The rule of thumb most marketers use: a 5:1 ratio of revenue to marketing spend is good, around 3:1 is the floor where marketing justifies itself, and 10:1 is exceptional. Useful as a compass, but the ratio hides more than it shows if you stop there.
ROI is revenue attributable to marketing minus the spend, divided by the spend, and every word of that sentence hides a decision. Attributable how, first touch or last? Revenue or margin? Including the salaries and tools behind the campaign or just media? Pick rules, write them down, and apply them the same way every month. The point is trend and comparability between channels, not the absolute number. Our cost per click calculator helps with the paid side, and how to reduce marketing cost covers what to do when a channel fails the math.
Here is what benchmark posts skip: an increasing share of buying decisions now happens inside AI conversations, where ChatGPT, Gemini, Claude, and Perplexity recommend brands and the buyer never clicks anything trackable. That demand shows up in your analytics as direct traffic or does not show up at all, quietly inflating or deflating every channel's measured ROI. Before you conclude a channel is broken, check whether AI assistants are sending competitors demand you cannot see: the free takes about a minute, and trying to fix your marketing covers the full diagnostic order.
Take a company spending 10,000 dollars a month on paid search that generates 200 leads, of which 30 become customers worth 1,500 dollars in gross margin each. Margin attributable: 45,000 dollars. Fully loaded spend, media plus the tools and the fraction of salary managing it, say 13,000 dollars. ROI is 45,000 minus 13,000, divided by 13,000: about 2.5 to 1, below the classic floor even though the media-only math looked like 4.5 to 1. That gap between media math and fully loaded math is where most inflated ROI claims live.
Now run the same discipline on every channel and one honest wrinkle appears: some customers arrive with no trackable touch because an AI assistant recommended you, and they surface as direct traffic or branded search. If that share is growing, your measured channel ROI is drifting away from reality, and the fix is measuring your recommendation share directly, which is what AI share of voice tracking exists for.
Channel benchmarks are useful as sanity checks and dangerous as targets, so take these with their caveats. Email to an owned list routinely clears 10:1 on media math because the media is nearly free; its honest denominator is the content and tooling time. Organic search compounds toward high ratios over years but looks terrible in its first two quarters; judging it monthly kills it exactly when it is about to pay. Paid search in competitive B2B categories often lives between 2:1 and 4:1 fully loaded, and that can be fine when customer lifetime value is high. Paid social ranges wildly with creative quality, the benchmark there is your own last quarter, not an industry table. And AI visibility is the youngest line: its spend is small, its revenue arrives disguised as direct and branded traffic, and its honest evaluation is share-of-moments trending against pipeline, the method in how to know if your GEO is working. Any benchmark that arrives without its measurement rules attached is a number someone needed for a slide.
Numbers move budgets only when they arrive inside a decision, so package the ratio three ways. For the CFO: fully loaded ROI by channel, trending four quarters, with the measurement rules footnoted, one page, no adjectives. For the growth decision: marginal ROI, what did the last dollar in each channel return, because averages hide saturation and the next dollar is the only one being decided. For the board narrative: blended acquisition cost against customer lifetime value, with the invisible-channel caveat attached so nobody mistakes attribution drift for performance change. Teams that ritualize these three views quarterly stop having ROI arguments and start having allocation meetings, which is the entire point of measuring anything.
No. Mature paid search in a competitive category can be healthy at 3:1 while email and organic often exceed 10:1 because the marginal spend is low. Compare each channel with its own history, not one universal bar.
Margin based ROI is more honest, especially with heavy discounting or fulfillment costs. If you use revenue, at least be consistent and say so next to the number.
AI recommended buyers often arrive as direct traffic or branded search, credit that legacy attribution hands to the wrong channel. Measuring your AI share of voice alongside channel ROI closes that blind spot.
Hold it to a longer clock, not a lower bar: brand work shows up as cheaper acquisition everywhere else, rising branded search, and AI assistants describing you correctly. Measure those, or brand becomes unaccountable.
Give any new channel two to three cycles to find its footing with a defined kill criterion agreed upfront. Channels killed in month one and channels immortal without numbers are the same mistake in opposite directions.
Suspiciously high ROI usually means underinvestment: you are harvesting easy demand and leaving scale on the table. The goal is the highest spend that still clears your bar, not the prettiest ratio.
Marginal decides spend changes; average decides channel existence. Most budget fights dissolve once the two questions stop being answered with the same number.
Fewer than it currently does, usually: two or three proven engines funded to their marginal limit, one experiment at a time, and the discipline to kill or scale the experiment on schedule.