Which ad metrics matter most in 2025

Marketing is increasingly full of metrics. Unfortunately, those metrics are often full of, well, something else. But as “performance marketing” becomes the norm rather than the exception, increasingly the industry revolves around metrics. So understanding them, how they work (and how they often don’t work), is increasingly important. Here are 18 of the most important metrics to understand. Ad Age Tech Power List 2025 We’re looking for top digital innovators from agencies, brands, platforms, ad tech companies and tech startups Enter here CPM Cost per thousand is an easy to understand number that makes the media world go round. But for a variety of reasons, it can be deceptive. The “C” tends to stand for media costs. Measuring the thousands is trickier, given the possibility of dubious audience counts, including “made for advertising” websites that deliver scads of viewable impressions through rapid ad refreshes, heavy ad loads and pop-unders that no one really can see. eCPM Short for “effective CPM,” or the cost of reaching real people who are part of the marketer’s target audience. If you reach 2,000 people, but only 1,000 of them are your target, the eCPM is twice as high as the CPM. Assuming you can trust the verification providers—media companies and databases used to calculate this number—it’s a better way of assessing true efficiency. aCPM Short for attentive CPM, this is a metric that adjusts CPM based on attention, for example by using total attentive minutes or only counting impressions that meet a base attention threshold. Again, it’s probably more useful than straight CPM, but tricky to define for pure effectiveness purposes. Does someone watching an ad for 30 seconds really deliver more impact than someone watching for 10 seconds? Not according to research presented by, say, YouTube. Then again, the Google unit has a vested interest in favoring shorter duration impressions (unless it’s selling unskippable ads.) Check out the latest Ad Age events and award programs CAC Customer acquisition cost, an all-in metric that covers all the marketing costs associated with landing a new customer, is the favored metric for increasingly ubiquitous performance marketing. What counts as “good” here really depends on the lifetime value of the customer (see below), which can vary widely by category. But a universal truth is that chief financial officers want it to get continuously lower. Working/non-working media ratio This is a metric that very much matters at many companies, even if it shouldn’t. It has many flaws, yet remains deeply engrained in finance departments, and, by extension, procurement departments, where it ultimately shapes agency and media contracts and marketing strategy. This is a measure of the ratio of paid (i.e. working) media to all the other associated costs that go into making ads and placing them (i.e. non-working media). That “non-working” stuff includes agency creative costs, production costs and media fees, including the numerous costs associated with programmatic media trading. Decades ago, marketers paid a media commission of 15% that covered creative, production and media placement costs, so the working/non-working ratio was 5.7 to 1. Today, say, in online media traded programmatically, just costs of demand side platforms, supply side platforms, data, verification and other often opaquely reported nickels and dimes add up to almost 30% of what gets sent to publishers on the open web, not even counting the costs of creative. So the working/non-working ratio is likely closer to 2 to 1. Ad Age Leading National Advertisers 2024 U.S. ad spending, rankings, analysis and profiles for the biggest marketers Click here The working/non-working ratio is one reason some marketers sign on to principal media deals in which agencies resell media rather than acting as agents. When they do, costs for fees (non-working media) essentially get folded into the “working” paid media side of the ratio. Marketers may end up paying more overall (they’ll likely never know) because the agency’s media cost is undisclosed, so it makes the ratio look better. Having media companies handle creative and fold that into the media cost has the same effect. Maybe the creative isn’t better, and the cost breakdown may be something of a mystery, but it makes the ratio look better. Conversion Often, perhaps ideally, a conversion is a sale. Realistically, that’s not always practical. Not all marketers have direct access to customer or sales data. For that reason, or because of different campaign objectives, marketers might define what constitutes a conversion differently, including filling out a form or watching a video to completion. In cases where none of those data types are available, survey-based research of purchase intent or brand favorability may suffice. Time to conversion An important part of properly attributing outcomes to various media or other marketing elements is understanding time to conversion. It usually isn’t instant (see last-click attribution below), according to Provalytics CEO Jeff Greenfield. Knowing how long it typically takes to move a customer to purchase is an important part of analyzing the role played by various media. Time to conversion varies by category. Lower-cost products bought more frequently generally take less time; big-ticket items take more. Impressions The growing prevalence of performance marketing has made conversion the key outcome most marketers track. It’s also generally steered more marketers away from tracking the impressions (number of times people encounter ads) from their campaigns, Greenfield said. That’s a problem, because brands ultimately can’t grow without new customers, and focusing on conversions as the main indicator of success can lead marketers to keep targeting the same well of customers who’ve already been converted. That in turn leads to the “doom loop,” where campaigns deliver ever-improving conversion rates but aggregate sales decline anyway, because some customers stop buying and new ones aren’t being reached. Beating the doom loop requires tracking impressions too, but Greenfield said performance marketers increasingly pay little attention to them. And while there are many ways impression data can mislead (e.g. impressions that go to bots rather than humans or happen in excessive clutter), they’re still important to track. Reach Maximizing reach is a way out of the doom loop, albeit an expensive way and increasingly difficult. Mass marketers like Procter & Gamble, which went too far chasing efficiency by narrowing their targets, ultimately blamed that for chronically declining market shares and shifted to ensuring the broadest reach of relevant audiences they could. Byron Sharp, a University of South Australia marketing science professor, has railed for years against targeting and loyalty marketing efforts that don’t let brands recruit new customers to make up for the ones they inevitably lose. Delivering reach is key to that. The trouble is, even if digital signals are mainly about narrowing targets, their loss, largely due to privacy restrictions and digital walled gardens, also makes it harder to ensure reach. Overcoming that signal loss is a key impetus behind the ANA developing Aquila cross-platform measurement, which, while it’s currently only set to cover the biggest digital players, would span media companies representing most U.S. spending. Super Bowl commercials Tracking the latest brand and agency plans Click here Frequency The other part of ensuring the “mental availability” that Sharp prizes is frequency. Too little frequency means people are more likely to forget the brand or the message. Too much turns people off and potentially delivers even negative returns. Just like with reach, measuring frequency can be difficult given the oddities of TV audience measurement, signal loss from digital platforms and walled gardens that make it hard to track people across media, and defects in programmatic digital trading. Recency Not all frequency is created equal. Assuming you can accurately measure how often people see your ads, if you deliver 100 impressions, spreading them out over 52 weeks is likely to work better than all in one month. People are more likely to remember the brand or product when they make purchases that way. Some ad formats that may do little to persuade (e.g. logos on athletes’ shirt sleeves or brief logo appearances on TV) are mainly about achieving recency. Match rate This is the rate at which two databases match. As marketers increasingly rely on their own first-party data, matching that against databases from third-party data brokers, retailers or media companies is an increasingly important way for them to leverage that data. Often this is done using email addresses encrypted (or hashed) to preserve privacy. The higher the match rate, the more data brokers and media companies get paid. Does this create incentives to fudge numbers? Sure. One publisher executive tested using AI to generate fake email addresses to see what results would come back from data brokers, finding some came back with “match rates” of 30% to 50%. Another marketer tried a highly reliable email database— that of its employees—and found very low, albeit apparently honest, match rates with a major data supplier. So while match rates are important, using them without testing is questionable. Attribution This means assigning fair and proportional credit to a medium or marketing tactic for an outcome, such as sales, site visits or purchase intent on the belief that more than one impression, message or medium is responsible. Last-click attribution This is the tendency to credit the “last click,” usually from search or a click on an ad, for the sale. It’s easy to measure, since it only involves data from one source. It’s also deceptive, because one source alone is rarely responsible. People often use Google search, e-commerce site search or AI search after encountering an ad elsewhere, hearing about something from friends or social media creators, etc. But this isn’t really just a digital problem. The old brick-and-mortar fallacy of crediting sales in stores only to retail promotions that happen in those stores, without accounting for the advertising that led to brand or product consideration in the first place, is the same thing. Multi-touch attribution This is the more sophisticated version of attribution, assigning weighted credit to a variety of messages, media and other factors along the so-called “purchase funnel” for the ultimate sale. It would be great if it were possible. Given the growing loss of identity signals in digital media and elsewhere, it’s increasingly hard to track people to help calculate this definitively. ROAS Return on ad spending, i.e. the sales that can be attributed to a campaign divided by the spending, is one of the simplest measures of marketing success, provided you can reliably get all the data on both sides of the equation. That can be easier said than done, primarily because returns can be inflated by piling in sales that would have happened even if the campaign never ran. Using econometrics (marketing mix modeling) or matched-market (A/B) testing can help get around this. ROI Sometimes conflated with ROAS (return on ad spend), it’s not the same thing. Return on investment for a campaign measures the incremental profit generated by a campaign, not the sales, and that will vary by company and product line based on the profit margin. High ROAS on a low-margin, money-losing product is generally not a good thing (though it can be where the goal is market penetration to build long-term profitable customers). Customer lifetime value Knowing the lifetime sales (or profit) value of a new customer is one key to determining ROAS or ROI. It’s impossible to know for certain, of course, but very possible to estimate based on a brand’s historical data and market research. Read Ad Age’s 2025 predictions stories here

Which ad metrics matter most in 2025

Featured Stories