01 Overview
I spent years treating display as a default line item. It was easy to buy, easy to report on, and easy to defend in a deck because the impression counts looked impressive. What changed my mind was not a single bad campaign. It was the slow realization that most of what I was paying for never had a real chance to be seen, and the small slice that was seen was often seen by people who had no reason to care. The format itself is not useless, but the way it is commonly bought makes it a poor fit for most direct-response goals.
So I stopped buying display in the open, untargeted way that most self-serve and managed buying interfaces encourage. I did not stop buying attention. I moved budget toward formats where the user has chosen to engage: search intent, retail and marketplace placements, newsletter and editorial sponsorships, podcast reads, and creator-led placements where the audience opted in. These formats cost more per impression, and that is the point. I would rather pay for a smaller number of deliberate exposures than a large number of accidental ones.
This article is about the metric that drove that decision, how to read it honestly, and how to think about scaling the alternatives without repeating the same mistake in a new channel.
02 What this metric really measures
The headline metric in display is usually the impression, and the more honest version of it is the viewable impression. An impression is counted when the ad is served, which is a statement about the ad server, not about a human. A viewable impression attempts to count when a meaningful portion of the ad was actually in the viewport for a meaningful amount of time. That is a much better proxy for attention, but it is still a proxy. It tells you the pixels had a chance to be seen, not that anyone processed them.
Click-through rate sits on top of that. In display, CTR is typically very low, and most clicks that do occur are not the product of considered interest. Some are accidental taps on mobile, some are curiosity clicks from a confusing creative, and some are generated by placements designed to capture misclicks. A high CTR on a cheap placement is often a warning sign rather than a win, because it usually means the creative is interrupting rather than informing.
What the metric really measures, then, is opportunity and friction. Viewability measures whether the opportunity existed. CTR measures whether the creative created enough friction or curiosity to produce a tap. Neither one measures whether the person wanted what you sell. That gap is why display reporting can look healthy while the business sees nothing.
03 How it moves with your levers
The levers that move viewability are mostly about where the ad appears and how the page is built. Placement quality matters more than almost anything else. Ads placed within the main content column, near the top of an article, or inside a feed tend to be seen far more often than ads parked at the bottom of a long page or behind lazy-loaded modules. Page speed and ad load matter too, because a page stuffed with many ad slots pushes each one further down and slows rendering, which reduces the share of impressions that ever become viewable.
CTR responds to a different set of levers. Creative relevance, size, and clarity move it, but so does frequency. Showing the same banner repeatedly to the same person tends to depress CTR over time and can push it toward zero. Frequency capping and audience refinement usually do more for CTR than any creative tweak. Targeting on declared or behavioral signals helps, but only when those signals are genuinely tied to purchase intent rather than inferred from a single page visit.
The lever people forget is the buy itself. Curated placements, direct deals with publishers, and allowlists tend to produce better viewability and better CTR than broad open-exchange buying, because you are choosing context instead of letting an algorithm optimize toward cheap inventory. Cheap inventory is cheap for a reason, and the reason is usually that it is hard to see.
04 Reading the number without fooling yourself
The most common misread is treating a low CTR as evidence that display does not work, or treating a high CTR as evidence that it does. Both readings ignore what the metric is actually capturing. A low CTR on a viewable, well-placed ad may simply mean the audience was not in market, which is fine if the goal is awareness. A high CTR on a poorly placed ad may mean you bought accidental clicks, which is not fine if the goal is revenue.
The second misread is comparing CTR across formats as if it were a single scale. Search clicks, social clicks, display clicks, and email clicks come from different levels of intent and are not interchangeable. A display CTR of a fraction of a percent and a search CTR in the single digits are not telling you that one channel is better; they are telling you the users arrived with different mindsets. Judging display against search on CTR alone is a category error.
The third misread is trusting attribution that gives display credit for conversions it merely accompanied. View-through windows can be generous, and last-click models can starve upper-funnel formats. Before you conclude that display is working or failing, look at whether the metric you are reading was measured the same way across the channels you are comparing. If it was not, you are not comparing performance. You are comparing measurement setups.
05 Benchmarks to sanity-check against
Rough benchmarks are useful only as a smell test, not as targets. Display CTR commonly falls somewhere in a band of about five hundredths of a percent to two tenths of a percent, with many campaigns landing near the low end. If you are seeing CTR far above that on cheap inventory, suspect accidental clicks or a creative that tricks people. If you are seeing CTR far below it on premium placements, suspect the audience or the offer rather than the format.
Viewability is often reported in a rough range of about half to roughly two thirds of served impressions, though the number depends heavily on the measurement definition, the device mix, and the page layout. Mobile in-feed placements tend to score differently from desktop sidebars. Treat any single viewability figure as a property of a specific buy, not of display as a whole.
Conversion rates from display are usually low and highly variable, so I would not anchor on a specific figure. The more useful sanity check is the ratio between viewable impressions and meaningful downstream actions. If you are buying a million impressions and only a tiny fraction are viewable and an even tinier fraction convert, the problem is not the benchmark. The problem is that you bought reach you could not use.
06 Scaling what works
Scaling display is easy and that is exactly the trap. You can add budget and get more impressions immediately, but the quality of those impressions tends to fall as you push past the inventory that actually fits your audience. CTR and conversion quality usually decay as you broaden targeting, lower your viewability floor, or accept more open-exchange supply. The metric that looked acceptable at a small budget often looks worse at ten times the spend, because the marginal impression is worse than the average one.
When I move budget to formats with deliberate engagement, scaling looks different. Newsletter and podcast inventory is finite, so growth comes from adding new partners rather than simply raising bids. Search and marketplace placements scale with demand but get more expensive as you compete for the same intent. Creator placements scale through relationships and repetition, not through a single dashboard slider. In each case, the constraint is attention, not inventory, and that constraint is what keeps quality from collapsing.
The practical rule is to scale the placement, not the format. Find the specific contexts where the metric holds up, then add more of those, and stop when the marginal placement stops behaving like the ones that worked. If you cannot find enough good placements to absorb the budget, the honest answer is that the budget should go somewhere else.
07 Frequently asked questions
Is display advertising ever worth buying? Yes, in specific situations. Retargeting known visitors, reinforcing a brand during a defined campaign window, and buying curated placements on sites your audience actually reads can all make sense. What rarely makes sense is buying broad, cheap, untargeted display and expecting it to behave like a performance channel. The format is not the problem; the buying method usually is.
What should I buy instead? It depends on your goal, but the common thread is deliberate engagement. Search and marketplace placements capture existing intent. Newsletter and editorial sponsorships borrow trust. Podcast and creator reads reach people who chose to listen. None of these are as cheap per impression as display, and that is the trade you are making. You are buying fewer, better exposures.
How do I know if I made the right call? Compare cost per meaningful action, not cost per impression, and make sure the measurement is consistent across channels. If the new formats produce a higher cost per impression but a lower cost per acquired customer or qualified lead, the switch paid for itself. If they do not, the problem may be your creative or your offer rather than the channel, and that is worth testing before you move budget again.