The state of Adservers in 2025
Adserving is a critical component of digital marketing that remains the key to serving the appropriate ad in the various situations that may present itself to an end user. But with the theme of both technological & regulatory changes having implications for every part of adtech, adserving is not immune to this. This raises the question, what is the value of an adserver?
What does an adserver do today?
The adserver market is constantly evolving, primarily due to the newer mediums like Connected TV, Retail Media & DOOH entering the market which still requires the ability to serve an ad in. It is also important to distinguish between the 2 sides:
The supply-side / publisher ecosystem, where the end goal is to be able to monetise through the medium of advertising
The buy-side / advertiser ecosystem, where the end goal is to create / build / measure brand specific ads across different platforms / channels
As with any tool, there are plenty of pros and cons to it. But if we concentrate into the different use cases / functions of an adserver, we can understand more clearly what the future state will look like.
Ad serving
By far the most critical element of an adserver is exactly what it says on the tin. The end goal of being able to respond to a bid request & generate an ad call with specific criteria / assets is ultimately what the adserver has done since its birth.
To some extent, this predates even OpenRTB and is a vital component of any modern day ad platform today. Every platform in some capacity adserves, but reliance on 3rd party tools differs depending on the platform. This is exactly where Google asserted its dominance over its rivals, by having adservers on both the publisher (formerly known as DFP through Google Ad Manager GAM) & advertiser (CM360) side, allowing them to dominate certain mediums like display but now very much under the microscope from all the antitrust cases ongoing currently.
Yes even those popular social channels all need to adserve, which is handled largely within their own systems. Even on a DSP like DV360, TTD or Amazon, when you upload a raw asset, the platform needs to host / adserve it on your behalf. With DV360, given Google have their own adserver in CM360, this effectively is making use of that technology but without the full capability if you were to use it directly.
The complexity comes with how the art of adserving aligns with other criteria within an ad impression, which will cover next. But this imperative part of digital advertising is going nowhere.
FUTURE STATE: SAFE
Dynamic Creative
If we move a step further than the standard ability to serve a creative asset as an ad on a given platform, the ability to personalise ads to individual behaviour / traits has seen constant upgrades to how adservers approach this.
This effectively brought in the era of DCO aka dynamic creative optimisation. Serving a personalised ad to the right individual at the right moment is still a viable tactic.
However over time, some of this capability has become more restricted and newer mediums also struggle to allow the same possibilities that may have existed in the past. A good example of this is in the retargeting space. Whilst in closed ecosystems there is still the ability to do this with other signals, the traditional open web way of using 3p cookies / mobile device ids has made this much more challenging. Whilst alternatives exist, none bring the scale of what advertisers may be used to right now.
Therefore there is currently a dual approach being seen:
Use signals that are more aggregated and not as dependent on technological / privacy changes. This could be things like contextual, location & time of day.
Or just go back to old school key value pairs, by passing through the relevant information in your adserved tags and matching this to bespoke creative messaging through your adserver
Initiatives likes Universal IDs, curated PMPs and even the Privacy Sandbox Protected Audience API do show that remarketing based DCO still has some life in it. But on the open web, it is likely to shift to more of a cohort approach, leaving the user level element to Social / Retail Media.
FUTURE STATE: SHIFTING
Verification REporting
Over time, adservers have played a big role in what reporting may look like. This can be from a more performance perspective but also this has been critical for billing / payment structures, especially when working between the buy & sell side.
However this has never been plain sailing for anyone involved in adtech. That 10% magic discrepancy number has been unequivocally part of every troubleshooting call between brands / agencies / publishers / vendors. The actual definitions of what should count as an impression has changed over time, particularly due to newer mediums in app / video but also the complexity of when exactly to measure this. Case in point is the constant challenges ad verification companies have with this exact conundrum.
There has also been a shift for certain adservers to incorporate ad quality / suitability into their offerings. Whether its Google’s free verification tool within CM360 or MediaOcean’s acquisition of Protected Media within Flashtalking’s ad stack, there is interest to monitor where exactly ads served.
The crown jewel was the birth of Viewability as a metric, which every platform / adserver attempted to create a means to track this. However due to the need to leverage JavaScript or VPAID in video to do this, there have been quite significant challenges / limitations with this approach.
But by far the most important verification metric is plain old impressions, which can then be tied to one of the more popular upper funnel KPIs of Reach / Frequency. This is where 3P cookies in particular had the ability to operate cross-publisher where applicable. But as this has become more challenging over time, the use of panel / conversion modelling is now rife within the reach space for even the adservers. Add in the newer CTV / Retail Media mediums, it becomes a challenge. A challenge that may be better solved by alternative solutions like Audience Project or Project Halo via ISBA’s Origin / ANA’s Aquila.
FUTURE STATE: SHIFTING
Measurement
Measuring success of digital advertising is arguably the most important concept within the life cycle of any campaign. And the role of an adserver in this has become pivotal for brands that have invested in an adserver as a means to measure cross-channel activity. In particular was the ability to measure beyond a click; with the use of post-impression attribution that is often seen within ad platforms themselves also being exposed on a wider level through enterprise adservers.
Within elements of programmatic activation, channels like Display will never look particularly good on a last click model so the ability to show a different perspective through post-impression was useful but somewhat questionable at the same time. Again this was made possible through the use of 3P cookies to stitch together conversion paths.
But over time, the trust & accuracy in adservers providing reliable attribution results has somewhat shifted. As Walled Gardens started to restrict 3rd party adservers from tracking alongside channels like YouTube moving more into a Data Clean Room space, this has meant the overall trackability has reduced.
There is still some hope, due to advancements in AI, conversion modeling and privacy enhancing technologies. We have seen the Attribution Reporting API from the Privacy Sandbox very much part of plans of adservers to integrate with to provide measurement without dependency on 3P cookies. But has also been a lot of failure. Look no further than both Meta (with Atlas) and Amazon (with Sizmek) very much trying to build smarter cross-channel attribution using an adserver but proving too much of a challenge to the point where it is no longer a solution.
On the publisher side, measurement is also somewhat critical to ensuring the value of each impression served. Whilst a lot of this may be revenue / margin related, the engagement metrics are also important, especially when trying to monetise above the fodl vs below the fold inventory. Something that very much went under the radar in recent times was Google getting rid of conversion tracking (Spotlights) on it’s supply-side adserver, which has meant there is less data beyond the ad call to work with & more dependency that the end advertiser can track performance to send back to the publisher.
Measurement is shifting no doubt, with the triangulation approach now even more common across the largest brands using MMM, experimentation & attribution. An adserver can fit into all 3 but has generally been focused on the last pillar. I personally still believe that a custom attribution model should go beyond look at click based solutions or a lot of the snake oil 1st party solutions that exist in the DTC space. But whether this is a critical use case for adservers will depend on your channel mix. Clearly with advancements in CTV & Retail Media, this is the new breeding ground for adservers to attempt to solve for, which is very much seen in their 2025 product roadmaps.
FUTURE STATE: SHIFTING
Closing Thoughts
Adservers, as with any technology are in the process of evolving to keep up with the increasingly fragmented & complex digital advertising space. There are use cases which remains important & where an adserver is very much at home with.
As emerging mediums continue to grow, the supply & buy-side of the industry will see newer solutions try to find a footprint in order to provide meaningful opportunities for brands / publishers. CTV in particular has seen plenty of innovation from the likes of Publica.
Though the largest question marks will be on the buy-side. Should brands continue to invest in a Google CM360 or MediaOcean Flashtalking as the big two or a smaller one like Adform / XR / Clinch. MediaOcean could be very interesting to follow given their recent acquisition of Innovid, however this may take a little while to fully integrate into a complete solution.
Personally, if you are someone that is heavily investing in Display & Video, an adserver still has some power. But the biggest headwind here is where AI will go, which could take some of the creative / measurement elements into a different space per individual platform.