Fix the System Not the Creative

Nothing "doesn't work"

Karim Atiyeh took over marketing at Ramp as the company's CTO, and describes treating it as a throughput-engineering problem rather than a creative one.¹ He stepped in at a moment when the lagging indicators looked strong but the leading ones worried him: conversion was already above 50 percent in many segments, monetization had a ceiling, the addressable market was still distant at under 2 percent of the corporate card market, and lead generation was quietly decelerating. Left unaddressed, he believed, "we'd be in trouble today."¹ His diagnosis of the cultural bug in the team was that people would try a channel, watch it fail, and conclude "that channel doesn't work for us."

Atiyeh rejects that conclusion. "It's not that the thing doesn't work, it's that you haven't figured out how to make it work," he says; "all these forms of advertising do work, otherwise other good companies wouldn't be doing it." A failed podcast ad, in his framing, means the wrong partner or the wrong format, not that podcasting is dead.¹ The move is to shift the question from which idea to what is broken in how the idea gets executed.

Attacking the pipeline, not the taste

Notably, Atiyeh does not claim the team needed better creative taste; he says they already had excellent people. Instead he attacked the system that produces the work. The bottleneck he found was that every asset required a brief, reviewed by the brand team roughly weekly, then assigned and produced, so even a ten-minute job took at least two weeks to get in front of anyone. In his words, "with that kind of clock speed it's impossible to get anything done." The fix was to compress the brief-to-asset cycle to about ten minutes using internal tooling that generates on-brand images quickly and removing the review queue. The payoff he names is more shots on goal: a faster cycle allows more experiments and more acceptable risk, "because we know that if it doesn't work out, we could try something else tomorrow."

Asked to diagram the resulting system, Atiyeh gives principles rather than an org chart: fast iteration cycles, explicit experimentation and feedback loops, and author accountability rather than committee. On that last point he is specific: the person with the creative idea owns and ships it, gets credit if it works and is moved off if it repeatedly does not, because "you don't get to ask 10 people and come up with a watered-down idea."¹

Arbitrage in fixed channels

The same engineering instinct leads Atiyeh to look for arbitrage in the cost structure of channels rather than in the creative. His example is a billboard: a new one costs roughly $100,000 and a second roughly $200,000, but changing a board the company already owns costs about $1,000. So he changed a single Union Square billboard seven times in a week, for around $107,000, which drives a story better than two static boards at $200,000 and costs less.¹

Atiyeh concedes limits. "Nothing doesn't work" can rationalize spending on a genuinely poor channel, so the discipline that saves the approach is the measurement loop, and he admits some channels, such as podcasts, get judged comparatively on audience fit rather than on tight attribution. Author accountability also needs a fair definition of how many failures is enough, or it punishes the risk-taking the system is built to encourage. He presents the pattern as an application of the same first-principles process he uses in engineering, pointed at a marketing pipeline instead.