Innovation.

The framework

1. 01

   Seventy per cent of the money, ten per cent of the return — and both numbers are right.

   The instinct on reading the inversion is to move money. Don't, or not much. Core innovation returns least per dollar because it's the cheapest, safest and most certain work in the portfolio — that's one fact stated twice, not two problems. What the inversion actually tells you is that the bands can't be compared at all: the core is bought at a known price for a known return, and the tenth is bought as an option on a return nobody can size yet. A portfolio that only funds what it can price has no options in it, and options are where the seventy per cent came from.

2. 02

   Seventy, twenty, ten is an average, not an instruction.

   Nagji and Tuff were explicit that it isn't a magic formula. It's a cross-industry average, and the variation around it is wide: consumer goods firms in their data sat closer to 80/18/2, mid-stage technology firms closer to 45/40/15. Competitive position moves it again — a business that's behind has more reason to bet big, a leader has more to protect. So the exercise isn't adopting the number. It's naming the ratio your industry and your stage justify, measuring how far your actual spend sits from it, and closing the gap on purpose rather than by drift.

3. 03

   Where does the transformational tenth get its money?

   Not from a levy on everyone else. The common approach — asking every business unit to contribute a slice of its budget to innovation — reliably backfires, because units don't experience it as supporting a good cause. They experience it as head office taking five per cent, and your innovation team becomes the villain of every budget meeting it isn't in. Core and adjacent work can sit inside the relevant unit's P&L, which is where it belongs and where somebody will fight for it. The tenth needs a different pocket: funded above the annual cycle, by someone senior enough that a soft quarter can't quietly empty it.

4. 04

   Write the hurdle before the first review, or the core's spreadsheet writes it for you.

   You already know the principle — an early bet is judged on what it settles, not what it returns. The work is writing the hurdle down in advance, because whatever you leave unspecified, the core's spreadsheet will specify. Net present value needs adoption rates and price points, and those need customers who can say what they'd pay — precisely what a transformational bet doesn't have. So name the question this bet exists to answer, the date it answers by, and what a clean no looks like. Google, they note, assessed transformational work on whether the company was likely to learn from it rather than earn from it.