Proof isn't persuasion

In the last essay, James Lind found the cure for scurvy in 1747 with two oranges and a lemon a day. The Royal Navy didn’t make citrus standard issue until 1795.

So why did it take 48 years to act on a test that worked?

It wasn’t because the science was weak. Lind was tentative about his own results and, by most accounts, a poor lobbyist. Many naval officers saw scurvy as a discipline problem, not a dietary one. And when the physician Gilbert Blane later pushed for citrus, the Admiralty said lemons were too expensive.

What finally changed it wasn’t a better trial. Blane had ships’ surgeons send him monthly sickness reports, so his case rested on evidence from across the fleet, not one ship. Then, once he had a seat on the Navy’s Sick and Hurt Board, he got a daily ration of lemon juice issued to every sailor.

Proof isn’t persuasion.

That’s the job of the third layer of the engine. Foundation makes the numbers trustworthy. Predictability tells you what works. Optimisation and scale is about getting more from what works - and that turns out to be far more about people than about spreadsheets.

Where I learned to think at scale

LinkedIn is where I learned to think at scale. It taught me to focus relentlessly on what matters, and it’s where I first saw a 10x big bet written into OKRs, alongside the day-to-day goals rather than in a separate innovation deck nobody opens.

I took that and built it into the operating model at Weploy, then at Rescale, and I’ve carried it into every role since.

So I’ve broken this layer into three parts: scale the engine rather than the channel, make the change stick, and leave room for big bets.

1. Scale the engine, not the channel

Once something works, the temptation is to pour money into it until it stops working. And it will stop.

The reason is simple maths. John Dawes of the Ehrenberg-Bass Institute, working with LinkedIn’s B2B Institute, showed that only around 5% of business buyers are in market for many products and services at any one time. The rest won’t buy this quarter, whatever you do. (Dawes himself says not to take the percentages literally. The point is the size of the gap.)

Optimise only for the buyers you can see and you get very efficient at harvesting the 5%, right up until you hit a ceiling. The other 95% are where next year’s pipeline comes from, and they only buy from brands they remember.

That’s why Les Binet and Peter Field’s work for the B2B Institute found B2B businesses grow best with roughly 46% of budget on brand and 54% on activation. Not because the split is magic, but because scaling one end of the funnel starves the other.

For me, optimisation means the whole engine - demand you create, demand you capture, and what happens after the deal closes. A channel that looks brilliant on its own can be quietly borrowing from next year.

2. Make the change stick

Scaling what works means changing how your own teams work. Not in a grand transformation programme, but every day - a new way for marketing to plan a quarter, a new way for BDRs to qualify and follow up an account, a new rule about what gets reported. That’s where most scaling lives or dies, and it’s much harder than it looks from the slide.

Here’s the thing I’ve learned the hard way. All change takes longer than you expect.

First, it takes longer than you’d think for the message to really sink in - the why and the what. Then the what is always harder than the status quo, so cementing new ways of working takes a huge amount of enablement and effort. People don’t resist change because they’re difficult. They resist it because the old way is easier, and they know how to do it.

What makes it stick? Knowing up front that it’ll take twice as long and double the effort, and planning for exactly that. And having objective gates to measure whether it’s really progressing, not just verbal agreement in a meeting. Nodding isn’t adoption.

You’ll often hear that “70% of change programmes fail”. It’s a great line. But when Mark Hughes went looking for the evidence in 2011, he traced five separate sources and found no valid empirical evidence behind the number. Change isn’t doomed. It’s just consistently underestimated - which is a planning problem, and planning problems can be fixed.

Blane, incidentally, did all of this. Evidence people trusted, a champion with authority, and a change built into the routine of every ship rather than left to each captain’s goodwill.

3. Leave room for big bets

The risk with a well-run engine is that it gets very good at small improvements and stops looking for big ones.

Astro Teller, who has led X, Google’s moonshot lab, makes a counter-intuitive point: a 10x improvement can be easier than a 10% one. Aim for 10% and you’re anchored to how things work today. Aim for 10x and you’re forced to rethink the whole approach.

The trick is making big bets part of the system rather than a distraction from it. That’s where OKRs earn their keep. When the 10x bet sits in the same set of objectives as the quarterly targets, it gets time, budget and review, instead of being the thing everyone means to get to.

To balance it all I use a rule I first heard on Shane Parrish’s The Knowledge Project podcast and have followed religiously since: borrow 70% of what you do from what’s already proven, reimagine 20%, and invent only 10%. It has a respectable lineage - in 2005 Eric Schmidt described Google’s version as 70% on the core business, 20% on adjacent projects and 10% on new ones. I like thinking of it as where the ideas come from, not just where the effort goes. Most of what works has already been worked out by someone. Your job is to know which 10% is genuinely yours to invent.

What good looks like

A few golden rules I use to check whether a team is scaling well:

• The whole funnel in one plan. Brand, demand creation and capture, and post-sale growth are planned and measured together.
• Diminishing returns watched. Every scaled channel has a point where the next dollar is tracked, not assumed.
• Change plans doubled. Timelines and enablement are planned at twice the instinctive estimate.
• Objective gates. Adoption is measured by behaviour and results, not agreement in meetings.
• A champion with authority. Every major change has an owner who can actually make it the default.
• A big bet in the OKRs. At least one 10x objective sits alongside the quarterly targets.
• 70/20/10 on purpose. You can say which work is borrowed, reimagined and invented.

For me, yes and no

There’s a fair challenge to the last part. When budgets are tight, aren’t big bets a luxury? Shouldn’t every dollar go to what’s already proven?

I understand the instinct. But in my experience efficiency and ambition aren’t opposites. At Docusign in FY26, sourced revenue per marketing dollar more than doubled.

In my experience, the 10% you invent is often what makes the other 90% cheaper.

Where I’d agree is on size. A big bet in a lean year should be small to start and quick to kill. Ambitious in direction, disciplined in spend.

Where to start on Monday

1. Check your funnel balance. Roughly what share of this quarter’s spend reaches buyers who aren’t in market yet? If it’s close to zero, you’ve found your ceiling.
2. Double one plan. Take a change you’re rolling out, double its timeline and enablement, and add two objective gates.
3. Write one 10x objective. Put it in next quarter’s OKRs, with an owner and a review date.

That’s the end of the series on the three layers - foundation, predictability, and optimisation and scale. I’d love to hear which layer your team finds hardest, because I suspect it’s rarely the one people expect.

Next, I’ll pick up the open questions the model leaves behind, starting with one I’ve learned the hard way: designing your go-to-market around how customers actually buy.