A company selling compliance software to regional banks put $8,000 a month into programmatic display. The quarterly report showed 14 million impressions at $1.70 per thousand and 19,000 clicks. The agency presented the buy as the cheapest reach the company had ever bought. A new marketing director asked for a placement report, the list of sites and apps where the impressions ran. The top placements were a free solitaire app, a celebrity gossip site, and a cluster of pages that exist to host advertising rather than to be read. Fewer than one impression in twenty ran anywhere a bank risk officer would plausibly spend time. The report was accurate. The buy was close to worthless.

Nothing in that report was miscalculated, and the same pattern sits inside most programmatic reports that nobody has questioned. This post explains what programmatic display is for, how display is bought, where display fits beside the other channel families, how to measure display, and the four ways display reporting overstates the channel. The series hub, How B2B Media Actually Works (and What It Cannot Do), covers the research behind the plan as a whole: the 95:5 rule, the buying committee, and demand creation versus demand capture.

What Programmatic Display Is For

Programmatic display is a demand creation channel. The Ehrenberg-Bass Institute's research on B2B buying holds that the overwhelming majority of category buyers, around 95 per cent, are not in market in any given period. Those out-of-market buyers cannot be converted this quarter. Those out-of-market buyers can be kept familiar with the company, so that the company comes to mind when a buying trigger fires months from now. Memory is the mechanism, and memory decays without refreshment. Our post on mental availability in marketing sets out that mechanism in full. Display's contribution to the mechanism is modest per impression and cheap per impression: a logo, a message and a category cue, glanced at rather than studied, repeated over months.

The practical job is background presence between exposures on other channels. A paid social campaign might reach an operations director a few times a month at a cost of tens of dollars per thousand impressions. Display keeps the company visible to that same operations director through the weeks in between, at a cost of a few dollars per thousand impressions. Display also carries a second job: account-list targeting. Upload a list of named target accounts, and display becomes the cheapest way to keep those accounts warm across the buying cycle. Gartner's research on the B2B buying journey describes that cycle as months long, looping rather than linear, and decided by a buying committee of six to ten people. Most members of that buying committee will never visit the website, and display can still reach those members, lightly and repeatedly, for less than any other channel charges.

How Programmatic Display Is Bought

Programmatic display is bought in automated auctions that run in the milliseconds while a page loads. The buying platform bids for each impression, priced per thousand impressions, across the open exchanges and across private marketplaces where publishers sell curated inventory. Targeting comes in three broad forms: contextual targeting against the content of the page, audience segments built from firmographic and behavioural data, and uploaded account lists. Cost per thousand impressions is driven by the quality of the inventory, the narrowness of the targeting and the format.

Format matters more than most plans admit. Standard banners sit in the margins of a page, and readers have learned to look past the margins. Native placements sit inside editorial feeds, styled to match the articles around the placement, and earn measurably more attention as a result. Native costs more per thousand impressions than standard banners. The premium buys attention, and attention is the whole point of a channel that works through exposure. A plan that buys the cheapest banners available has usually bought the inventory nobody looks at.

Placement quality is where most display budgets are wasted. The cheapest impressions on the open exchanges concentrate in three places: made-for-advertising sites, which are pages built to host ads and recycle arbitraged traffic rather than to be read; app inventory, where a banner sits under a game a distracted commuter is playing; and below-the-fold slots that load without ever entering the visible screen. Viewability is the share of served impressions that appear on the visible screen for long enough to be seen; the common industry standard counts an impression as viewable when at least half the ad is on screen for at least one second. An impression that was never on screen did nothing. Set a viewability floor on every display buy: an instruction to the buying platform to buy only placements whose measured viewability stays above a chosen percentage, so impressions that load off screen or below the fold are not bought at all. A viewability floor only removes unseen impressions. Passing the floor does not make an impression valuable: a viewable impression on a made-for-advertising site is still an impression nobody in the ideal customer profile received. Inclusion lists, which restrict the buy to named publishers, are the strongest single control. The course Media and Measurement carries the method for building those controls into a buy.

Where Programmatic Display Fits in the Plan

Display sits beside paid social inside the demand creation budget, and the two channels trade off against each other in a specific way. Paid social, covered in Paid Social in B2B: Reaching Buyers Who Are Not Looking, offers role-level targeting accuracy and a feed environment where strong creative can hold attention for seconds. Display offers neither. Display's targeting is looser, and display's attention per impression is a glance at best. What display offers in exchange is price: frequency at a fraction of the cost per thousand impressions. Run the two channels together with distinct jobs: paid social builds the memory with richer creative, and display refreshes that memory cheaply between the social exposures. Display inventory is also one of the places where retargeting runs. Retargeting serves an audience of already-engaged people across display, social and video inventory, and Retargeting in B2B: The Most Overcredited Tactic in the Plan covers why retargeting's numbers need separate scrutiny.

Set and enforce a frequency cap: the maximum number of times one person is shown the ads in a month. A B2B account list of 400 companies contains a few thousand relevant people at most. Without a cap, the auction serves the most impressions to the people who are online the most hours, so a part of the audience receives dozens of impressions a month and the rest receive almost none. Creative wears out on small audiences: after enough repetitions the same banner stops adding memory and starts to register as wallpaper, and past that point extra frequency buys irritation. Cap frequency per person, spread the budget across the list, and rotate creative before wear-out arrives. In my experience, programmatic display earns a minority share of the demand creation budget, and display earns that share only after paid social is properly funded, because cheap frequency is worth little until better-targeted channels have built the memory that display refreshes.

How to Measure Programmatic Display

Clicks are the wrong measure, and the reasons are structural rather than cosmetic. Display click-through rates run at small fractions of one per cent, so the clickers are a tiny and unrepresentative sliver of the people reached. A meaningful share of recorded clicks are accidental, a thumb slipping on a phone screen, and a further share are fraudulent, generated by software rather than by buyers. Worst of all, optimising a display buy to clicks steers the budget toward the placements that produce clicks, and the placements that produce clicks are disproportionately the junk placements: game apps with buttons beside the ad slot and made-for-advertising pages built to harvest accidental taps. A display campaign optimised to cost per click will steadily move its budget onto the worst inventory available.

Measure display the way demand creation is measured. In quarter: viewable reach and frequency within the ideal customer profile or the account list, which answers whether the right people were present when the ads appeared. As leading indicators: lift in branded search volume and in direct traffic, and the count of target accounts newly engaging with the website or the sales team. On a lag: pipeline among reached accounts, compared with matched accounts that were not reached, read two to four quarters after the spend. Binet and Field's analysis of the IPA databank explains the lag: brand-building effects compound slowly and persist, while activation effects spike and fade, so a demand creation channel judged inside one quarter will always look like waste. What a display report can never provide is proof of cause from attribution alone, and why marketing attribution misleads covers the reasons.

The table below is a worked example for a company selling supply chain analytics to mid-market manufacturers. The company runs two display buys in parallel at the same quarterly spend. The open-exchange buy is optimised to cost per click. The curated buy runs on an inclusion list of trade and business publishers, with a viewability floor, optimised to viewable reach inside the 400-account target list.

Example figures for illustration. Engaged target accounts: accounts on the 400-account list showing new website visits or sales conversations during the quarter.

The open-exchange buy wins every volume metric: more impressions, more clicks, a far lower cost per click. The curated buy is the only buy that put the ads in front of the audience the company sells to. Any report ranking the two buys on clicks or on impression volume would move the budget in exactly the wrong direction. Rank the two buys on viewable reach inside the account list and on engaged target accounts, and the curated buy is the only one doing the job display was hired for.

Where the Numbers Lie

Display reporting overstates the channel in four recurring ways.

1. Clicks are reported as results when almost nobody clicks on purpose

A display report that leads with clicks and click-through rate is reporting noise. The rates are tiny, a portion of the clicks are accidental taps on mobile screens, and a further portion are generated by fraudulent software that clicks to make junk inventory look responsive. The people who deliberately click banners are not a fair sample of the buyers the campaign reached. The deeper damage arrives through optimisation: a buy tuned to cost per click shifts spend toward whatever placements produce clicks, and accidental and fraudulent clicks concentrate on the worst placements. Remove click metrics from display evaluation entirely. Report viewable reach within the ideal customer profile, and let the website's own analytics record whatever visits arrive.

2. Cheap impressions concentrate where nobody in the ideal customer profile is looking

An impression total in the millions looks like scale, and the cost per thousand impressions looks like efficiency. The auction produced both numbers by finding the cheapest available inventory: made-for-advertising sites, app slots and below-the-fold positions. The compliance software company in the opening scenario bought 14 million impressions and almost no exposure among bank risk officers. Volume metrics look strongest when the buy has drifted to low-quality placements, because low-quality placements are where impressions cost least. What to do: pull the placement report every month and read where the money went, restrict the buy to an inclusion list of named publishers, and set a viewability floor so unseen impressions are not bought at all.

3. View-through conversions claim credit the impressions did not earn

A view-through conversion is recorded when someone converts after an impression was served to that person, with no click involved. Display serves impressions in the millions, so almost anyone who converts through any route has an unnoticed display impression sitting somewhere in the trail. The display report then claims conversions that paid search, paid social, the sales team and the website produced. The claim grows with impression volume, so the cheapest and least attentive buys claim the most credit. Treat a view-through conversion as a record of exposure, never as a record of cause. Where the claim matters for budget, test the claim: compare conversion rates between reached accounts and comparable accounts held out of the campaign.

4. Reported reach counts traffic that includes machines

Open-web reach figures count served impressions, and a share of open-web traffic is not human. Automated browsers load pages, register impressions and inflate the counts, because fabricated inventory costs almost nothing to produce and sells into the same auctions as real inventory. The cheapest exchanges attract the most fabricated supply, since fraudulent supply concentrates in the exchanges with the least vetting. The result is that reported reach overstates the number of humans reached, and overstates most exactly where the impressions were cheapest. Ad blockers push in the same direction from the opposite side: a share of the real human audience never receives the impressions at all, and that share skews technical. What to do: buy through curated inventory where the publisher is known, apply independent verification to the buy, and treat open-exchange reach claims as an upper bound.

What Programmatic Display Cannot Do

Programmatic display cannot capture demand, because no impression is summoned by a buyer's query the way a search ad is. Display cannot carry the demand creation job alone, because a glanced-at banner builds less memory per exposure than the video and feed formats paid social and online video deliver. Display cannot convert: clicks are rare and partly accidental, and conversion is not the job display is funded to do. Display cannot run unattended, because an unmanaged buy drifts toward made-for-advertising sites and app junk within weeks. Display cannot reach the audience segment that runs ad blockers, and that segment is large in technical categories. Display cannot prove its own contribution from platform reporting, because clicks are noise and view-through conversions are claims. What display can do is one thing well: keep the company visibly present to the ideal customer profile and the account list for less money per exposure than any other channel in the plan.

Programmatic Display FAQs

Why is the display click-through rate so low, and does the low rate matter?

Display click-through rates are small fractions of one per cent because banners are ambient: people glance at banners while doing something else, and almost nobody interrupts a task to click one. The low rate does not matter, because display works through repeated light exposure that builds and refreshes memory, never through clicks. Judge display on viewable reach within the ideal customer profile and on lift, and stop reporting the click-through rate at all.

What does a very low CPM signal in programmatic display?

Usually inventory quality. The auction prices impressions by demand, and the impressions nobody else wants are cheap for a reason: made-for-advertising sites, app slots beside game buttons, and positions that never enter the visible screen. Check the placement report before treating a very low cost per thousand impressions as good news. The fair comparison is cost per viewable impression within the ideal customer profile, and on that comparison the cheap open-exchange buy often turns out to be the most expensive way to reach a buyer.

What is a made-for-advertising site?

A made-for-advertising site is a page built to sell ad impressions rather than to serve readers. The operator buys cheap traffic, loads each page with a high density of ad slots, refreshes those slots aggressively, and sells the resulting impressions into the open exchanges at low prices. The impressions are real in a technical sense and worthless in a commercial sense, because no buyer chose to spend time there. Inclusion lists are the reliable defence, because an inclusion list admits only publishers someone has named and checked.

Are native placements worth the higher cost?

Often, yes. Native placements sit inside editorial feeds and match the form of the content around the placement, so readers give native placements more attention than banners in the page margins. Display is paid for exposure, and exposure without attention is close to worthless, so paying more per thousand impressions for a format people notice is frequently the better trade. The comparison to run is cost per viewable, attended impression within the ideal customer profile, not cost per thousand impressions served.

How long does programmatic display take to show results?

Delivery metrics arrive immediately: viewable reach and account match are visible in the first weeks and confirm the buy is running against the right people. Leading indicators, such as lift in branded search and direct traffic and newly engaged target accounts, move over one to two quarters. Pipeline among reached accounts arrives two to four quarters after the spend, in step with the buying cycle. Binet and Field's IPA databank analysis shows why the delay is structural: brand-building effects compound slowly, and short evaluation windows systematically undercount brand-building effects.

Go Deeper

This post carries the judgement: what programmatic display maintains, why clicks mislead, and why placement quality decides whether cheap reach is reach at all. The method sits in the course Media and Measurement, which covers inclusion lists, viewability floors, frequency caps and account-list measurement step by step. The free module in B2B Marketing Fundamentals covers the ideal customer profile that every display audience inherits. The memory mechanism that display exists to serve is the subject of mental availability in marketing.

Sources

• Ehrenberg-Bass Institute for Marketing Science, 2021, How B2B Brands Grow: the finding that around 95 per cent of category buyers are out of market in any given period, and the role of memory and mental availability in which suppliers are considered when buying begins.
• Les Binet and Peter Field, The Long and the Short of It, IPA, 2013: analysis of the IPA effectiveness databank showing that activation effects spike and decay within weeks while brand-building effects compound over years, and that short evaluation windows shift budgets toward activation.
• Gartner, The B2B Buying Journey: typical buying groups of 6 to 10 decision makers, a long non-linear journey that loops through research and evaluation, and most buying time spent away from suppliers.