Via Negativa: Improve By Subtraction, Not Addition

Every roadmap meeting ends the same way: someone proposes adding a channel, a tool, or a message. Via negativa asks the harder question first, what should you remove?

Quick Summary

• Via negativa, Nassim Taleb's term for 'the negative way', says removing what's harmful is more reliable than adding what's beneficial.
• Negative knowledge (this doesn't work) is more durable than positive knowledge (this works), it rarely gets overturned by new data.
• Subtraction is lower-risk than addition in uncertain systems: you're closing a known leak, not betting on an unproven idea.
• P&G cut its brand portfolio from roughly 170 brands to about 65, and the brands it kept accounted for 90% of sales and 95% of profit.
• A simple audit-and-cut framework beats another brainstorm for new tactics, most teams have more to gain from stopping than starting.

The Principle: Knowledge Grows Faster By Subtraction

Taleb's core claim in Antifragile is unglamorous but hard to argue with: we are much more certain about what is wrong than what is right.

If a campaign loses money for six straight months, that's settled. You know it doesn't work, and no future data point casually reverses that.

If a campaign made money last quarter, that's a hypothesis, not a fact. Markets shift, algorithms change, and audiences move on, so 'this works' has a shelf life that 'this doesn't work' does not.

That asymmetry is why Taleb calls negative knowledge more robust. Via negativa in practice means acting on what you're already sure of (kill the underperformer) instead of chasing what you merely hope is true (launch the untested idea).

Marketing teams do the opposite by default. Adding a channel feels like progress, cutting one feels like admitting defeat, and that emotional bias is exactly what makes this principle worth naming out loud.

Why Subtraction Beats Addition When You're Uncertain

Addition and subtraction carry very different risk profiles, and the difference matters most when you can't be sure what will happen next.

Adding something new is a bet on an unknown. A new paid channel, a new tool, a new messaging angle, all of these might work, might do nothing, or might quietly cannibalize what already works.

Removing something bad is a bet on a known. You already have the data, the underperforming channel has been losing money or attention for months, and stopping it locks in a gain you can already see.

There's a compounding effect too. Every tool, channel, or message you keep adds coordination cost, someone has to maintain it, report on it, and defend its budget line at review time.

A bloated martech stack is the clearest version of this. Enterprise teams juggling dozens of overlapping platforms routinely find that rationalizing the stack, not expanding it, is what unlocks growth, because fewer, better-integrated tools beat more disconnected ones.

Subtraction doesn't guarantee a win, cutting the wrong thing still costs you. But its downside is capped and knowable, while addition's downside is open-ended, and that's the whole case for defaulting to the cut.

The Audit-and-Cut Framework

Via negativa needs a process, otherwise it's just a slogan. Run this quarterly, not as a one-off cleanup.

1. List everything currently running. Every channel, campaign, recurring content series, and paid tool, no exceptions for "it's always been there."
2. Score each one on evidence, not vibes. Does it have six-plus months of data showing a clear result, or is it running on a hunch?
3. Cut anything with a settled negative result first. These are your highest-confidence removals, don't wait for a "better time."
4. Flag anything unmeasured as a removal candidate. If you can't tell whether it's working, that's itself a reason to question keeping it.
5. Reinvest the freed budget in what's already proven, not automatically in something new.

Subtraction Wins You Can Actually Point To

P&G's 2014 brand cut is the textbook case. The company moved to divest or discontinue more than half its brands, keeping roughly 65-80 core names like Tide and Pampers, which delivered 90% of sales and 95% of profit across the prior three years.

That's the asymmetry in one data point: the removed brands weren't adding much, and cutting them freed focus and budget for what was already winning. P&G ran the same logic on SKUs in 2023, its CFO noted that the bottom 25% of SKUs contribute very little to absolute retail sales, so slimming the shelf was framed as a growth move, not just a cost cut.

Checkout funnels tell the same story at a smaller scale. Baymard Institute's long-running usability research finds that the average large ecommerce site could lift conversion by roughly 35% purely through checkout UX fixes, most of them subtractive: fewer form fields, dropping forced account creation, removing unnecessary steps. Nobody added a persuasive new element to get that lift, they took friction away.

The pattern repeats everywhere: growth from removing a bloated tool stack, a redundant channel, or an extra form field, arrives faster and more predictably than growth from adding one more thing to the pile.

Key Takeaways

• Via negativa means improving by removing what's confirmed harmful, not by adding what's merely hoped to help.
• Negative knowledge (this doesn't work) is more durable than positive knowledge (this works), so cuts based on settled data are lower-risk bets.
• Subtraction has a capped, knowable downside, addition's downside is open-ended, that asymmetry is the whole argument for defaulting to the cut.
• Run a quarterly audit-and-cut pass: list everything running, score it on evidence, cut settled underperformers first, reinvest in what's proven.
• Real wins back this up, P&G's brand and SKU cuts, and checkout UX research showing ~35% of conversion sitting in friction you can simply remove.