Natural Intelligence’s Phenomenal Story: A Tech Perspective
This blog is adapted from a talk given at Microsoft Israel as part of the CTO Breakfast Club series in February 2020. All images have been taken from management / company sessions.
When I first joined Natural Intelligence two years ago, the company was approaching a decade of incredible profitability and growth since its founding in 2009.
But the company’s founder and CEO Nir Greenberg came to the realization that NI had barely scratched the surface of its potential.
This is the story of my first two years at Natural Intelligence — a story of cultural transformation derived from technology and product, which I had the privilege to lead.
Bootstrapping and Opportunistic Growth
Natural Intelligence had been around for nine years by the time I got there, and was already a global leader in multi-vertical online comparison marketplaces. Basically, we help people make decisions on what products to buy by presenting them with shortlists of the best brands for them (www.top10.com is a great example).
The world of online services comparison was brand new — a vast blue ocean waiting to be sailed. The company succeeded and opened new verticals at a breakneck pace, getting bigger and bigger through opportunistic growth. Nir took no outside investments and always maintained profitability while putting profits back into the business.
Shifting Mindsets
Around 2017 the realization was dawning that despite the great growth that the company had experienced, the opportunity was even bigger.
People love to compare products and services of every kind, and really need help making decisions on which products to choose. Therefore, the potential of comparison sites is much, much larger; and despite the success in our first decade, we had only scratched the surface of market potential.
Nir envisioned tech as being NI’s “Exponential Engine,” a main pillar in the company strategy which will expedite growth. The vision was to turn our underlying technology into our competitive advantage and an enabler for business scalability. To bootstrap this transformation Nir recruited Moti Karmona as a product and e-commerce leader and myself to lead the Tech.
Rebalancing
NI started as a bootstrap company built on its performance marketing expertise and successfully evolved without having tech as a substantial pillar. The business departments had been leading the way and defining both the ‘WHAT’ (to develop) and the ‘HOW’; it also had its own dev teams dealing with the more delicate tracks.
Rebalancing, and getting important stakeholders to buy into changes was done incrementally. Small successes built trust up over a two year period. It was a process of learning how to meet the company’s truly big challenges. I believe that our super unique “setup” — Nir the Founder is also the CEO, sole owner & the Board, was the key in our ability to drive cultural change through the company effectively; another key success factor was to rebuild the unit’s manpower, reaching the right talent that can gain trust and deliver sound solutions; and lastly, it was the collaboration, shared goals and engagement of many super talented employees from across the company.
In the following paragraphs I will share additional factors that in my view made this journey successful.
Reflection
An important factor in instituting such a big change is to create a set of baseline KPIs to accompany the journey — to let the company look in the mirror and see how we’re progressing with time. The management reviewed these KPIs periodically. At the end of 2019, we were out of red, and into the yellow and orange range.
This was an important tool to assess our progress in a consistent way. Transformation means the ability to push the horizon forward. It has to be holistic. It’s not enough to have only two reds become green.. The purpose is to get everything gradually from red to orange to yellow to green.
Setting the House Rules
It is important to create a framework that enables an effective transformation process and reduces the friction between the teams. We focused on the following 3 pillars:
1. Gradual Transformation — When the business is profitable and going forward at 200 mph, you have to swap out parts piece by piece. A phased implementation of this kind of change lessens risk. You have to get everyone used to changes, and not try to give them the full transformation all at once. This also serves well my philosophy to meet production as early as possible.
2. Product Leadership — A tech transformation may sometimes be led by engineers, but in our case this couldn’t have worked. If you’re trying to reverse engineer 20 years of work (4 years * 5 engineers), the result can’t be good. We didn’t want to try to clean up what had already been done. Sifting back through no longer relevant logic and instances from 4 or 5 years ago was not possible. We insisted that Product redefine, together with the business, is the right approach for us.
3. No Rollback Policy — Implementing a series of changes that affected the core KPIs of the business could create a lot of tension and noise…. just when we needed to focus on the execution. We decided that the only way to succeed is to move forward and adopted a ‘NO ROLLBACK’ policy.
Measure & Visualize
I’m a true believer in creating internal and external visibility. The focus on such visibility in my teams is motivated by a very selfish reason — I want to identify potential risks and mitigations without needing micro-management. This is an extremely important factor in the success of a transformation process since it provides more immediate feedback on our progress and creates trust between groups, i.e., visibility at the team level, enabling management to track execution, while creating trust among the business stakeholders.
It’s important to measure tactical successes. It gives you and other stakeholders an understanding that the vector is going in the right direction. One example of such a tactical thing we measured is how many data-related incidents we had. We saw a drastic decrease, and were able to present a positive objective measurement.
This was part of building trust in the solution for us and the business. In addition, it provided a confident boost for the developers.
Innovate — Identify Where Magic Can Make the 10x Impact
Amid this continued transformation you can’t give up on innovation. While working toward the goal, you still need to identify opportunities for the magic touch that can give 10x impact.
This serves:
a) The company’s growth.
b) Retention — transformation can seem like a sisyphean process. It’s hard to retain talent who want to be on the cutting edge, making those 10x contributions. They won’t stick around for two years because you’re nice. They need a little bit of magic amid something that can be a bit bland.
Two successful examples from the last 2 years:
1) Experimentation platform — In 6 months we cut the time for web experiments (A/B testing) by 50% by using a different statistical model (moving from a Frequentist to a Bayesian approach). This enabled the company to scale up our web experimentations significantly. More details can be found here.
2) Contributing to open source — We have built our machine-learning pipelines foundation with the intention of sharing with the open source community. In May, our project was accepted to the Apache Software Foundation as an incubator project — Apache Liminal — (https://incubator.apache.org/clutch/liminal.html). This is a huge achievement for myself and for the company. Building a community around the project in such a domain has a great benefit for the company, since more developers (from other companies) can evolve it faster, and it has a great impact on retention and hiring talent.
Creating a North Star
A big part of realizing something abstract like an “Exponential Engine” is to apply the concept on a more practical level that connects the different parts of the company. Getting buy-in from all these departments was critical: Business, Ops, Product, Tech, Marketing. You want and need all these people with you on the journey.
The above slide about our plan to “Automate Everything” was presented rather late in the process in a company meeting. It explains the idea of the Exponential Engine and how it’s relevant to every stakeholder in the company. Every person can relate to one of the junctions, and can see clearly where it will meet them in their day-to-day work. This says, “friends we’re going to automate the business together and connect business, tech and product to drive business scalability.” I wish we had drawn this vision earlier in the process.
Part of the challenge of this kind of transformation is that it’s a culture change. The idea of automation sometimes creates fear, which has to be dealt with as well. The goal isn’t to replace people with automation, but rather to let them scale up, and focus on the essence, not be spread thin with repetitive tasks.
The New Balance
The past 2 years at Natural Intelligence have been an amazing experience. We have rebuilt NI’s Cloud & Analytics infrastructure and even managed to make some 10x impacts along the way. We have reached a new kind of balance, but keep fine-tuning and finding the best ways to push the company toward reaching its enormous potential over the next few years.
