Monthly Archives

January 2018

Software startups: Beware ‘magic’ bullets

By | Startup general interest, Uncategorized | No Comments

I just read ‘Need More Time’? Guideposts For Tech Founders Going To Market When No Market Exists which is full of great tips for what they call ‘pre-chasm’ enterprise startups. The term ‘pre-chasm’ is a nod to Geoffrey Moore’s 1998 classic Crossing the Chasm and refers to companies that may have sold to early adopters, but haven’t yet found a way to sell to the mainstream. Getting sales going in those early years is terrifically challenging and requires great product and great sales. There are lots of common pitfalls that founders fall into and the whole post is well worth a read, but I want to highlight two sections which cover mistakes that in my experience many founders are prone to making.

  1. Over-value conversations and even deals with large enterprise customers. Here’s how they put it:Surely people paying you money for expertise is a strong signal you’re heading towards product-market fit? The twist is: In much-hyped new technology areas, before there’s a big market, it’s not uncommon for startups to close high dollar PoCs and even some large contracts simply because companies are happy to be educated by startups.This practice is particularly common in fintech.
  2. Believe that channel partners will accelerate sales. Here’s how they put it:I see this play out in new markets again and again: Pre-chasm enterprise startups throw time and resources at indirect sales channels (including OEMs, etc.) in the hopes that someone else’s sales team can do a better job than your own. Or, assuming it will accelerate sales, they will spend a lot of time with technical or channel partners … [but] rarely will indirect channels for enterprise devote real resources to help push someone else’s product to market. As for VARs, they typically only provide fulfillment in the early days (and if you’re really lucky, deal registration for qualified leads) because they aren’t structured to carry pre-chasm products — i.e., pitching, educating, hiring the right sales force. They’re good at distributing things where there’s already an educated customer base.Nine times out of ten (or more) direct sales is the only way for early stage startups.

As easy as… AI

By | Startup general interest, Uncategorized | No Comments

Google is pushing hard to make artificial intelligence as easy to access as cloud computing, building services that reduce both costs and the technical skill required from users. To my mind, there is a strong parallel with how Amazon Web Services made it easier and cheaper for companies to build web apps.

Throughout 2017 they made steady advances, releasing Google Cloud Machine Learning Engine and grew Kaggle, their community of data scientists and ML researchers, to more than one million members. They now have more than 10,000 businesses using Google Cloud AI services, including companies BoxRolls Royce MarineKewpie and Ocado.

And now they are introducing Cloud AutoML, which promises to:

Help businesses with limited ML expertise start building their own high-quality custom models by using advanced techniques like learning2learn and transfer learning

Google seems to be in the vanguard, but Amazon and Microsoft are pursuing similar agendas. For AI startups this means that machine learning expertise will become relatively less important whilst access to data and customer understanding will rise in significance. Given that ML expertise has been in short supply we can expect to see a sharp rise in the number of high-quality startups using machine learning to make better products (as distinct from low-quality startups that claim to use machine learning but don’t really). It also means that the opportunity space will tip towards applications and away from infrastructure.

At Forward Partners we’ve had “Applied AI” as one of our two focus areas for investments for around six months now. We define Applied AI as anything that mimics human cognition (that’s the AI bit) in an application applied to real-world problems using well-understood technologies. We insist on ‘well-understood technologies’ because as an early stage investor we want to be funding companies that can get products to market in predictable time-frames rather than what I sometimes unkindly refer to as research projects. From our perspective, services like Cloud AutoML are great because they add to the available suite of well-understood technologies and therefore increase the range of startups we can back. This is a trend we can expect to continue as Google, Amazon and Microsoft offer more services in this area as they compete to keep people inside their ecosystems.

Sometimes best practice isn’t best

By | Startup general interest | One Comment

The blogosphere has transformed entrepreneurship. Twenty years ago there were precious few resources available for founders and most either had to find an advisor who had done it before or rely on trial and error. That increased the power of the network effects at the heart of startup hubs where it was easier to find those conversations – especially Silicon Valley. These days it’s all different and a decent guide to doing just about anything is only a few clicks away. Here at Forward Partners we’ve contributed our fair share of such guides at The Path Forward.

However, whilst all this great content is amazingly valuable for entrepreneurs it gives them a new problem. If they try to follow the best practice advised for fundraisingrecruitmentbrand, writing code, product management, and growth then they will quickly run out of hours in their day, at least in the early days when resources are scarce. Moreover, this advice isn’t only coming from the blogosphere, it is coming from investors, board members, mentors, accelerator programmes and friends at other startups, making it harder to deal with than it should be.

I’m not criticising here. Much of the advice is highly-insightful and well-intentioned, and I’ve doled out my fair share (including on this blog). What I am saying is that founders need something more. They need to work out where in their companies they should apply best practice and where they should not.

For most startups these days the first answer is product. If we look at the three biggest startup successes of recent years, Google, Amazon and Facebook, then it’s clear their early success was underpinned by true excellence in product. However, this hasn’t always been the case and isn’t the always the case now. To go back a generation of startups, the success of Oracle and Microsoft came more from being great at sales and partnerships than from being great at product.

Still, for most startups today best practice in product, including customer development and lean development principles, will be critically important. But great product is rarely sufficient on its own, and most successful companies will have an additional spike or two in the other areas listed above.

The task for founders, then, is to first identify the areas where they will excel. That will be determined in part by the market they are in and in part by the experience and capabilities the founding team brings to the table (but don’t make the mistake of focusing where founders are strong if it isn’t right for the market). The second task is to work out the minimum requirement in all the other areas. That’s complicated, and will change as the business matures and standards rise across the board. It’s also something that the blogosphere doesn’t help with.

A simple ‘where we will excel and where we will do just enough’ framework will also be helpful when founders are talking with mentors and advisors. Too often I see mentors frustrated when their advice isn’t actioned and entrepreneurs avoiding mentor conversations for fear of leaving with a list of recommendations they won’t have time to implement. When a startup is just a handful of people it’s ok to be great where it counts and average where it doesn’t matter so much.