Paulo Cunha is Co-Founder, CEO at ShiftForward, and will present at this year's ATS London. Here he discusses the paths that can be taken for those seeking to innovate and customise in ad technology, and how most will look beyond the monolithic stack as a solution.

For those not paying attention to the technical side of digital media, ad technology systems have come a very long way since their inception in the 90s. Particularly fueled by the recent Real Time Bidding architecture, the scalability requirements for most systems have grown by orders of magnitude - businesses now have the desire, and commercial ability, to evaluate peaks of hundreds of thousands of ad delivery opportunities per second (Queries per Second in industry jargon) in a relatively quick manner. Before RTB, this level of scale took years of business development to reach, if at all.

However, its not just the scale that changed: the depth has substantially increased too. Media businesses now expect an ever-growing number of features, as standard, from their platform; and this trend will continue, as more and deeper functionality comes out of research. Dynamic creatives, behavioural and socio-demographic targeting, mobile ads, social, contextual, predictive targeting, measurement and attribution, ad exchanges, forecasting, automated yield and inventory management, RTB - to name just a few - were, not that long ago, unproven ideas in which only visionaries invested. When their use reveals positive impact on ROI, they get promoted to 'must-haves' and go into every ad tech platform’s unspoken list of 'standard features we must implement before we die'. Problem is, that list grew. A lot. The result is ad technology functional complexity now being on par with the most advanced software engineering projects being executed today, in any industry.

Consequently, today's innovations are about building relatively small improvements across the stack, or digging deep into very specific focused areas. Complexity has evolved so far that only very few, deep-pocketed, companies are able to build an entire and complete ad tech stack from scratch; and even then, building the whole stack may not be feasible, and is definitely not cost-efficient. Although in some situations only using third-party technology isn’t an option. It may not provide enough room for customisation, for innovation, for control over the technology or for preventing vendor lock-in. Using a combination of multiple third-party technologies can only go so far in mitigating these issues - although when integration is secured through app marketplaces and third-party APIs it can definitely help. Still, without some form of technology ownership, the potential for the extraordinary is seriously limited.

So, if developing all the pieces is out of the question, and using just third-party platforms doesn’t fit the bill, is there a middle-ground? A combination of selected third-party technology, with key components kept internally, might be the answer. The good news is technology vendors are making it easier by moving away from closed product models to more open and mashable designs, to be used and integrated by other technology developers.

This means building and controlling an independent platform doesn’t have to be done completely from scratch. Common ground is being established in ad technology. Projects like OpenRTB are providing standards - and not just on the real-time bidding interfaces - that help technology developers know what is expectable from their and other systems, from the design of taxonomies to advertising business entities such as 'impressions', 'banners', 'devices', 'sites' or 'segments' and their relationship. It’s a combination of common ground like this, together with API-driven product architectures, that enables a stack to be put together using multiple technologies and vendors with reduced effort. (an API is a way for one piece of software to interact with another piece of software without having to know how it actually works).

Even though not all technology vendors have this open design approach yet, some of the harder and most commonly used ones do exist and the scope will broaden as more businesses realise the power of technology ownership and customisation, and vendors realise this as a huge opportunity and not a threat.

The RTBkit, open-sourced by Datacratic earlier this year, is a good example of this trend. It implements a fully functional real-time bidding software framework and can reduce the time and effort of development, scaling and maintenance of a real-time bidder. Vendors are invited to build their own plug-ins to attach to the RTBkit to enhance its functionality, and Datacratic has already made their own optimisations plug-ins available for licensing.

Google is following the same steps and is sharing their own bidding framework, the Open Bidder, for developers to use. Ad serving companies have increasingly focused on expanding the provision of better APIs - some even place 'Full APIs' as their top differentiator, like Adzerk. Closer to home, ShiftForward has released a complete ad forecasting solution designed to be easily integrated through APIs with any ad platform’s back-end, with more stand-alone modules planned for the future.

Another big innovation in ad tech of late has been the concept of the 'app market'. It has already made its way into various different types of technologies - from large ad exchanges to smaller start-up ad servers - and is a way to overcome the complexity challenge and foster some degree of independent innovation. Companies using these platforms can build their own features on top of the platform or purchase functionality from tech providers who sell their wares on these markets.

In many cases, this is a win-win-win situation: sellers can focus on specific feature sets and find access to demand; media companies can easily acquire and integrate new functionality; the platform increases its stickiness and proposition value. Trouble is, there are limits on how much innovation is possible to achieve within one closed ecosystem. At some point, having an independent, owned and operated technology stack may be the only way forward - for both technology vendors and media companies. Although many won’t find the need to go down this route, others may find their very first customisation requirements are only realised this way.

Put that together with a flexible, low-maintenance infrastructure from the likes of Amazon and there hasn’t been a better time to claim a stake in ad technology innovation.