To make money, digital publishers run ads. The stronger the brand, the more advertisers are willing to pay for access to the publisher’s audience. Success comes with strong brands and large audiences.
Maximizing monetization means making the most of millions of page views — each representing the potential for multiple ad requests. To keep pace, publishers need to close deals and place ads at a dizzying rate. Approached through manual means, it’s simply impossible; which is why programmatic ad brokerage has become so universal.
While programmatic advertising automates many of the market interactions, it doesn’t automate everything and it certainly doesn’t optimize the process for maximum monetization. Plus, the scale of ad activity made possible with programmatic brokerage brings its own set of challenges.
Once monetization issues are detected, the full flow of revenue cannot be restored unless and until these 5 boxes are checked:
Satisfying all five of these demands can be a slow and grueling process. Below, we delve into each one and explain the complicating factors.
One might think that validation would already be satisfied as part of the detection process. Sadly, that's not normally the case.
To ease the burden of detection, publishers typically turn to rule-based alerts and data visualization dashboards. But these tools operate on a mathematical level, not a business level.
They lack context-awareness and they don't understand how the data relates to the business.
As a result, their validation is purely a function of statistical significance.
But even when something is statistically significant, it's not necessarily operationally significant. Blind to this distinction at the point of detection, publishers are left to deal with a lot of false-positives.
Working through that list of potential issues to sort the meaningful from the trivial is obviously important, but it can also be very time-consuming.
When data is reviewed more or less in the order it’s received or arbitrarily stacked, you can spend just as much time working on minor issues as you spend on those issues that really matter. That translates to money left on the table. Which is why prioritization is so important.
Without seeing the whole field at once and having an idea of the risk-reward presented by each possible play, you’re incapable of making an informed decision.
To prioritize tasks, you need to be able to systematically and quantifiably compare them. That means understanding the financial and operational implications. Ultimately, you'll want to assign a dollar value to each issue.
Unfortunately, to assess the bottom-line impact you'll sometimes need to scratch the surface of an investigation. Which is why proper prioritization can take up a good chunk of time.
Even after validating and prioritizing an issue, it still needs to be traced to its point of origin. This means investigating the data to pull apart all the different entanglements and dependencies to arrive at an understanding of what's really going on.
In this respect, the standard set of tools and technologies are of very little use. Instead, this type of analysis work is conducted manually by humans experts.
Monetization professionals — understanding the data context and business landscape — are the ones to tasked with getting to the root of the matter. And their root cause investigations can span hours, days, or even weeks.
The process is typically slow going and gobbles up valuable skilled (wo)manhours. All the while, the issue continues to rob the business of revenue.
Just because an issue can be influenced doesn’t mean it can be controlled — or at least not with perfect precision and responsiveness.
Because publisher monetization is so dynamic, multilateral, and multivariable, there’s normally more than one factor contributing to a given trend.
Ask yourself these questions:
The answers to these questions should be weighed in view of one another. In the absence of a deep monitoring solution to automatically detect, validate, investigate, and prioritize issues, this step can pose a real challenge.
More often than not, this is something that AdOps gauge informally on the basis of gut feeling and prior experience/knowledge. It's not a very scientific approach and it leaves a lot of room for error.
Now it's time to plan your attack. Of course, it’s not just the plan that matters. Ultimately, it comes down to the execution. Sometimes it’ll be simple and straightforward. Sometimes it won't be.
This opens up an additional and often considerable time gap. Sometimes even when the precise actions required are perfectly mapped out, they’re still delayed by human bottlenecks and competing interests.
It shouldn’t be very hard to see how all these steps could add up to a lot of time between an issue’s discovery and its resolution. The consequences of that gap can be harsh.
Consider for example that one of your ad units is exhibiting low viewability. Suppose that the issue costs you $1,200/day. Running through the whole process of validation, prioritization, investigation, consideration, and remediation, it can easily take three days to put the problem to rest. Over the course of those three days, you’ll have lost nearly $3.6K.
Now imagine that there are a couple different issues like that every day. Over the course of a month, you’re looking at more than $65K of lost revenue.
Consider a case where you’ve noted low eCPM for ad unit “MREC_banner_middle” when served to US audiences on mobile.
Now you need to understand the impact. You crunch the numbers for the affected properties, platforms, and inventory, and you conclude that the issue is costing you some $800/day.
But what’s the root cause? You investigate and as you work through the usual suspects — Impressions, Requests, Fill — everything seems to be in order. As you dig deeper, you notice that AdX’s win rate is down.
Not only that, but its CPM is also down. You continue to poke around and eventually discover that a long-running and high-CPM Programmatic Guaranteed (PG) campaign recently ended. The loss of that campaign dropped AdX’s bid participation and the overall CPM for the unit.
Where does that leave you? Obviously you can’t control demand outright. At the same time, you’re not altogether without recourse.
This is a perfect example of how solvability is not always black or white, but more often grey.
You can bring the issue to Sales and have them do what they can to rustle up a new PG deal. You can adjust AdX flooring to try and boost auction participation and increase competition. You could integrate new demand networks to compensate for the AdX drop. Or you could mix and match these approaches.
Still, none of these options are guaranteed to work. And even if they do work, it will take time for results to be felt. Worse still, some of these remediation strategies can actually backfire to further harm revenue.
Due to its questionable solvability, this issue probably shouldn't jump to the front of the line. But how do you prioritize against other similarly complicated issues?
Worse still, by the time you know enough to put the issue on the backburner, you've already lost hours down the rabbit hole; hours you could have spent on other more solvable stuff!
This is the pain felt by publishers all over the world. Even when everything is handled with care, it’s still a slow and inefficient process.
Because each step of the process relies on manual reviews that can only be undertaken upon the completion of the step prior, revenue leaks persist long after detection.
If, however, publishers were able to automate the process and run through steps 1-4 in parallel, the gap between detection and correction could be virtually eliminated. That's what oolo does.
Using deep monitoring to contextually and operationally analyze data, oolo simultaneously detects, validates, and investigates the issues affecting publisher monetization. When something actionable is found, oolo sends an alert that includes prioritization, root cause explanation, and suggested follow up — putting you in position to act quickly and effectively.