How VCs See Your KPIs
nfx.com
nfx.com
Nope.
Daily / weekly / monthly plain text email or emailed spreadsheet is way better. The problem with dashboards is that when the underlying numbers change, they can change the figures for the past and there won't be a record of it. If it's in an email someone can bring up this discrepancy.
Also, lots of startups waste a lot of time building a dashboard not realizing that they're now building two apps. "We don't need login, we'll just IP whitelist it." Then a VC wants to see the numbers. Then he wants to forward it to his partners. Now if your dashboard goes down the CEO starts freaking out about how it makes the company look during their raise.
The rest of the advice is spot on though. Especially keeping the number of KPIs down.
This isn't to rule out pulling in other sets of numbers, but those should be done by a Data Analyst and they should be one-time analyses that can be re-run by the analyst later if needed.
> Also, lots of startups waste a lot of time building a dashboard not realizing that they're now building two apps.
Setting up a Grafana dashboard is a lot easier than setting up crons that send emails. Auth is built in, including OAuth.
It's also trivial to have your data adjust to time bounds that you can change on the interface.
> when the underlying numbers change, they can change the figures for the past and there won't be a record of it. Any important numbers should be graphed on a time series chart, and you won't have to track down emails and compare across different tabs.
Not really. It would take me about 5 minutes to setup an extra worker that sent emails for an existing app.
Plus, it's a strategic decision to share numbers with a potential funder. That decision should be made each time, rather than allow ongoing analysis.
> Any important numbers should be graphed on a time series chart, and you won't have to track down emails and compare across different tabs.
I agree that important numbers should be graphed on a time series chart, what I'm saying is that sometimes changes in code can change numbers in the past. If each email has that month's spreadsheet, complete with numbers going into the past, and someone says "hey wait a minute, I thought churn in July 2017 was 4.5%, I remember because it was our worst month!" Then you can go back to see what the email said.
The author doesn't seem to recognize that KPIs would be subject to Goodhart's Law. ("When a measure becomes a target, it ceases to be a good measure.")
I am willing to bet that something like "retention" or "net new accounts" are relatively resistant to GL.
https://money.cnn.com/2017/08/31/investing/wells-fargo-fake-...
I've been in tech 14 years, and I've never worked anywhere where a manager discussed KPIs with people on my level. But the companies I worked for also had billions in the bank, so maybe this just implies that companies of that size don't need engineers to look at KPIs.
Get rid of the stupid TV dashboards and put up a decent Monet. It sucks working in a physical manifestation of a computer.
Indeed. Things are a little different at a startup when there are ten of you sitting in one big room and the business lives or dies based on what each and every person does in that room.
For some reason, this line put a bad taste in my mouth. Probably an association with too many bad advice articles on the internet where the author felt the need to establish authority on the subject using a line like this rather than letting the article stand on its own merits.
> If they don’t have a daily report or tell me it's going to take a few days, the conversation is over.
So he's a big fan of having the company KPIs displayed on monitors throughout the office, but if anyone actually uses those monitors to get their info on KPIs then the conversation is over. Got it.
Whether that's worth the half day to setup is a different question.
(1) To go to the trouble to measure something and publicize it to employees, investors, etc., first need a good reason for measuring that something. E.g., why do we measure MPG, MPH, engine temperature, time since last oil change, ..., unemployment rate, labor participation rate, economic growth rate, inflation rate, prime rate, money supply, balance of payments, mortgage default rate, federal budget deficit/surplus? For each of these, and jointly for several of them, we have good reasons, from owning a car to macroeconomics.
(2) When we seek to measure something, we should be clear on what in principle we are trying to measure and, then, pay due attention to the accuracy of our measurements, e.g., reliability and validity.
The OP ignored (1) and (2).
My wife, Ph.D. in mathematical sociology, understood reliability and validity and much more. Similarly for my brother, Ph.D. in political science. The social sciences are such indefinite fields that they have to be careful about what they are measuring, why, what the measurements mean, and the accuracy.
As a graduate student, Ph.D. in applied math, I understood reliability and validity, and at least several, likely nearly all, my fellow graduate students and faculty did also.
The OP is by a VC who wants "warm introductions". Well, likely the people he knows don't know about reliability or validity, and the people I know do. So, we won't have colleagues in common with no possibility of a "warm introduction" in either direction. E.g., one of my dissertation advisors was later President at CMU. I'm sure he knew about reliability and validity, but it sounds like we couldn't get a "warm introduction" via him in either direction. The Chair of my Ph.D. orals committee was a Member, US National Academy of Engineering, and I'm sure he knew about reliability and validity; same problem, not much chance of an introduction!
Uh, the author of the OP is interested in "huge ideas". Hmm .... From people who don't know about reliability and validity? How about measure preserving in ergodic theory and applications to machine learning and computer security?
My startup is supposed to be of high interest to nearly everyone on the Internet -- "huge" enough? But to evaluate the work, need to dig into and evaluate the crucial, core, original applied math secret sauce based on some advanced pure math prerequisites -- the OP author is qualified to evaluate such work?
In what sense is he really interested in and able to evaluate or work with "huge ideas"?
Or, if he invests and we have our first BoD meeting and his big contribution is to insist that we allocate time, money, and effort to lining the walls with big video screens with "key performance indicators" changing in real time, uh, to "motivate" the employees? Uh, might we have a study of the effect on the employees, motivate them, distract them, or irritate them? My a priori guess -- at best another case of the Hawthorne effect