Launch quickly, inspect closely, iterate rapidly

One of my biggest learnings from our recent “Grand Orange Bag Days (GOBD)” sale has been the about the make-or-break impact quick inspect-iterate cycles can have on growth. If you are interested in stories of how rapid iterations coupled with tight feedback loops helped us get insane growth, you will find it interesting.

I believe that most product launches need to be followed by rapid “inspect <-> iterate” cycles to make sure we learn deeply and maximise impact.

These cycle are best served with a simple and instantly actionable stretch goal.

For most consumer internet products, this should mean multiple iterations everyday till we clearly know the next big problem to solve (or till we give up).

I have a couple of stories to explain what I mean.

Story 1 — A flawed dream

This story starts with a dream we had. We dreamt that our service partners (micro entrepreneurs who do order delivery for us on a pay per delivery basis) will one day partner with us to sell our products in their neighbourhood.

With sale around the corner, someone had the bright idea of driving sign-ups through our fleet. We saw the perfect opportunity to test our hypothesis. I thought — “If these guys can sell our sale, I’ll bet they can pretty much sell anything else as well!”.

But when we release the sign-up referral program for them (they earned Rs. X on each sale sign-up they got), there was very little uptake in sign-ups. It seemed like a failed experiment.

Luckily, we decided to dig deeper. After talking to a few service providers, we realised that the problem was that none of the partners knew about the program. It was a communication problem!

On pointing this out to our last mile team, they doubled-down on their communication efforts. Things immediately started to move!

At this point, we set a stretch goal for this project— “If we can get 50% of all sign-ups in a day from partner referrals, it would mean this network can do much better than any of our existing marketing channels at less than one-tenth the cost!”

Initially we grew fast on the back of improvements in communication and iterations we did based on insights we got from calling and meeting service partners. But after a while the numbers started to stagnate, no matter how hard we pushed.

Digging deeper into this revealed a conclusive insight — “Service partners are excellent at execution, but most of them are not so great at selling”. This is where we realised pushing more was a waste of time. More specifically, on inspection, the next iteration didn’t seem doable within the sign-up time limits. Unless we did something radically different, we concluded that partner referrals can only get us 5–10% of daily sign-ups. They still came at less than 1/10th the cost of most channels. So while it was disappointing at the time, rapid inspect-iterate cycles did give us significant impact and insights in hindsight.

Sign-ups through delivery partners, date-wise

Story 2 — How we kept 5x consumers happy

This story begins with a fight.

I and Suneet (our customer relations in-charge) were shouting on each other at the top of our voices, late in the night. Early access to sale was about to open in the morning.

Suneet wanted me to put a pop-up on the payments page, telling customers about a tip to maximise cash-back. He believed this would improve customer experience and lead to lesser cancellations.

I was fighting him because I thought the pop-up would block the payments screen. I believed this would spoil the payment experience, especially for not-so-savvy customers, and would lead to lesser checkouts.

We decided to go Suneet’s way, with a 10% control group setup my way to compare and find out the truth.

Next morning, I checked to realise Suneet was right. There was in fact a ~5% increase in conversion on the payment page because of the pop-up. For me, an old belief of never adding steps in a funnel was broken.

Long story short, we went crazy with in-app messaging after that! We started monitoring chat channels and quickly identified most frequent pain-points customers had around the sale.

At first, the biggest customer anxiety was “whether I need to apply a coupon code”. People were scared of checking out, fearing if they would miss out on the “100% cash-back” offer we were running. Another common question was whether the offer was applicable on every product.

We quickly changed all our creatives and communication to address these concerns. These questions instantly dropped from chat queries, we had managed to reduce CRM load by such simple solutions!

Before (left) and after (right, with checkboxes highlighting answers to customer anxieties)

The next biggest concerns now were around some special deals we were running in the sale, which were hard to understand for consumers. We added a few helpful in-app messages to explain these and triggered automated notifications on coupon code failure explaining the offer.

After solving for quite a few more such concerns, we reached a point where the most common concerns that remained were consumers not believing us because 100% cash-back was too good to be true for them!

We decided to create an automated chat flow (using a tool we were still evaluating at the time) to answer most common queries without manual intervention. With an amazing effort from our support team and help from the chat tool vendor, we were able to iterate on the flow as well to automate most of the inbound queries related to sale.

A by-product of these in-app “interventions” was that almost all of them increased conversions on the app. So reducing customer anxiety does lead to business growth. Hmmm.

Once we got familiar with this new weapon, we expanded our focus area from reducing anxiety to driving checkouts and user growth through sharing. Because of these rapid inspect — iterate cycles, “in-app messaging” became the new skill we acquired quickly which unlocked new possibilities.


Frankly, by far the biggest reason the sale quadrupled our revenue was because of the awesome 100% cash-back deal which was too good to be true for customers (more learnings about having extreme simplicity in communication for customers, maybe for another post). But I think it was this constant inspection and iteration by all teams, driven by the omni-present daily GMV ticker and 2 stand-ups a day across all functions (another whole set of learnings there about team-work right there), that made sure we multiplied and maximised impact.

Launch quickly, inspect closely, iterate rapidly was originally published in Lambda – The Grofers Engineering Blog on Medium, where people are continuing the conversation by highlighting and responding to this story.

Source: Grofers

Leave a Reply

Your email address will not be published.