Product startups lessons learnt (personal analysis)

Personal Experiments · Source

Context

I’ve failed with 5 product startups and am currently running one that is, at best, a lukewarm success (S$8k in MRR + roughly the same amount in one-off consulting contracts in non-Covid times as a solo bootstrapped founder).

Writing this document to summarize my own lessons for running companies, so that I don’t repeat the same mistakes again and can also share this with potential co-founders to help them decide if i’m a good fit for them.

Personal principles for startup

1. In the absence of network effects, there is no early mover advantage.

The first mouse often gets snared by bad customer support terms, marketing the wrong value-prop, and not knowing customers' ROI from its products. The second mouse learns from the first, and steps over its carcass to avoid the trap and get to the cheese.

Take the example of visual search. Visenze seems to be the most well-capitalized visual search company right now with the best partnerships. It was founded in 2012, but there were a ton of visual search companies founded before this or around the same time that ended up failing. Kuznech (founded in 2011) raised $2.3M and died. Meanwhile, Cloudsight (also founded in 2012) raised a $7.5M Series A round and is still running as a profitable company with high single digit million dollars in annual revenue.

Personally, when I have tried to sell completely new products to customers (like this), they have always shown interest but never bought anything. But selling products in markets customers already understand (like this) has resulted in a lot of conversions

2. When working on SaaS products, stay away from large customers at the beginning.

Getting a large customer at the start can feel like a huge win+validation for you – and can be very tempting given the money (6 figure annual contracts) that they offer. But working with these customers can be a time and energy sink. They often have sub-optimal internal teams who can’t extract much value from your product without hand-holding. Moreover, they often buy based on how well you manage the RFP process and not purely based on the merits of the product – these sales can lead you to think that you may have product-market fit when you actually don’t.

Moreover, working with these customers can lead the founders to go away from adding features and talking to new users and into the minutiae of customer support. Sometimes, this is useful. Other times, it’s a waste of time.

Lastly, if a single large customer knows that they account for 30% or more of your total revenue, they will often try to bully you and get you to internalize a lot of support and customization costs. This can lead you into a trap where you are stuck with large enterprise customers instead of creating truly self-serve solutions.

Start with startups and smaller companies (up to $500/month contracts) instead. Though it’s important to have a plan to eventually scale to large companies and 6-7 figure annual contracts.

3. Care about scalability from day 0

When working on a data-heavy project, make sure that you can create products that are scalable and can handle millions of requests a day without breaking. When working for a data startup in 2015, we were able to onboard a very large customer in Dubai. But our system broke under their load (around 5 million pageviews per day), and caused massive issues on their website for a day. After this, we lost the customer and also other leads that we were developing in the Middle East

It’s not enough for a data API to give great results when tested locally or shown in a demo. It has to be able to work at scale.

4. It’s better to serve a currently small/medium-sized market that is rising fast (AI, wearables, biotech), than a large market that is shrinking or plateauing (media, retail, construction)

I have spent a lot of time working in the media industry. It’s great for your ego (everyone recognizes your clients, you often get to appear on TV/newspapers or in online videos that are watched by hundreds of thousands of people). But it rarely translates to long-term revenue. This is because of two reasons:

Sunset industries are extremely inefficient and are bogged down by layers of bureaucracy. This means that founders have to get sucked into internal office politics for customers to extract the most value out of their product. This limits their time which could otherwise have gone into growing the business
It’s difficult to attract talent to serve sunset industries, as smart people prefer to work with clients that are more future oriented

On the contrary, starting with a market that is currently small but is growing rapidly helps startups iterate quickly with customers that are happy to experiment, and helps them avoid the temptation of too much revenue too early. It’s interesting that most startups that made it really big ($10B valuation) either spent their first year serving no large businesses, or in serving small businesses (Google, Facebook, Shopify, Cloudflare, MongoDB, Bytedance, Stripe, and more)

5. Be fair to good talent – be ruthless with those who don’t perform

Smart people are a huge asset. I have had the pleasure of working with some incredibly smart people by hiring them as consulting researchers or designers when they were still students or had just switched to a new career path. I paid them fairly and gave them a 30-40% bonus whenever they yielded results that were far better than expected. Moreover, I gave them permission to use any IP they developed for my company for their own projects – as long as they were using it in ways that didn’t compete with my company.

This resulted in them being motivated to work, creating some excellent technology, and in actively providing support when needed even after their consulting term had ended.

Conversely, I have also terminated agreements with people within a week if it became clear that they were not a good fit and that continuing on a project with them would be a waste of time. Tried to do this as humanely as possible, but was fairly darwinistic in decision making. Would like to continue sticking to this – so that the bar for quality remains high while money (which will be a scarce resource) is conserved.