Benjamin Graham, the father of value investing and Warren Buffett's teacher, built his entire philosophy around one idea: never buy a stock at a price that only works out if you are right. He called the gap between what you pay and what the asset is actually worth a margin of safety, buying at a discount steep enough to absorb bad luck, bad timing, or your own bad math.
Marketers plan the opposite way. We build a campaign model on our best-guess conversion rate, our best-guess CAC, our best-guess launch date, and then commit the whole budget to that single number. Graham would call that reckless.
What margin of safety actually means
A margin of safety is not "add 10% contingency" as a token gesture. It is a deliberate acknowledgment that your forecast is a guess dressed up as a number.
Graham's own rule of thumb was concrete: don't pay more than 15 times earnings or 1.5 times book value, a discount built in specifically so "human error, bad luck, or extreme volatility" wouldn't wipe you out (TradeAlgo). Translate that into marketing: don't commit 100% of a launch budget to a channel mix that only breaks even if your projected CPA holds exactly.
Gartner's 2025 CMO Spend Survey found marketing budgets flatlined at 7.7% of company revenue, and 59% of CMOs still say they lack enough budget to execute their strategy (Gartner). Tight budgets and zero slack are a dangerous combination, because there's no room left to absorb the surprise.
That's the setup. Here's where it actually bites you.
Where marketing plans skip the discount
Three places margin of safety usually gets skipped entirely:
- Budget: the media plan assumes today's CPC holds for the whole quarter, even though average Google Ads CPC jumped roughly 13% year-over-year in 2025 and rose across 87% of industries tracked (Search Engine Land).
- Timeline: the launch date is set from the "everything goes right" version of the project plan, with no buffer for a legal review, a creative reshoot, or a platform API change.
- Forecast: the revenue target assumes last quarter's conversion rate repeats exactly, with no room for seasonality or a competitor's price cut.
Each of these is a single point of failure disguised as a plan. Fix even one and the whole campaign gets sturdier.
A plan that only works if every assumption holds isn't a plan, it's a bet. Margin of safety is what turns a bet back into a plan.
Building the discount in
You don't need Graham's exact math, you need his instinct: price in the ways you could be wrong before you commit.
- Budget in bands, not points. Model your CAC forecast as a range (say, $40-$65) and size the campaign so it still clears a minimally acceptable ROI at the pessimistic end, not just the midpoint.
- Stagger the spend. Commit 30% of budget to validate assumptions, hold the rest until early signal confirms the plan, rather than firing the full budget on day one.
- Pad the timeline where you have the least control. Legal, procurement, and third-party integrations are where delays actually happen; build buffer there, not in the parts you control.
- Keep an "if I'm wrong" line item. A reserved 10-15% of budget earmarked specifically for correcting course, not spent by default.
None of this is pessimism. It's the same math that let Graham keep buying through downturns other investors couldn't survive.
A team plans a $200,000 quarterly launch assuming a $50 CAC, which would deliver 4,000 customers. Building in a margin of safety, they model the pessimistic case at $70 CAC (a 40% miss) and check the plan still clears breakeven at 2,850 customers. It does, barely. They commit $140,000 upfront and hold $60,000 back, releasing it only once the first two weeks of real CAC data confirm the campaign is tracking toward the optimistic end rather than the pessimistic one. When a platform algorithm change pushes real CAC to $68 in week three, the plan absorbs it instead of blowing the quarter's budget in the first month.
Why having a plan isn't the same as using one
Here's the part that should worry you more than any single bad forecast: most marketing teams already know they need a buffer, and build one anyway that never gets used. A Gartner survey found 81% of marketing leaders have contingency plans in response to disruptions, but only 21% of them actually follow those plans when disruption hits (Gartner).
That gap is the real failure, not the absence of planning. A margin of safety that lives in a slide deck and gets overridden the moment a VP wants the full budget spent faster isn't a margin of safety, it's decoration.
A few reasons the 79% don't follow their own plan:
- The reserved budget gets reallocated under pressure the moment a quarter looks soft, because "unused" money reads as available money to whoever controls the spreadsheet.
- Nobody owns the trigger for when to use it. A buffer with no named owner and no predefined trigger condition just sits there until someone spends it on something else.
- Using the buffer feels like admitting the original plan was wrong, so teams delay pulling it until the damage is already done.
The fix isn't a better contingency plan, it's a pre-agreed trigger: name the specific metric, the specific threshold, and the specific person authorized to release the reserved budget before you ever need it.
Why this compounds
A team that always plans to the edge of its assumptions eventually gets burned by one wrong assumption at the worst possible time, a platform algorithm change, a category-wide CPC spike, a slipped launch date that collides with a competitor's. A team with margin of safety absorbs the same shock as a bad quarter, not a bad year.
The discipline isn't glamorous. But neither was Graham's, and it's the reason his students outlasted the traders who bet the whole position on being right.
Next time you build a plan, ask the Graham question before you ask the growth question: how wrong can I be and still be fine?