2024 Reflections on Building Refrens.com

Sunset at Promthep Cape, Phuket. November 8 2024.

Continuing from my last year’s update.

Product

  1. We had a 2.5X YoY growth in monthly revenue. 8% monthly compounding. Lots of processes changes. This year I am expecting a 4X year.
  2. The product is expanding to become the financial department for business of all sizes – from new age startups to scaled up public companies.
  3. We are an accounting app now. This year we will add the accountant and CFO layer, thanks to AI.
  4. There are marketing and sales modules that are branching out due to customer demand. Each distraction new direction in SaaS is just one more feature to add.
  5. Banking relations is still a challenge. Of all the contacts I have in fintech, its still unsolved for us. Banks had opened up their APIs for a short period during 2018 to 2022. Lately, banks in India have become less open and that restricts what our product can do. Hopefully this year we should be able to resolve that.
  6. Service as a Software or remain a product company? We have thought of this pre-AI boom also but now this is on everyone’s mind. We will resolve the dilemma this year. For now it seems we will enable accounting teams and services companies to serve their businesses better, that will keep us light. But then the value attribution won’t come to us. Dilemma!?
  7. We have paying customers in 124 countries all grown orgnically. This year we will focus on just 4 countries and try to scale them with more predictability.
  8. In some of the cohorts that we want to win in India, we are now number 2 in terms of new user acquisition and product maturity.
  9. Adding the cherry on the cake is worth more than the cake but you still have to make the whole cake first. Now that the base product is mature, we added to 2 new modules on top. Each of these module sells for 3X to 10X the price of the base product.
  10. SaaS becomes a commodity very fast. That makes it a popularity contest. Some really bad accounting software from the non-internet era still gets mentioned by users. That still keeps us our marketing team on toes.
  11. The moat is the scale. Scale of the product that only requires adding different cherries/toppings. And scale of recurring revenue that makes adding the topping possible even in a bad quarter.

Finance/Revenue

  1. We started the last year with 11 months of runway in the bank. After 12 months we have 7 months runway. This month we might actually become cashflow positive.
  2. Revenue does solve a lot of problems.
  3. We expanded some teams faster in anticipation of revenue. We could have waited one more quarter. There is no right timing here I guess.
  4. Accrued revenue vs cashflow is a weird thing we had to adopt this year. Still makes little sense to me. Lots of financial engineering in startups happen around this and leads to poor outcomes. We are sticking to cashflow for internal MIS.
  5. When you are growing well, profitability becomes a choice.
  6. VCs still don’t love SMB focussed startups. My thesis is that growth curves are not very exciting, there are can rarely be a 10X year. And the user persona does not resonate with them.

Team/Org building

  1. Team size has grown 40% in last 12 months.
  2. We have added new roles in every team. Trying to create a multi-level hierarchy. This has slowed us down but gives us more visibility. I am hoping to see some gains because of this in the current quarter.
  3. Each major feature now has an owner. They are measured on revenue but they do not do sales directly. This is the layer between product manager, tech, sales and support. At smaller scale this would not have been possible.
  4. For a brief period I had a young college grad help me with new initiatives. In about 4 months he helped me start 3 new things that would have otherwise been stuck in my head forever. I will open this as a permanent position.
  5. We are taking “what cannot be measured, cannot be improved” very seriously. While sales people get measured easily, others are difficult to do. New age teams do not like to be measured on number of hours in office, neither is it the right proxy for work. So we came up with some crude numbers. Implemented it in 2 teams and the output has improved significantly. The variance between the highest performer and lowest performer has reduced drastically. We will refine this number further.
  6. Some teams are now being given cash incentives, over and above their promised CTC. By the start of next quarter I am targeting to have variable incentives for every team member. It helps the managers give better direction to each member. Else there is too much chaos and too many people running in multiple directions.
  7. One person, one task, one metric. We have to adhere more to this.
  8. I tend to hire people for intent over skill. This has led to multiple finger burns. We changed the hiring process and it has helped keep my biases in check.

Ecosystem

  1. Apart from Saasboomi, in Chennai and Surat, I attended a lot of new events. It feels good to meet users in person.
  2. Lots of learnings from those events. AI will kill SaaS, to AI will grow SaaS bigger.
  3. We have a growing set of well wishers, who I have never met in person, who send us new users and potential hires. Really thankful for this.
  4. The number of startups building in our space is increasing. Narrative is changing from Accounting is a solved problem, to Accounting is a boring SaaS, to All B2B fintech will offer an accounting product.

Random

  1. People still read Refrens as “Refrain” and type it as “Refern”. The trailing “s” is ignored. Not sure why. It is a play on Reference, so it should be pronounced as “ref-uh-rens” or “ref-rens”.
  2. I need to write more often, beyond twitter.

2023 Reflections On Progress of Refrens.com

Satpura Jungle, Madhya Pradesh, India. On an early morning drive to see some lions. 24th Dec 2023.

Product:

  1. We grew 3x in terms of Monthly Revenue. 10% MoM compounding. We need 3 times 3x and 2 times 2x to go 100X. But we need a 10X year before that. 
  2. We are an invoicing and accounting app. People love us (and pay) because we handle the complete journey from lead to quotation to invoice to accounting.
  3. We are not a market network, yet, what we originally set out to build. This keeps me mentally occupied. But right now the focus is to grow revenue for what we have.
  4. Consumer product – each customer is not worth enough to call. B2B – Each customer is worth going on a feedback call.
  5. We have paying users from 103 countries. I still don’t know if this is good or not for an early-stage product. We partially designed for this but never optimised for this. Happened to us organically, so yeah, it should be fine. But are we serving them well?
  6. Direction bothers me more than speed, on weekends.
  7. Speed bothers me more on weekdays. I don’t get a clear sight of direction in day-to-day work.
  8. Our target market is large, this helps us make decisions for the long term.
  9. Build essential features first. Add speed. Then fuss over a great design. People don’t notice good design much. Some people notice bad design.
  10. You can only know what is an essential feature if you talk to customers every day. Or live with them. Or you are them.
  11. When you don’t have PMF you tend to go breadth-first and end up having multiple projects that are loosely connected. When you have PMF, you can go depth-first and it is easier to sequence the priority for a PM.

Finance/Revenue:

  1. Some problems can be solved only by adding more people. I should have realized this much earlier. I wanted to solve the revenue problem with tech.
  2. Users pay a significantly higher premium if the product is complete for the JTBD.
  3. There is no bigger fear than that of money running out. Specially once you think you have something meaningful built. I have stopped thinking about negative scenarios for the past few months. We will sail through fine.
  4. Time or opportunity cost rarely bothers me. I am told it should bother me.
  5. Passive income to support a founder’s personal expenses takes active energy. If the family is helping you, it takes energy to manage them.
  6. I have stopped looking at marketing charts on Google Analytics and now look at daily revenue charts. Partly because GA4 broke some of my favourite reports. Partly because there is nothing interesting on Twitter.
  7. Having a pulse of cash flow helps keep my anxiety low. I want to get over with tracking cash flow for 6 months. Hopefully soon.

Team:

  1. Distributed team across Bangalore and Surat (and some WFH) helps keep the costs low. Branding of Bangalore with the cost of Surat. It was initially done because we (Mohit and me) were already in 2 cities when the product took shape. Sometimes it makes me feel that this is slowing us down.
  2. Having a few canons in the team can do wonders. Helps distribute founders’ workload. Don’t need too many canons though.
  3. Delegate things that don’t directly affect the product. Don’t delegate other stuff until there is a very strong reporting and feedback loop.
  4. Interns bring a different level of energy to the workspace. They are generally the happiest people on the floor. And that is contagious.
  5. People Managers are generally grumpy. We need fewer Managers.
  6. Any freedom that breaks a process for the rest of the team is not freedom. It is a discipline issue that must be addressed immediately. Demand for more Freedom is just less belief in the mission.
  7. Letting people go is fine. They already know that it is coming. Your team knows it earlier. We let go 3 people this year. All in less than 2 months from joining.
  8. One person one job is very difficult in the initial days. Creates too much cost load. In later stages, if one person has multiple meaningful tasks, it is a sign of low skills, either of the worker or the manager.
  9. The larger the vision, the better the people who want to align with you.
  10. Having a dedicated office space improves my communication. We took over the Surat co-working space in November. I can now shout across the corridor or sing to relieve my stress. I cannot think of going back to the older style.
  11. We are 54 people now, including interns. We need to 4X this in the next 12 months.

Ecosystem:

  1. Community events have been really helpful. People go to any extent to help you if they see that you are passionate.
  2. Stories are good. Numbers are better.
  3. Meeting in person and sharing your problems works best to build relations with investors. Current and new. We have 65 investors on capable and about 100 including the ones through syndicate.
  4. After taking investment you are competing with your investor’s other portfolio companies from very different sectors for their attention. If any of their portfolio companies grow faster than you, you will lose attention.
  5. No one likes slow movers. They stop cheering for you if you don’t keep fuelling their excitement.

Understanding The Economics Of Edtech Startups

I spoke to a few early-stage EdTech startups over the last few weeks. I had multiple questions answered. Here I am looking at the unit economics of such platforms.

The economics is mostly dependent on the student-teacher ratio. And that defines the whole experience as well.

There is broadly 3 range of ratio here:
1. 1to1 – WhiteHatJr.
2. 1to Few – Cohort based. GreatLearning.
3. 1to Many – Recorded classes. Byju’s.

The cost of acquiring a user(learner) is the same for everyone. The gross margins are 20% to 92% depending on the student-teacher ratio.

A 20% margin business for a non-branded teacher is impossible to break even. In D2C, 70% gross margin is supposed to be healthy. The same should apply here.

What would work well?
A hybrid model, where pre-recorded classes are followed with a few cohort-based workshops and then 1-1 doubt solving. This is the ultimate package that removes the creators branding requirement also. And gives control to the ed-tech platform.

Problems as compared to offline education
1. A 1-hour recorded video is about 3hrs of live class equivalent. Generally, recorded videos are condensed & edited, which results in poor learning because there is no buffer time to make notes or to think.

2. Chat for doubt solving is flooded with 100s of students. The rate of resolving doubts is poor.

There are 3 types of learning and all 3 are measurable in their own respects. :
1. Certificate-based – Mostly upskilling courses.
Branding of the certifying institute. – GreatLearning has Stanford, IIT etc.

2. Outcome-based – Competitive exams for admission.
The past success of institute – “AIR 1 is our student”.

3. Hobby Learning – ECA.
Branding of the teacher – Check Frontrow – Celebrity teachers.

All the 3 above are socially tangible.

Unlike learning and gaining knowledge. If you are just curious and have the capability to “learn on your own” – Youtube is enough. Alas, most of the millennials haven’t been taught to learn on their own. We need to be guided.

Understanding The Opportunity In Creator Economy

Originally posted as a Twitter Thread.
Interesting how Threads have become easier to write than blog posts.

We are trying to decode the creator economy that we use to guide us at Refrens .

The most important thing before we start. All creators must have their own audience. If your audience is owned by a platform that might go hostile later (a la FB) you are swimming in the wrong sea.

What you monetize is your audience. Your audience trusts you, your face, your voice, your brand. So like a teacher or doctor, your work is not delegatable. You only earn from as many as you can directly reach.

The creator economy falls into 2 brackets with respect to where their money comes from.
1) Those who teach the audience a new skill – and the audience pays. Example – @VaibhavSisinty@LiveFromALounge
2) Those who sell the audience a new product – and a brand pays. Example – @Bhuvan_Bam – also called Influencer

Change in Execution:
A. Some are creating digital products for other creators.
Lack of potential in capturing value, i.e. collecting fees for advice, enforces this. So you create an enabler product/platform instead of advising.

B. Some creators are now starting to own the brand that they promote ( a la Ramdev-Patanjali)
A broken influencer to sales attribution system enforces this.

What happens to other creators – like writers on @TeamPratilipi or @YourQuoteApp? They fall somewhere between 1 & 2. The audience pays, not to learn but to consume the creation. Freelance writers used to get fixed-fee from editors. Now they are micro-publishers themselves. (Read more about micro-entrepreneurs on Platforms)

Products in the creator or freelancer economy can do 3 things:
1. Provide a tool to create – Video editing tools, podcast creation tools etc.
2. Provide a platform to distribute – Reach and manage the audience.
3. Provide a platform to manage the business – payments, leads etc.

A good platform must do 2 of the 3 things above.
1. Youtube does all 3 of the above. But #3 partially only.
2. Spotify does 2 directly and 1 through Anchor.

Doing only 1 of 3 above means you are doing nothing specific to this audience. You would be anyone like Figma/Freshworks, with lower ARPU. You will be a commodity tool.

The challenge is in selecting a target market that is still large enough even when you combine 2 of the 3 above.

How You Can Exploit The Tolerance Level of Users To Increase RoI – Some Examples

When pricing a new product, understanding the tolerance level of the user is important. Knowing what is the maximum you can charge for a product without turning down your customers helps you maximize your revenue. This tolerance level theory applies to every feature optimization. The idea of exploiting tolerance level in product design is to move away from being a purist and keeping interactions clean to finding the right balance between clean product and clean product that makes money.

To be able exploit tolerance you need to understand what your true product is. Example – For most of India, a flight booking OTA’s true product may be booking the cheapest flight easily. Cheapest is the real product, easily might be exploitable for tolerance. A food delivery companies real product is food quality and delivery time, interface of app might be exploitable for tolerance.

Here are some examples that will help you understand what negotiables can be exploited for tolerance. Some product managers might want to call it a Growth Hack.

  1. Way2SMS – A free SMS sending website. From the time of landing on homepage to sending the SMS, a user is shown about 15 ads. You might call it too much but if you understand the users’ profile you will be less worried about interface and more worried about delivery time of SMS. Way2SMS’s delivery time is under 2Seconds, even in peak hours.
  2. Akosha –  A freemium dispute resolution system for consumers. Akosha has pivoted to being Helpchat. Akosha used to send a notice to the disputing service provider on behalf of the consumer for free. If the dispute isn’t resolved, they charge the consumer Rs.500 to follow it up. I and friends have used Akosha thrice to solve disputes worth Rs.13K, Rs.30K, Rs.1K successfully. In all 3 cases Akosha did not charge, because the free service was good enough. Was there room to charge a fee/tip after the service? Absolutely.
  3. GoZoomo: A friend used Gozoomo recently to buy a second hand car. They are doing a lot of offline work with the RTO for him, all for free. Charging for actual expenses borne on behalf of the consumer wouldn’t hurt.
  4. Pinterest – If you land on Pinterest from Google, the first 2 folds are visible without login but when you scroll down further, you are asked to signup. Good balance of freemium.
  5. Quora– Like Pinterest, Quora only allows you to read the answer that you directly landed on, everything else requires you to log in. Sometimes the user has no problem signing in, it’s just that you haven’t asked him well enough or you have given him a “Skip for now” button.
  6. FindYogi – At FindYogi, we ask the user to login to see coupons applicable on a product. Most of our signups come from that 1 feature.
  7. Coupon sites – They open the destination site before showing the actual coupon code. Does the user have a problem with that? Well, their growth doesn’t suggest that.
  8. Insurance sites – Most of the sites ask you for contact details before showing a quote. If you are really serious about buying  insurance, you may not mind it. Afterall, what’s the use of spending so much marketing money to bring the user on the site and not even creating a hook to contact them again.
  9. GoDaddy – From the time of selecting the domain to buy to actually finalizing your order, they try to upsell you 3-4 products viz. SSL, Hosting, email, related domains. Does it hurt? Well what are the chances that he will bounce from step 2 and go to a competitor to buy a commodity product like a domain? Low revenues hurt more.
  10. BigBasket – Untill recently, the earliest delivery slot that you would on BigBasket would be atleast 36hrs away. What really worked for them is the fullfillment rate. They were so good at getting what you ordered that they were ready bet on it with a 50% premium on refunds made for non-delivery. And when they promised a time-slot they really delivered then. Delivery slot is a negotiable, partial fulfillment or delay is not.

Not exploiting tolerance is like leaving money on the table. And once you know what the negotiables are, you will work harder to improve the non-negotiables. A lot of times maximizing the negotiable tolerance actually helps you discover a sustainable business model.

The important thing with tolerance level is that your margin in the tolerance exploitation is the opportunity for a new player to get it. BigBasket vs. Grofers is a good case here.

The Overload Of User Interfaces – Who Should You Develop For?

Back in 2005-06 the world had almost come to working with browser apps only. This was after we had started giving up on Yahoo chat as a desktop downloadable tool. Soon after mobile started picking up and WAP sites got popular. Later in 2007-08 iPhones were launched and all hell broke loose. Web developers now had to develop for environments that were different from the traditional web. Nokia’s symbian was still popular in developing economies. And then Android also caught up.

As if all the fragmentation wasn’t enough, browsers like Opera and UC web were running their own standards of HTML and no-JS sites. Google Chrome was lagging in mobile web story due to low bandwidth in developing economies so they started re-directing traffic through Google Web Light. They recently introduced AMP (Accelerated Mobile Pages) as a standard for content sites on low bandwidth. In a bid to reduce bandwidth overload for extended usage some developers took to Single Page Applications (SPAs) as well but that had it’s own challenge in terms of first time load and SEO.

With the whole fight around apps/no-apps, Google is now introducing app streaming for users with high-bandwidth access. So what does a developer do?

We tried to do a break down of all potential interfaces. Here’s what we got. Might be useful to think around this.

Screen Shot 2015-12-28 at 6.30.08 PMWeb interface break down for India. (Click to expand)

Facebook’s Copy Writing Change For Events RSVP Is Small But Magical

Facebook’s events’ RSVP earlier had 3 responses – Going, May Be, Not Going. Lately, I have started seeing the responses as – Going, Interested, Can’t Go.

  • The change from May Be to Interested may not mean much as both are non-confirmatory but Interested is more positive than May Be. 
  • Changing Not Going to Can’t Go is a major change. Can’t Go already has a sense of willingness to go but declining due to uncontrollable circumstances. The aim of the event RSVP is to capture data for the host in the most positive way, and this change will enable more people to respond without hurting the host’s sentiments or feeling guilty about it. Facebook has done this earlier also.

The UX of Social products has a lot to do with copy writing. Those little changes will make all the difference in engagement rates. Read more here.

Update: Facebook has 2 sets of responses depending on whether the event is public or private. This makes the detailing all the more interesting.

FirstCry’s User Generated Content Strategy For Reviews

FirstCry, ecommerce company for baby products, is asking shoppers to upload pictures and videos of their child with the product. These pictures would be showcased on the product page.

first cry

This feature is a win-win-win for existing users, potential new users and the site.

1. Rewards – For the reviewer. Seeing your kid featured on a popular ecommerce site is a huge ego boost for the parents. This is almost like a mini celebrity stature. This is lesser effort than writing a 200 word review.

2. Testimonial – It’s a proof that someone else has bought the product from this site. It is more than just saying “Genuine buyer”.

3. Better Decisions – Since these are genuine non-paid pics, shoppers are more like to trust them and can easily visualize real world view of the product.

In the days of selfie, expect this trend to catch up with other ecommerce portals.

Cleartrip Activities: More Than A Product Vertical

Cleartrip has launched Activities. As the name suggests, the section lists various activities that you can take up on weekends or otherwise. While it might look like a new product vertical for travelers, there is more to this product vertical, or rather engagement strategy.

There are 3 things to note about this feature:

1. It is app only.
2. The launch cities are not where people go to travel but where the Cleartrip’s audience lives.
3. The activities chosen are for locals, not for travelers visiting the place. Most of them.

For a segment like travel booking, which happens couple of times in a year, most people would uninstall the app after a use. This feature gives the users a reason to keep the app and open it every week. I would expect Cleartrip doesn’t limit itself to paid activities only.

It is very important for transaction systems to have an engagement feature else the acquisition cost is not limited to every new user but for every new transaction. And with transactions being a commodity, the product might just become very difficult arbitrage game.

Your product always needs to do more than just the feature that makes you money.

Incase you like this feature, Explara and Townista are doing this decently.

Flipkart had recently promoted an engagement tool on its app, albeit short lived – Read about the #ThumbThing.