However, this does not mean that we built everything from scratch. We just chose the right tools that integrate well with the one’s we build. We use a combination of tools like Branch, Segment, BIME etc.
Importance of App Store Optimisation
All of us will admit that app discoverability is hard! Pay attention on this early on and make sure people see you for relevant search terms. Do this to improve your organic downloads over a long run.
With the growth projections at 1 Mn users, the platform team was constantly asking itself the question — How do we make the app snappy enough for our users? An average user browses our main screens (Feed, Brand, Category, Collections), dives into few products, adds them to cart and makes an order. We wanted to make this entire process seem lightning fast. This would mean bringing our API response times to < 40 ms.
In this process, we also realised that a major section of our latency was network delay and it stemmed from the fact that our front end servers were located in AWS US East Coast while our users were in India. Much to our advantage AWS launched its services in Mumbai around the same time and we migrated all our front-end servers to Mumbai. We then started experimenting with various API optimisations to get the best performance from our servers. The end result was response times of less than 40 ms on one of our most busiest days ever.
We are the first ever marketplace in India to host a 45 day long sale that mirrored the in-store offerings of all our 200+ brands across the nation.
In doing so, we provided every brand an opportunity to further capitalise on their already existing partnership with Fynd by ensuring sell-through of competitively priced merchandise.
While we focussed on customer activations across all possible channels months in advance, which is essentially creating demand on Fynd — a lot of prior work went into aligning brands on the platform to ensure sufficient supply and availability of the best-priced products for every kind of shopper.
Two months prior to our upcoming ‘The Grand Brand Sale’, every brand was contacted and apprised of all our efforts. Moreover, the brand growth team worked alongside every point of contact to ensure we had all their catalogues, flawless real-time inventory/discounts, brand assets and special offers active during the term of our sale.
The next milestone for us is 10 million customers and we plan to achieve this by February 2017.