To track your progress (and raise your round)
If you are building (and fundraising) for a company with deeper technology, it is likely that you will not have the “traditional” metrics by which investors would evaluate a more traditional startup at the various investment stages.
By traditional metrics, I am referring to things like revenue, GMV, MRR, churn, AVC, etc. If you approach an investor for a Series A investment for an e-commerce play, she will have enough data points from having seen dozens of other pitches to know what range of GMV and CAC/LTV she would like to see for you to be “Series A” ready.
If, however, you’re building, say, a new proprietary technology, data set or computational model that brings this data to life, it will, in my experience, always take you longer to get the data, tune the data and prove that the data adds value. Your customer Proof of Concepts might last longer, the size of your technical team might have to be deeper, if you’re building IP, your legal bills for patent filings might be larger.
It’s unlikely therefore, that you will have the traditional metrics an investor will traditionally use to map traction.
It should be your job, therefore, to define what metrics are most relevant for what you are building, and to anchor progress against those, to continuously report against them to your team, existing investors and Board, and to then be able to prove to future investors that the metrics you promised at Seed you are on track to achieve by Series A (or, as is usually the case … to that Seed extension).
Biotech provides an analogy for this: Series A rounds in Biotech are quite big relative to “tech” and yet the technical risk at Series A is still very, very high. But because there is a “pipeline” for therapeutic discovery and testing, that anchors progress, the progress through that pipeline becomes the metric, allowing investors to track progress, even while accepting significant technical risk.
To draw another example of an anchor point, the Imagenet competition, which measures the ability for a machine learning algorithm to identify objects in pictures, helped define the framework of “progress” of the various competing teams. No team before 2012 had achieved greater than 75% accuracy until the team of Geoffrey Hinton, Ilya Sutskever, and Alex Krizhevsky applied their deep-learning algorithm. This milestone metric unleashed the Cambrian explosion of innovation and funding around deep learning. Defining the 25% metric as a milestone would have been the relevant anchor.
To bring it closer to home, and with their permission, I draw on two real world examples that I have been close to:
Aire uses proprietary and third party data combined with machine learning to create a modern credit bureau which predicts financial risk at the individual level where backwards-looking incumbent bureaus cannot. Because @airelabs clients (lenders such as credit card companies, lease providers, point of sale providers) need to get comfortable that the Aire score correctly predicts individual lender risk before issuing real credit, they need to test the data with a small subset over a long period of time. On average, they need six months to see if a new credit card holder defaults or does not at the rate that Aire predicted. It is after this long POC process that clients will open up broader parts of their customer base onto Aire’s algorithm accelerating that traditional metric, revenue. Aneesh, the founder of Aire, therefore, had to create other metrics to anchor progress for clients, existing investors, like me, and future investors. The first milestone he anchored was a regulartory one. He spent significant time and resources to gain FCA approval in the UK as the first regulated credit bureau since the late 1990s. The FCA approval provided not only a defensible moat, but also validation that the nation’s financial regulator felt Aire’s technology was accurate enough to be deployed with real citizens seeking real financial loans. To then prove the model, he anchored around a second metric: AUC (Area Under the Curve) a well understood metric in risk modelling circles, which indicated how accurate a model is in predicting behaviour. His clients understood AUC… and after a few tries, he finally got me and the rest of the Board to understand it as well. He proved his algorithm worked by backward testing AUC against $10 Billion worth of real-world financials loans, which allowed him to prove to clients (and investors) that, given the data, he could tune the data and prove the model.
Aneesh anchored forward momentum around regulation and an industry-accepted metric that proved that his data models worked against real-world lending books, which is what allowed him to raise his Series B earlier this year as more traditional metrics like revenue or MRR ramped up and his sales pipeline accelerated its conversion.
Another example is Wefarm, a London-based startup that is helping to digitise 80% of the world’s food supply chain, which primarily comes from single-farm farmers in emerging markets. These farmers don’t have smartphones, laptops or GPS enabled tractors… but they do have phones with SMS. Kenny, the founder, set out to connect, digitise and help these millions of farmers starting with a Quora-like Q&A farmer-to-farmer network (with advanced NLP in the middle to digitize and log the questions and answers in their local dialect). From a social network perspective, he has seen insanely impressive growth in his ability to enlist farmers, starting in Kenya and Uganda. In a year he grew by over 1 million farmers posing and answering questions about their crops and farm animals. He now has direct engagement with 25% of the farmers in those countries and statistically has built the ground-truth data source of who is planting what, their yields, virus outbreaks, pesticide used, etc. This data, at scale, is a gold-mine to the whole supply chain (and the individual farmer).
What he does not have yet is a large revenue stream. He had purposely chose not to sell the data or layer on commerce until he had network density. This was never going to be an investment for a VC who wanted to purely hear about GMV or MRR. Instead, Kenny anchored progress around two metrics: usage growth, which he nailed, and then messaging density. To any VC that’s evaluated a dating app, chatter (messages sent and responded to, ie flirting) is one of the healthiest signals of an engaged community. In the same way, Kenny was using questions asked and answered as signals of the health and value of his network to his users. Anchoring around his network density and user engagement (and vision) allowed Wefarm to raise a Series A led by True Ventures. Late last year, WeFarm announced that (with users in only two countries in Africa) “Wefarm now shares more content than Stack Overflow and has more content contributors than Wikipedia.”
As these examples show, if your vision requires that it take longer to build, tune and prove the data and the algorithms, define and anchor that metric early, and have everyone on your team and on your cap table understand it intimately to allow you to track progress against it and to build momentum for the next time you meet an investor for the next round.
Investors should (or I would argue must) be willing to learn and accept new metrics if they are to agree with you that achieving your long term objectives will create something of significantly greater enterprise value than purely CAC/LTV.