Why are RATs not more popular?

Our core approach to developing novel products and services it to model value from the outset. Such a model starts with huge unknowns, but this nothing to fear: use a tool that copes with uncertainty like the excellent Guesstimate, in which each spreadsheet cell can be a probability distribution.

Armed with such a model you can calculate the expected value of your project which, more than likely, starts negative. You can also plan your next iteration to tackle the greatest uncertainty in the model. As you chip away at uncertainties (Can you do every step of what you offer customers? At acceptable cost/performance? Do your customers understand what you’re offering them? Do they want it?) you reduce the uncertainty in the value and the expected value changes.

A good time to switch to aggressive scaling is when that expected value gets high enough (eg, pick a multiple of the expected development cost).

A colleague pointed out an article about “riskiest assumption tests” with the note “isn’t this what you say?”. It is a nice way of summing up the idea. I was surprised I hadn’t heard the term before, but it seems pretty new to Google too:

So I now have a quick version of my advice:

Don’t waste time on a “minimum viable product” (MVP) or “minimum lovable product”. Do an experiment that does no more than necessary to address the riskiest assumption you’re making — a “riskiest assumption test” (RAT).