01 Overview
Display advertising is often judged by a single blended number: the average cost per click. That number looks reassuringly small, and it is easy to build a budget around it. The trouble is that an average hides the shape of the spend. Some clicks are genuinely new visits that would not have occurred otherwise, and some are clicks from people who were already on their way to the site. When those two groups are blended together, the reported cost per click can look like a bargain while the real cost of buying additional behavior is quietly much higher.
The 36-cent problem is a way of naming that gap. It is not a claim about any particular advertiser's account or a promise about what any campaign will pay. It is a reminder that the marginal click, the next one you buy, is usually more expensive than the clicks you already bought. Early clicks come from the most responsive people in the cheapest inventory. Later clicks come from harder-to-convince people in more contested placements. If you plan against the average and then scale, the average stops describing what you are actually paying.
This article is about reading display cost metrics honestly. It covers what the metric really measures, how it responds to the levers you control, how to avoid fooling yourself with a flattering average, rough benchmarks for sanity checks, and what happens when you try to scale a format that looked cheap at low volume.
02 What this metric really measures
Cost per click in display is usually calculated by dividing total spend by total clicks. That is a clean arithmetic operation, but it answers a narrower question than most people assume. It tells you what you paid, on average, for every click the system recorded. It does not tell you which of those clicks were caused by the ad and which would have happened anyway. A person who sees a banner, ignores it, and later searches for your brand and clicks an organic result is not the same as a person who saw the banner and clicked through immediately. Blended cost per click treats both as equivalent if the click is attributed to the campaign.
The more useful concept is incremental cost per click, sometimes called marginal cost per click. This asks a counterfactual question: if we spent a little more, how many additional clicks would we get, and what would each of those cost? That figure is almost always higher than the blended average, because the cheapest and most responsive audience is consumed first. The gap between the two is the leak. It is not fraud and it is not necessarily waste; it is the natural consequence of buying the easiest conversions first and then assuming the rest will behave the same way.
A related distinction is between clicks and view-through or assisted actions. Display often contributes to a later action without receiving a click at all. If your measurement only counts clicks, you may understate display's role. If your measurement counts every later action as a display win, you may overstate it. The metric is only as honest as the attribution window and comparison group behind it.
03 How it moves with your levers
Creative relevance is the first lever. A display unit that matches the viewer's current context and intent earns more genuine clicks per impression, which lowers the cost of each real click. Stale creative, by contrast, gets ignored, and the clicks that remain skew toward accidental or low-intent taps. Refreshing creative on a regular cadence keeps the audience from developing banner blindness toward your specific units. The lever is not more impressions; it is better-matched impressions.
Audience and placement quality is the second lever. Broad reach buys cheap impressions, but cheap impressions are cheap for a reason. Narrowing to placements and audience segments with demonstrated interest raises the click rate and usually raises the cost per thousand impressions at the same time. The net effect on cost per incremental click depends on which force is stronger. Frequency capping is a third lever: showing the same person the same message repeatedly produces diminishing returns and can inflate reported clicks from a small group while adding little incremental behavior. Capping protects the marginal cost from drifting upward.
Bid and pacing strategy is the fourth lever. Aggressive bidding on the cheapest inventory can win a lot of low-value clicks. Careful pacing and value-based bidding tend to raise the blended cost per click while lowering the cost per incremental outcome. Finally, the post-click experience matters. A slow or confusing landing page turns a paid click into a bounce, which makes the click look expensive in hindsight even if it was cheap at auction. Measurement window and attribution model sit underneath all of these, because they determine which clicks get counted at all.
04 Reading the number without fooling yourself
The most common misread is treating a low average cost per click as proof of efficient acquisition. A low average can mean you are buying a lot of clicks from people who were already going to convert, or from placements that generate accidental taps. The average is a description of the past, not a forecast of the next dollar. Before you celebrate a cheap cost per click, ask what share of those clicks came from users who had already been exposed to your brand, and what share came from new-to-brand audiences. If the cheap clicks are concentrated in the already-exposed group, the number is flattering but not incremental.
A second misread is comparing cost per click across formats or campaigns as if the units were identical. A click on a display banner and a click on a search result are different behaviors with different intent levels. A display click may be a curious tap; a search click is often an explicit request. Treating them as the same unit of value leads to budget decisions that look data-driven but are actually apples-to-oranges. The fix is to compare cost per incremental outcome, not cost per click, whenever the outcome definitions differ.
A third misread is ignoring the denominator. Clicks can rise simply because impressions rose, without any improvement in efficiency. Always pair cost per click with click-through rate and with a measure of incremental reach. If click-through rate is flat and impressions doubled, the cost per click may look stable while the marginal value of each additional impression is falling. Reading the metric in isolation is how leaks stay hidden.
05 Benchmarks to sanity-check against
Display cost per click varies enormously by market, placement quality, audience definition, and season. Rough directional ranges are more useful than precise targets. In many contexts, blended display cost per click lands somewhere in the low single-digit dollars, and incremental cost per click can be a meaningful multiple of that blended figure. Treat any specific number, including the 36-cent framing, as an illustration of a shape rather than a target you should expect to hit. Your own historical marginal cost is a better benchmark than any published range.
A practical sanity check is to compare cost per click against cost per thousand impressions and click-through rate. If cost per thousand impressions is stable and click-through rate is falling, cost per click will rise even though nothing about your buying changed. If click-through rate is rising while cost per thousand impressions rises faster, cost per click may rise while efficiency improves. The three numbers move together, and reading them as a set prevents single-metric tunnel vision.
Another check is to segment the average. Break cost per click down by placement, by audience segment, and by frequency bucket. You will usually find that a small share of inventory produces most of the cheap clicks and a large share produces expensive ones. The blended average sits between them and describes neither. Segment-level ranges are rough, but they reveal whether your cheap clicks are concentrated in a place you actually want to buy. Benchmarks are a starting point for questions, not a verdict.
06 Scaling what works
Scaling a display campaign changes the economics of the clicks you buy. At low spend, you can cherry-pick the most responsive audiences and the cheapest quality placements. As budget rises, you exhaust that pool and move into less responsive audiences and more contested inventory. Marginal cost per click tends to climb, and it can climb quickly once you pass the point where the cheapest supply is gone. This is not a failure of the campaign; it is a normal property of auction-based buying. The mistake is assuming the average you measured at low spend will hold at high spend.
To scale without losing the plot, watch the marginal cost rather than the blended average. Increase budget in steps and measure the cost of the additional clicks each step produces. When the marginal cost approaches the value of the outcome, you have found the practical ceiling for that creative and audience combination. At that point, the way to keep growing is to change the inputs: new creative concepts, new audience segments, new placements, or new geographies. Expanding the supply of responsive inventory is what resets the marginal cost curve.
Frequency management becomes more important at scale, not less. A larger budget spent against a fixed audience means more impressions per person, which raises the risk of fatigue and of counting the same responsive users repeatedly. Capping frequency and rotating creative keeps the incremental pool fresh. Finally, revisit the measurement window as you scale. Longer windows tend to capture more assisted actions and can make a campaign look more efficient than it is. Keep the comparison group honest so that growth in reported clicks reflects growth in real behavior.
07 Frequently asked questions
Is a low cost per click always good? No. A low average cost per click can reflect a high share of clicks from people who would have converted anyway, or from placements that generate accidental taps. The number is only good if the clicks are incremental and lead to outcomes you value. Pair it with new-to-brand reach and post-click behavior before drawing conclusions.
Why does my cost per click rise when I increase budget? Because you are buying deeper into the auction. The first dollars reach the most responsive people in the cheapest quality inventory. Additional dollars reach less responsive people and more contested placements. Rising marginal cost is expected; the question is whether the additional clicks are still worth their price.
How do I measure incremental cost per click? Compare a group exposed to the campaign with a comparable group that was not, or pause and restart spend in controlled steps and observe the change in clicks. The difference in clicks divided by the difference in spend approximates the marginal cost. It is an estimate, not a precise figure, but it is far more honest than the blended average.
Should I optimize to clicks at all? Clicks are a useful intermediate signal, but they are not the outcome. Optimize toward the action you actually care about, and use clicks as a diagnostic. If click quality is poor, a low cost per click is a warning sign rather than a win.
How often should I refresh creative? There is no universal cadence. Watch click-through rate and frequency together. When click-through rate declines at stable frequency, or frequency climbs at stable click-through rate, it is time to rotate creative and re-examine the audience.