That question sounds like boardroom philosophy, not product management. It isn’t. Once you know what you’re centered on, a surprising number of "should we do this?" debates stop being debates.
Why "more options" isn’t the answer anymore
The usual advice to founders is to build optionality: try more things, hedge more bets, diversify the portfolio so no single failure sinks you. McGrath argues that this old playbook is wearing out. Markets are fragmenting into smaller, more personalized segments, and even a stable, well-known product can face a shifting audience underneath it. She points to Tide detergent — the product hasn’t changed, but P&G has had to explain fabric softener to younger consumers as "hair conditioner for clothing," because the market around a supposedly universal product quietly split into different audiences with different vocabularies.
For a startup, the equivalent moment is smaller but just as disorienting: a feature that worked with your first 50 users doesn’t resonate with the next 500, and it’s unclear whether that’s a targeting problem, a product problem, or just noise. Without a clear center, every one of those moments turns into a fresh argument. With one, it turns into a filter.
What a "center" actually is
A center, in McGrath’s framing, is the organizing logic that decides where resources go, what gets killed, and which opportunities are distractions dressed up as growth. It isn’t a mission statement bolted onto a wall. It’s closer to a company’s own definition of strategy — the set of choices about where it will compete and how it intends to win.
A center can be a customer problem you exist to solve, a technical capability you’re better at than anyone else, a specific kind of friction you remove, or a community you serve. McGrath’s example is Novartis: when Vas Narasimhan became CEO, the company spanned pharmaceuticals, consumer health, eyecare, and generics, on the theory that diversification would cushion patent expirations. Narasimhan decided that theory was false — what had actually carried the company through hard years was its science — and refocused Novartis on innovative medicines, selling or spinning off what didn’t fit. That doesn’t mean every diversified company is making a mistake; it means one specific company found that its historical hedge wasn’t doing the job it was supposed to do, and defining a sharper center made its next round of hard calls faster.
That’s the practical payoff for a product team: a center doesn’t tell you which idea will win. It tells you which ideas are even worth arguing about.
What the center changes in a product decision
The same weak signal — a spike in a niche use case, a slowdown in signups, an unsolicited partnership offer — can lead to opposite decisions depending on whether it touches your center or sits outside it.
| Signal or opportunity | If it fits the center | If it doesn’t fit the center |
|---|---|---|
| Unexpected demand from a new segment | Worth a small, fast test to see if it deepens your core problem-solving | Worth noting, but not worth diverting a roadmap |
| A competitor’s AI feature | Ask how AI could reinforce what you already do best | Resist copying just to keep pace |
| An investor or partner pushing a new product line | Explore only if it strengthens the core capability | Treat as a distraction, however attractive the check |
| A metric drifting from plan | Investigate — it may be telling you the market has changed | Still investigate, but weigh less heavily against core bets |
The table isn’t a formula; it’s a habit. The center doesn’t remove judgment — it gives judgment something to check itself against.
Weak signals are clues, not verdicts
Here’s where a lot of teams go wrong in the opposite direction: they treat any early signal — a few enthusiastic users, one viral post, a single big customer request — as proof they’ve found something real, and they overcommit before the signal has had time to firm up.
Research on weak signals in other uncertain sectors is useful here, even outside product management. A study on energy-sector foresight defines a weak signal as an early, ambiguous indicator of possible change that lacks broad awareness and carries low predictive certainty — its value lies not in forecasting precisely, but in broadening what decision-makers consider possible. Firms with more disciplined signal-watching practices have shown meaningfully better long-term financial performance in longitudinal research, though that link is associative, not a guarantee tied to any individual company’s next move. The point isn’t that weak signals are useless. It’s that they’re inputs for a hypothesis, not a verdict on one.
flowchart TD
A[Weak signal observed] --> B[Check against the center]
B --> C[Assign confidence: noise or worth testing]
C --> D[Run a small test]
D --> E{Result}
E --> F[Iterate or retest]
E --> G[Revise hypothesis or stop]
Confidence here is a judgment call informed by evidence, not a statistical score — the discipline is in writing it down and revisiting it, not in treating it as proof.
AI belongs after the center, not before
McGrath is direct about a common mistake: leaders "starting with AI and figuring out how to use it," when they’d be better served starting with strategy and backing into where AI fits. For a startup, that means asking what your center requires — faster discovery, less friction, better matching — before asking which AI tool to bolt on. AI is genuinely useful for scanning more signals than a small team could track alone, but it doesn’t replace the judgment of deciding what those signals mean for you specifically.
Write it down: a decision log that outlives the meeting
Teams don’t need a heavier process — they need a lighter, honest record. A simple log with five fields does most of the work: the belief you’re testing, the signal you observed, its possible implication, the next test you’ll run, and whether you’re continuing, retesting, redirecting, or stopping. Not every deviation from plan belongs in the "failure" column; sometimes it’s just the market telling you something you didn’t know.
The center doesn’t predict the future — it organizes your response to it
None of this eliminates uncertainty, and it shouldn’t pretend to. A clear center won’t tell you which idea succeeds, and a weak signal won’t tell you what’s coming next with any precision. What both give you is a faster, less personal way to decide what deserves another test, what needs rethinking, and what no longer belongs — which, for a team drowning in plausible ideas, may be the only kind of clarity that’s actually available.
