AI Broke the Buy vs. Build Decision
An engineer says they can replace a $100,000 product in a month. These days they’re often right. Building it was never the expensive part.
Build the next version of your company.
You can already see what this has to become. Getting there means decisions across product, systems, marketing and cost at the same time, and everyone advising you has only ever run one of them.
I work with software and SaaS CEOs at $10M to $50M on what the next version has to be, what’s stopping it, and what closing that gap is worth.
Companies are organized into departments. Customers aren’t.
So when growth slows we go looking for the department responsible. Pipeline’s weak, that’s a marketing problem. Deals won’t close, sales problem. It’s a reasonable way to think, and I’ve watched it be wrong more often than right. The expensive problems live in the seams between departments, and nobody owns a seam.
Read five of these and you’ll know what it’s like to have me in the room.
Written from the middle of live engagements. Not frameworks, and nothing to download.
- Note 188 min
- Note 179 min
The first guess is supposed to be wrong
A division full of tiny software products that couldn’t make money. My first explanation was wrong, and being wrong fast was the only reason I found the right one.
- Note 169 min
Buying the Same Revenue Twice
Retained ARR, new ARR, cash and freed-up time are not the same dollar. Sort that out first, then go read the customers you already have.
- Note 157 min
But everybody did their job...
We blame the department showing the bad number. The real cause is usually upstream, in a seam nobody owns, and AI makes it harder to see.
The agency reported 11.7×. The bank statement said $918.
A category-leading brand with a 77-year heritage was steering its spend by an agency dashboard. Reconciled against the bank, one “winning” campaign had made $918 on $600 of spend. The reporting looked fine, so nobody had reason to check it. So we stopped buying more traffic, tuned the store and the email flows, brought marketing in-house and taught their own team to run it with AI. Same traffic, conversion from 0.97% to 2.73%. Every number in it reconciles to their own books, and they run it themselves now.
0.97% → 2.73%
Site conversion after the adjustment, on the same traffic. A leading 77-year-old skincare brand.
5 → 1
Competing visions, then one accountable owner. A leading $100M engineering consultancy.
Five strong leaders. Five good visions. One decision nobody had been asked to make.
One of the leading firms in its field had grown a $3–5M software business on the strength of its consulting. The team was capable and working hard, and leadership could see the output wasn’t matching the effort. Fifteen interviews in, the lowest scores weren’t in engineering at all. They were above the code. Five leaders held five different visions of what the software business was for, so any of them could veto and none could greenlight. The company had never formally decided it was building a software business, because until then it hadn’t needed to. Seven symptoms, one decision underneath all of them.
Where I learned to read across the lines.
Move faster, make more money, get rid of work that shouldn’t exist. Different rooms, same job every time.
- GrowthLoopHead of Engineering through the company’s scale to eight figures. The team tripled; the complexity didn’t.
- Discovery ChannelProduct lead.
- PerformLineFractional product lead. Four people took a new product category from concept to public launch in under seventy days.
- Enterprise automationTen years of workflow and automation software, client-facing from the first day, inside FMC, Dow Chemical, Abbott Labs, GlaxoSmithKline, Towers Perrin, Campbell Soup and TJMaxx.
- AI since 2016Nearly a decade before it became a line item. It’s the instrument in most of this work and the reason for none of it.
I look at what the business actually did before I form a view.
So we can decide what to do next, and what it’s worth.
Ask people what’s holding the company back and you get people-shaped answers: a decision nobody made, a thing nobody owns. Start instead with what the business actually did, the billings, the cohorts, the spend, every lost deal rather than a sample, and the roadmap comes back with numbers on it.
Which matters, because a roadmap without economics is a wish list. The point isn’t the finding. It’s knowing which three things to do first and what each one is worth.
Then your people run it. Every engagement has ended with the client owning a capability they didn’t have: PerformLine’s team still builds on that platform, the skincare brand runs its own marketing, the agency’s designers run the automation. A fix that needs me in the room forever isn’t a fix.
- Where this has to get toThe next version of the company, in your terms, not a framework’s.
- What’s in the wayAcross product, systems, marketing and cost, read from the numbers first.
- What closing it is worthEach move priced, ranked, and honest about confidence.
- Who builds each partYour people where they can. Hands I trust where they can’t. Never me.
Both engagements above are anonymised, one at the client’s request and one at mine. Every number reconciles to the client’s own reporting.
Know where this has to go, and not how?
Twenty minutes. Tell me where you want this company to be in two years and I’ll tell you what I’d go look at first.
Start a conversation
I take a handful of these a quarter.