With programmatic inventory costs escalating due to factors such as made-for-advertising sites, invalid traffic, and various other conditions, demand partners are now seeking new solutions.

In this guest post, Mintegral’s General Manager for the Americas, Jeff Sue, examines the options for demand platforms to access supply in 2024 as they look to tackle rising costs. Sue details the advantages and disadvantages of three feasible options: ad exchanges, working with demand-agnostic exchanges and building your own SDK to integrate with supply partners.

It’s getting tough out there for demand platforms, as they face the rising cost of the supply of programmatic inventory. Rising levels of invalid traffic and MFA (Made For Advertising) sites, together with the sheer number of intermediaries inserting themselves into the bidstream and taking their cut, are pushing up the cost of inventory. A June 2023 Association of National Advertisers (ANA) report revealed that MFA sites account for 15% ($13 billion) of annual ad spend, with the percentage of programmatic auctions coming from MFA publishers estimated to be around 20%. The same report also found that the average programmatic ad campaign runs on an astonishing 44,000 websites – not so much a long tail as a never-ending one.

So, what are the options for demand platforms and the costs associated with each? To my mind, there are three, each with its own pros and cons, so let’s look at each in turn.

Option 1 - Work with the ad exchanges

Ad exchanges offer rapid access to inventory, enabling demand platforms to get up and running quickly with minimum effort. They are a well-established route to market, offering demand platforms quick and easy access to millions of impressions, with a wide choice of ad placements and the ability to control spend and frequency.

But this ease and speed of access come at a price - take rates are high for ad exchanges, averaging around 20%, sometimes even higher. Precise figures are hard to come by, as the fee varies significantly from one exchange to another, so in some instances, it’s much higher. In addition, pricing is often dynamic, so it can be challenging for demand platforms to plan for their cost of supply.

Additionally, pricing is not always transparent, so demand platforms may not know what they are paying for access to premium inventory. You may be content that you’re buying on Candy Crush through an exchange, for example, but before you get too complacent, you need to know what that access is costing you. 20%? 30%? Maybe even more. So, it's not as simple as saying: “I have access to Candy Crush, job done.” You need to test out multiple exchanges and constantly diversify and refresh your exchange mix to find the lowest cost of supply.

There are two other downsides to working with exchanges. The first is that it’s easy to fall into the trap of spreading your demand too narrowly across a limited number of exchanges. While it involves more effort, demand platforms should constantly research and expand their partner list to diversify their supply portfolio to ensure they are accessing the best supply at the best price. It’s important to test multiple large exchanges to see your spending and effective costs per bundle ID.

Finally, some exchanges often offer inferior renderability for more sophisticated ad formats like video and playable units. This is because most ad exchanges also run their own ad network and tend to focus their time and resources on optimising their own demand or ad formats first.

Option 2 - Work with an SDK-less mediation platform

The second option is to integrate with an SDK-less mediation platform like Amazon TAM (Transparent Ad Marketplace), Amazon UAM (Unified Ad Marketplace) or Nimbus. These have the advantage of lower take rates - 2.5% for Amazon TAM and 10% for Amazon UAM. These fees are charged to the demand partner at the end of the month and technically, the publisher gets 100% of the bid. They are also easier to integrate with as there is no SDK to build, integrate and maintain. Additionally, in some instances, Nimbus, for example, has no demand side fees, and they do not have their own demand sources. This means there is no potential conflict of interest. They are simply incentivised to get as many demand platforms on board as possible and serve the ad to the highest bidder.

A potential downside of these platforms is that you are relying on their SDK to do all the rendering that you'd need and to be able to handle all the ad formats that you need. This is an issue for demand platforms that need to run video ads, as some of these SDK-less mediation platforms are display-focused, and their solutions and capabilities for ad formats beyond straight display are much more limited than what a typical exchange or your own SDK can offer. You may also have to consider the amount of scale or overlap, when compared to a traditional SDK-mediation platform.

Option 3 - Build your own SDK and integrate with supply partners

The third option for demand platforms is to build their own SDK and have it integrated directly into apps. This usually happens through a mediation layer, which needs to certify your SDK and enable bidding functionality. The advantage here is that mediation layers typically charge a much lower bidding fee of around 5%. This fee is also charged at the end of the month, and technically, the publisher gets 100% of the bid. Or if you opt for the waterfall model, there are no fees, though bear in mind that the waterfall is widely seen as older tech with inefficiencies and has been deprecated by big players like Meta and Google. But the bidding fee is totally transparent – it’s written into the contract. In addition, the demand platform has its own SDK, so it can control the creative rendering rather than relying on an external exchange’s capabilities in this respect.

The downside is that creating an SDK, integrating it with supply partners, and building an SDK network take a significant amount of time, resources, and money - years rather than months and millions of dollars.

To give an example, let’s say you’re an advertiser with a $1 CPI, buying on Candy Crush through an exchange. So, after the exchange takes its 20% (give or take a few per cent), your buying power is now effectively 80 cents going to Candy Crush (assuming no other intermediaries taking their cut). In the same scenario, if you build an SDK, the SDK has a direct relationship with the app developer, and if you can get yourself integrated, you’re only going to be paying 5% as opposed to 20%, so your buying power is technically $1 (or $0.95 when you factor in the bidding fee you’ll have to pay later).

But this doesn’t happen overnight. You would need to hire a business development team to sell the SDK to King for Candy Crush, convince them that it’s worth pulling engineering resources off other projects, like IP updates, or creating a new level for the game, to integrate the SDK, put the necessary bidding adaptors in place, and go through the due legal process, until one day - probably a year or more from when you started the process - you wake up, and voila, you have an SDK integration. Oh, and repeat that same process for every other app developer you need to integrate with in order to get to the ROI figure you need to hit.

So, before taking this route, the demand platform needs to do the maths and determine whether the money saved on the take rate and the additional revenues they can make by being more in control of their own destiny justify the time, effort, and investment required.

Finding the most value

So which way should a demand platform turn? There’s no one-size-fits-all answer. It depends on where you are in your programmatic journey and what you’re aiming to do. Do you need to access supply quickly and are prepared to pay for it? The traditional exchange is probably the way to go for the scale and proven performance.

The other option is the SDK-less mediation platform, where you can diversify your supply mix and reduce costs by leveraging Amazon TAM, Nimbus, or one of the other SDK-less mediation platforms. They may not have all the ad formats or ad rendering capabilities some demand platforms need, but they do offer lower take rates and easy integration.

Or do you have the time and the resources to build a business development team to create direct relationships with publishers and can see the ROI in doing so? Then build your own SDK network.

Each option has its unique value in the ecosystem. It’s perfectly feasible to run your demand simultaneously through the exchanges, an SDK-less mediation platform, and via your own SDK, which is exactly how we work at Mintegral.

Edited by Paige Cook