Start 'em, Sit 'em revisited: publisher priorities for H2 2025

Back in January, as we approached the NFL playoffs, Rob and Craig Leshen from adops.com put publishers’ H1 priorities into two buckets: Start ‘ems and Sit ‘ems.

Six months on, they’re back in action, Fantasy Football-style. Let’s take a look at who’s breaking out as a starter, and who’s still warming the bench.

Rob: Craig, welcome back for another edition of Start ’em, Sit ’em. A lot has changed since January, including a new name for you guys - adops.com - congratulations! Six months ago, we huddled up to set the game plan for publishers in H1 2025 - efficiency, partner reporting, contextual, and curation were all in play. Fast-forward to H2: what’s changed?

Craig: Rob, great to be back in the fantasy draft room for another round. And thanks, we’re loving life under the new banner of adops.com. You’re right, the hot industry topics for H1 were all about building momentum: fixing partner reporting, implementing automation, experimenting with contextual.

Some plays are working, some need tweaking, and there are definitely some new names on the roster. It’s been a fast-moving six months, and the playbook keeps evolving.

Rob: So what’s the number-one Start ’em right now?

Craig: So aside from the day-to-day pressures of driving efficiency, figuring out how to automate certain aspects of ad ops, and keeping an eye on curation, there are a handful of buzzwords that have emerged since January that publishers need to understand in order to determine where they should come on the agenda for H2. These include sell-side decisioning, agentic AI, neuro-contextual targeting, and identity 2.0.

Rob: Ok let’s start with sell-side decisioning and agentic AI. Because to my mind, they go hand in hand in various ways.

Craig: Exactly, Rob, and this is where things get really interesting. Sell-side decisioning means publishers aren’t just taking orders from the buy side - they’re starting to call their own shots. It’s real-time logic baked into their ad stack that helps optimize yield: which SSPs to send a bid to, when to block formats, how to price impressions. Now enter agentic AI - that’s the quarterback. Instead of just analyzing data, it acts. It can forecast, adjust floors, or re-prioritize demand partners mid-flight, without human input.

In parallel to this, both sales and ops teams are exploring agentic AI to streamline their own workflows, marking a significant development in the type of AI-driven automation we talked about six months ago.

In both cases above, publishers are exploring the shift from static dashboards to agentic AI systems that execute decisions autonomously. In other words, we’re moving from assistive to autonomous. Publishers who’ve built a clean, efficient stack in H1 are now in a position to unleash this kind of automation.

Rob: Alright, next up - contextual was having its big moment back in January, when we were definitely saying goodbye to cookies for real this time. So now that cookies are still around, where does contextual stand today?

Craig: As you say Rob, contextual has risen in line with the demise of the cookie. Whereas it used to be a Plan B for cookieless, now it’s Plan A for smarter targeting. In fact, contextual has evolved once again, with a lot of noise recently around a ‘neuro-contextual’ technology. It’s not hype anymore - it’s starting to show real traction in the market.

We’re no longer just talking about “is this page about sports or news?” Much like the human brain relies on neural networks to detect patterns and interpret meaning, emerging contextual technologies apply similar principles to move past basic content classification. These systems - driven largely by AI - now aim to understand the deeper layers of consumer behavior underlying their actions, such as interest, intent, and emotion.

And there will be other advanced data-driven targeting tech emerging that we’ll all be talking about by the end of the year. The key is for publishers to get on the train now and start testing new targeting methods across all channels - display, video, CTV - so they can both recognize revenue for H2, but also build for the future.

Rob: Ok, let’s move onto identity. Again, the cookie saga has caused a massive shakeup here - why is this a hot pickup?

Craig: We’re pretty much entering the era of identity 2.0: frameworks built for a post-cookie world that are privacy-safe, interoperable, and empowered by AI. Today’s solutions typically blend deterministic data (e.g. hashed email or login-based identifiers), probabilistic signals (device IDs, contextual behavior, IP, etc.), and AI and machine learning to stitch identities across devices and channels in real time. That’s why it’s becoming a core pillar rather than just an experiment.

These tools also inform curation by defining high-value audiences, as they give SSPs, DSPs, and buyers access to persistent, first-party-derived user profiles. This allows curated marketplaces to be built around precise audience definitions, on top of contextual signals or basic demographics.

Rob: So publishers should start building an identity strategy now?

Craig: Yes, especially as it complements contextual. Think of it as a hybrid approach: contextual for scale, identity for precision.

Rob: Got it. And you mentioned identity is linked to curation - is curation still a priority?

Craig: The late-2024 hype around curation has steadily subsided, with many publishers finding that curation partners didn’t deliver predictable lift, and sometimes eroded yield. The category is valid, but it’s not plug-and-play. You need control over your data and a clear ROI path. So yes, have a curation strategy in place, but keep pilots small and margin‑controlled; only scale if return is predictable and client demand justifies it. Essentially, keep it tight, track everything, and walk before you run.

Rob: Any other Start ‘ems before we head over to the bench to review the Sit ‘ems?

Craig: Retail media continues to be a priority for our clients as we head into H2. In fact, it’s predicted to be bigger than CTV, audio, linear TV combined in 2027. What’s more, research shows that programmatic infrastructure directly influences retail media budget: 96% of marketers would prefer DSP buying for retail media, while 49% say DSP access would prompt ad spend toward retail networks and publisher off-site placements.

So it’s important that publishers make sure they’re on top of standing up GAM360, but also the data, attribution, audience segmentation and reporting that goes into a successful retail media strategy.

Rob: Ok, what’s not making the H2 Start ‘ems roster, but needs to be considered ahead of 2026?

Craig: So we need to talk about AI and automation in ad ops workflows - two important yet distinct things, which have a place as both a Start ‘em and a Sit’em. The preface here is that if you have the bandwidth to give this some attention now, then it should not be a H2 Sit ‘em for your company.

Automation is the parent term to AI and other types of technologies and products: it means “I need to find a way to make this process better.”  That “way” might happen to incorporate a certain amount of AI. From a workflow standpoint, as an ad ops team, you need to think about what processes you need to automate, and then figure out if you can do it on your own, or which outside companies or platforms you should bring in to assist.

But if your plate is too full, or your dev schedule is jam packed for H2, you might not be able to deploy a new technology straightaway or modify your current process or workflow. That’s when it becomes a Sit ‘em - you’re not forgetting about it, but it instead becomes an H1 ‘26 agenda item.

So the Start 'em here is that there are opportunities for automation and efficiency today. The Sit 'em is the fact that you work in ad ops, so you’re never not busy, and automation and AI will still be there in six months. There's also so much noise around AI that you don't have to be deploying every new shiny object today. It will evolve, and new products and offerings will be available in H1 ‘26.

In Q4 ‘25, when everyone's trying to hit numbers and chase the almighty ad dollar, you’re going to see a lot of revenue coming your way. If your time is limited, you’re better off making sure your programmatic stack is set up correctly, you’re ready and able to tackle all of your incoming direct business, and your ad stack is optimized.

Rob: Great. Any other Sit ‘Ems?

Craig: Last time we talked about exploring different engagement channels including subscriptions, paywalls and events. Subscription and paywall strategies can still drive revenue, but only for publishers with strong brands, committed users or scale. Adoption remains selective, and many still focus on ad‑first models.

Rob: So in conclusion, good old ops automation is still evolving, with tech and AI promising to push deeper into workflows, and also into the monetization bidstream. We’re also not just talking dashboards and reporting - but QA checks, pacing reviews and revenue forecasts, too. But I think we can all agree that humans will still oversee the tech, and the decisions it makes. Like how a defensive Special Teams unit watches the whole football field.

Craig: Exactly. The machines need to be monitored, taught, updated and adjusted to keep up with any additional tech the publisher deploys within their ad stack, and more. Publisher ad ops teams that have already invested in automation are now layering on real-time alerts and suggestions, and well beyond that. You need to have proficiency in this to make sure the automation tech you’ve deployed is doing its job in the best possible way.  This is where outsourcing partners and tech providers are essential.

H1 2025 was about laying the groundwork. Now, armed with new insights and an evolved playing field, H2 is where your fantasy season kicks into high gear. Publishers should play to their strengths, keep an eye on the market, and make sure that both their ad ops and ad tech rosters are ready to deliver.

Rob: Before we go, last September, I said that in a year’s time, we may be able to automate and outsource many of the tasks on our plates today, and that 2025 would be the year of outsourcing. Is it? 

Craig: Yes it is. If you're not doing it yet, get help in cleaning up your stack, because it’s one of those things that is always going to be a Start 'em. If you look at a Fantasy platform’s weekly report, it won’t tell you which superstars to start or sit, because you start your superstars every week. As a publisher, automating as much as possible - whether that’s campaign management, monetization management, or reporting - is a given, and not something to question. You should always be finding ways to delegate and/or outsource whenever possible, especially when you can lean on external expertise to do it more efficiently. 

To learn more about how a strategic partner like adops.com can help you fine-tune your roster for H2 and beyond, connect with adops.com today.

This is content created in paid partnership with adops.com. We only feature partners who we believe bring real value to the publisher community.