Evolving our investment strategy
This is a long post (1,900 words). For those of you who are time poor here’s the tldr:
Forward Partners operates a focused investment strategy because it helps us make better investment decisions and provide better support to our companies.
A good focus area for us is one that can generate 50+ deals and where we can build some generalised expertise that helps with our decision making and value add.
Until now we have focused on marketplaces and next generation ecommerce
Recently we evaluated lots of options and did a deep dive on Applied AI before selecting it as our next area of focus
For the three and a half years that we've been going, Forward Partners has operated a focused investment strategy. We observed that small transactions of all types are increasingly moving online and backed the companies that were helping to accelerate that trend. That meant lots of consumer and small business focused marketplaces and next generation ecommerce companies. Lost My Name, Appear Here and Thread are three of the better known examples, but overall there are 37 companies in that portfolio.
We chose to be focused for three reasons. First, and perhaps most important, being focused enabled us to build up expertise that resulted in better investment decisions. Specifically, we feel we have strong capabilities in working out whether customers will value products highly and whether it will be possible to market them cost-effectively online. Secondly, we have seen so many similar companies now that we have a good sense of what they should be achieving by when. We are better able to see problems coming and advise on strategies to work around them. Being expert in an area makes us better board members and hence better able to win deals with the best entrepreneurs. Finally, focusing allows us to add more value operationally so our companies can execute faster and with higher quality. The companies we back often share the same challenges as each other and because we focus our team has solved those problems many times over.
However, venture capital is a dynamic business and good focus areas don't last forever. We are still seeing lots of marketplace and next generation ecommerce opportunities, but as we move into our second fund we decided to add another focus area to make sure we will continue to have enough high quality opportunities to invest in over the next four years.
Our first step was to define the what we mean by a “good focus area”. For us the following characteristics are important:
Will generate 50+ deals
We can build knowledge that's broadly applicable across the focus area and gives us an advantage versus other investors
We can articulate a few underlying investment theses
We can articulate use cases
Suitable for early stage investment
The UK has some kind of advantage
Then we had a high level discussion about what areas we might focus on next. A couple of interesting things came out of that. Firstly we like to invest in sectors that are rising from the low point of the Gartner Hype Cycle. Investing at this point leverages our key capabilities of assessing whether customers will love products and whether companies will be able to market them cost-effectively. If we get the timing right then mass adoption should be achievable. Investing with this strategy means we don't chase the very rapid value appreciation that sometimes occurs at the beginning of the Hype Cycle, but we think the benefits of focus outweigh the cost of the lost opportunity.
The other interesting point to come out is that investing in deep tech at the very earliest stages is difficult. One of the key drivers of success for us as a fund is backing companies that make rapid progress and are able to raise up rounds a year or so after we invest. To do that they must pass valuation milestones. With ecommerce and marketplace companies those milestones relate to sales and unit economics and are easily demonstrable. Progress at deep tech companies, on the other hand, is based on internal development milestones and it's difficult to predict how next round investors will respond. Until a product is released and is in the hands of customers, which can take years, the only evidence of success is internally reported improvements in algorithms and the production of code. I’m sure there’s a way to solve this for deep tech investments, but we haven’t figured it out yet.
The next stage for us was to brainstorm potential areas of focus. Each member of the investment team went away and over a couple of weeks contributed ideas to a shared Google Doc. Then we reconvened with the objective of choosing a single area on which to focus. Via a process of discussion, voting and then amalgamation of ideas we decided to look seriously at making “Applied AI” our next focus area. That would mean investing in companies that were using well understood artificial intelligence techniques to build new and superior products.
We felt that Applied AI is attractive because:
It’s a broad enough area to generate 50+ deals
Is one where we already have knowledge and could could go on to develop a deep expertise in the different techniques and their application
Is at the right point in the Hype Cycle and plays to our strengths in evaluating demand
The major concern we had is that AI more generally has been a popular investment theme with other investors for some time and we wanted to make sure that Applied AI is sufficiently differentiated to be a viable investment focus for Forward Partners.
We decided to go away and do some work to improve our understanding of the area with the aim of answering the differentiation question and convincing ourselves more generally that Applied AI has the potential to yield a flow of high quality investment opportunities over the next 3-5 years.
To that end we sought to answer the following questions:
What are the AI techniques that can be applied cheaply and predictably by startups?
What capabilities do those techniques enable? (e.g. natural language processing enables conversational interfaces)
What use cases can these techniques be put to? (e.g. conversational interfaces to FAQ databases can improve customer service)
Are there enough use cases where the addition of ‘intelligence’ makes the product meaningfully better?
How can Applied AI startups meaningfully show progress in their first year of operations?
How much AI talent is required at pre-seed and seed stage Applied AI startups and can we find enough companies with that talent?
How can we add value to Applied AI companies?
What are some hypothetical strategies for Applied AI startups to obtain the data they need to train their algorithms? (Addressing the “cold start” problem.)
The first three of these questions relate to the size of the opportunity set. To choose Applied AI as a focus area we had to believe there is the potential for 50+ deals that would make sense for us. To get an answer we mapped an extensive list of Applied AI techniques against the Gartner Hype Cycle, and put them into a spreadsheet linking them to the capabilities they enable, then linked those use cases to capabilities, and finally the use cases to ideas for companies. After that we scored the company concepts based on their attractiveness as Forward Partners investments and looked to see how many high scoring opportunities there were. Fortunately there were many.
Once we had comfort on the size of the opportunity we turned to the final three questions which relate to whether the opportunities will work as early stage investments. Our approach this time was to hold workshops and meetings with people who had experience of building applied AI businesses. Thank you in particular to Matt Scheybeler, Steve Crossan, and Martin Goodson for helping us with this part of the journey.
One important learning at this point was that in the early stages of Applied AI startups the artificial intelligence component isn’t that complicated. We heard multiple times that you can get 80% of the way there with statistics, that almost any AI technique will get you the next 10% and that it’s only when you get to the last 10% that you need to get clever. That was great to hear for two reasons:
Most startups with true potential don’t get to the last 10% in their first couple of years so hard to find AI talent isn’t a prerequisite to get started.
Our existing strengths in building products that resonate with customers and driving growth aren’t eclipsed by a requirement for deep tech knowledge - i.e. we can help.
The other important point we learned is that Applied AI startups can get product to market quickly and drive predictable value appreciation in the timeframe of a pre-seed or seed investment. We talked through numerous real and hypothetical examples and got confident that when we make Applied AI investments they will be able to raise their next rounds at a good step up in valuation. That’s one of the most important questions any VC has to answer and we were pleased to find that because they can get started with simple algorithms, Applied AI startups aren’t different from other software startups in this regard.
The final piece of our investigation was to think about the “Cold start” problem. We talked about three different data strategies for Applied AI startups and what that would mean for us:
Founders have access to some proprietary data
Founders have an innovative idea for using publicly available data
Founders will generate data from their business and develop algorithms later
In the first two of these cases Forward Partners needs to evaluate whether there is value in the data pre-investment and to help the founder extract value from the data post investment. In the third case we need to be able to evaluate whether the business will be able to generate data, and then if they can the evaluation is the same as in the first two cases. All of this points to us enhancing our data science capability at Forward Partners.
Our conclusion therefore, is that Applied AI is an attractive focus area for Forward Partners. It looks promising that there will be the required volume of dealflow, we can see how an early stage investment strategy will work, and we can leverage our existing strengths to help businesses in this new area. The only new requirement is that we enhance our data science capability.
Hence for the last couple of months we have been targeting Applied AI deals alongside our traditional focus area of marketplaces and next gen ecommerce. Wherever possible we like to take an experimental approach so we have decided that we will run with it until the end of the year and then evaluate. In parallel we are investigating what sort of data science capability we need. That will in large part be determined by the sort of opportunities we see and end up investing in, so for now we are relying on relationships with people who help us on an ad hoc basis with a plan to bring the capability in house when the picture gets clearer.
And I’m pleased to report that we have already made our first two Applied AI investments. Neither is announced yet, but watch this space :)