Behind the Impression Count
Notice plans are measured in impressions because impressions are easy to count and easy to specify in a contract. The figure documents that media ran, and it gives counsel and the court something concrete to evaluate. It is a standard and necessary metric, and it belongs in every notice report. But the count answers a delivery question, not a notice question. What an impression counts varies by platform. The major platforms count an impression at or after the point the ad begins to render, while other inventory may log one earlier in the delivery chain, before the ad could be seen. Either way, the count does not establish that a person was there to see it, that the person belonged to the intended class, or that the placement gave the message a fair chance to register. Each of those is a separate question with its own measurement, and treating the impression count as the answer to all of them overstates what the number proves.
That distinction matters more as digital delivery carries more of the notice burden. Adequacy turns less on how many impressions a plan reports and more on what those impressions represent: whether they were viewable, served to the intended class, delivered in environments where notice can register, seen often enough to land, and counted free of fraud and waste. A program can deliver tens of millions of impressions and still leave much of the class effectively unreached, if those impressions were low-visibility, off-audience, or never rendered to a person at all. A smaller, higher-quality delivery can do more real notice work. The impression count cannot tell the two apart; that depends on the composition behind it.
What matters is not how many impressions were delivered, but what kind of exposure they represented.
Viewability is the gap between counted and seen
A counted impression and a viewable impression are different counts. An ad can begin to render, and so be counted and billed, while it sits in a background browser tab, loads below the fold and is never scrolled to, or appears on screen for a fraction of a second. It was delivered, although it had little chance of being seen.
The industry built a standard around this gap. Under the Media Rating Council (MRC) viewable-impression guidelines, a display ad counts as viewable when at least 50% of its pixels are in view for at least 1 continuous second, and a video ad when at least 50% of its pixels are in view for at least 2 continuous seconds.[1] The MRC audience-measurement standards go further, treating the viewable impression, not the served impression, as the minimum qualifying unit for measuring reach and frequency.[1]
For notice, the implication is worth keeping in view. Reach and frequency carry weight in a declaration, and the standard-setting bodies build those figures on viewable impressions rather than served ones. Viewability can be measured today through accredited third-party verification, though it is typically a paid add-on rather than a default, so whether a given program reports it is often a matter of scope. Distinguishing viewable from counted impressions is a reasonable direction for notice reporting to move, and a large impression count can be entirely legitimate under the current standard while still sitting on a low viewable rate, so volume alone is no sign the ads were seen.
Audience composition is what separates reach from notice
In the impression count, an ad shown to a class member and an ad shown to someone outside the class look exactly the same. For notice, they are not equal.
Honesty about the current environment matters here. The audience signals available for targeting have become less precise, and in some matters are not available at all, so some share of out-of-class delivery is unavoidable, and that share has grown. The better practice is to work with that limit rather than paper over it. That means modeling in-class composition openly, documenting the basis for it, and sizing the plan so that in-class exposure reaches the level adequacy requires. A program built on geographic and demographic targeting can be entirely defensible, provided the documentation describes what the targeting can and cannot do and the volume is scaled accordingly.
The reporting point follows from that. No report can measure, after delivery, how many impressions reached the class. The defensible approach is to model that share in planning against a defined audience proxy, record how the model was built, and have counsel accept it as part of the plan, rather than present a single reach figure the data cannot actually support.
Inventory and environment shape exposure quality
The same impression count can come from very different places, and the place matters. A message rendered on a well-trafficked publisher page or in a premium video environment has a different chance of being seen than the same message on a remnant banner or a site that exists mainly to carry ads.
The Association of National Advertisers programmatic transparency work found that a meaningful share of open-web programmatic spend reached made-for-advertising sites, pages built to attract ad dollars rather than readers, increasingly including AI-generated sites created for that purpose, and that a substantial portion of spend went to impressions that were never measurable or never viewable.[2] The share going to the worst inventory has fallen sharply since, but the ANA's more recent benchmarks find the gap between buyers who actively manage quality and those who do not still widening, with viewability and measurability the main dividing line.[2] None of this means display should be avoided. Used to extend reach against curated, screened inventory, it earns its place; the problem is only when cheap placements are used to run up an impression total. Advertisers control that line through exclusions and manual review. The point is narrower. Two plans reporting the same impression total can deliver very different amounts of real exposure, depending on where those impressions ran.
For a notice program, inventory quality is therefore part of adequacy, not only efficiency. Adding low-visibility inventory to reach a headline number raises the impression count without raising the amount of notice actually delivered.
Verification keeps the count honest
Bot traffic, click fraud, and other invalid traffic add impressions to the delivery log without a human ever being present. Accidental clicks and low-attention environments add a different distortion, a real person who never meaningfully received the notice.
Left unaddressed, this activity would pad a delivery count with impressions no person ever saw. The major ad platforms filter invalid traffic before it is counted, and independent verification tools can confirm that ads ran in appropriate, brand-safe environments. A plan that applies both and says so stands on firmer ground than one that reports raw delivery without addressing it.
A low CPM is only a saving if the impressions work
Cost per thousand impressions, or CPM, is a standard way to price and compare media. A lower CPM looks like a saving, and sometimes it is. But CPM measures the cost of delivery, not the cost of notice. A low CPM bought on inventory that does not render, does not reach the class, or runs in low-attention environments can cost more per unit of real notice than a higher CPM bought on quality inventory.
The more useful way to think about cost is per viewable, in-class impression, though that is a planning estimate, not a figure delivery can measure. Judged that way, the cheaper line item is often the more expensive one. This is the practical reason quality inventory is worth the cost in a notice plan. Adequacy has a floor: meaningful exposure to the class. Impressions that do not clear it do not help a program get there, no matter what they cost per thousand.
What this means for adequacy
Rule 23(c)(2)(B) calls for "the best notice that is practicable under the circumstances," a standard that asks about method and reasonableness, not volume alone.[3] Courts have long leaned on Federal Judicial Center guidance that treats a program reaching roughly 70 to 95 percent of the class as generally reasonable.[4] Both the rule and the benchmark rest on the same quiet assumption, that the reach behind the percentage was real.
That is where exposure quality does the work. Two programs can report the same reach and stand on very different ground, depending on whether the impressions behind it were viewable, in-class, and verified, or whether the figure was lifted by low-visibility placements, off-audience delivery, and traffic that never reached a person. An administrator who can account for the composition of the delivery, and not only its size, is in a far stronger position if the program is examined.
Questions counsel and courts can ask
For counsel and the courts that review notice programs, a few questions help tell media volume apart from meaningful notice. Not every program will have a full answer to each today, and the data behind some of them is still maturing. But the questions point in the right direction, and a program that can speak to them is on firmer ground:
Was the plan designed with viewable impressions in mind?
How was in-class reach modeled, and what does the model assume about audience composition?
What types of inventory carried the impression volume, and how were placements chosen for visibility?
What invalid-traffic filtering was applied, and is delivery reported gross or net of it?
Asked early, these questions shape a better plan. Asked at review, they help counsel and the court gauge whether the reported volume stands for notice.
Where this is heading
The standard for notice has not changed. The medium that carries it has. Impressions remain the unit most notice media is bought and reported in, and they belong in every notice plan. On their own, they are not proof that notice was given. What turns delivery into notice is the quality of the exposure behind it, whether it reached the right people, whether it ran where the message could be seen and shown often enough to land, and whether what got counted was real. A report that describes that, and not only the size of the buy, is the one that holds up.
None of this is abstract for us. We plan against a defined audience proxy, document how we approximate the class, and verify where the ads run, because that is what lets a reach figure hold up when a court looks at it. The parts the tools do not yet make easy, we treat as the work still ahead. As more notice moves to digital, the administrators and counsel who ask what a media number represents, rather than how large it is, will have the better answer when a court asks.
References
Viewable impression and audience-measurement standards. Media Rating Council — Viewable Ad Impression Measurement Guidelines (link); Media Rating Council — Digital Audience-Based Measurement Standards (link).
Programmatic inventory quality, made-for-advertising spend, and non-viewable / non-measurable spend. Association of National Advertisers — Programmatic Media Supply Chain Transparency Study: Complete Report (2023) (link); ANA — Q1 2026 Programmatic Transparency Benchmark (link).
Class notice standard. Federal Rule of Civil Procedure 23(c)(2)(B) (link).
Class notice reach benchmark. Federal Judicial Center — Judges' Class Action Notice and Claims Process Checklist and Plain Language Guide (2010) (link).
Alexander Stephens is the Client Director of Legal Notification Programs at Élan, where he serves as a strategic partner to some of the country's leading Class Action Administrators. He has directed more than 50 complex, multi-channel notice campaigns spanning digital, print, radio, and press across a broad range of class action and mass tort matters.
