Tom Barnett is the Managing Director at Switch Concepts. Here he discusses the importance of the customisable real-time platform when trading ad inventory at the impression level, the huge potential for publishers to increase yield, and the critical piece the decision engine will play in all automated trades.

It could be argued that the online advertising trade has been stuck in a perpetual bear market since the dot.com bubble burst. That is to say that the overall average clearing price of inventory (ecpm) has been in a fairly steady state of descent. This is not to be confused with the fact that the online advertising market has grown year on year in that time – I am talking particularly about the price at which an advertising impression, in all formats, trades. It is easy to argue that this is a very clever ploy on the behalf of advertisers, which, whilst not entirely untrue, is not the major factor. I think the major factor has been widespread inefficiency in the marketplace.

The online advertising marketplace is really very young by comparison with its older brother, the financial markets. And we are only just beginning to see it start to take some form. With this in mind, it’s more than reasonable to expect that in formative years things do not run like clockwork. Some of the contributory factors to the inefficiencies of the last decade included:

- Lack of clarity around the actual cost of an impression. With so many participants trading on the same inventory, on the same impression even, there was a kind of upper and lower quartile effect on the actual “price” of inventory. Result, play it safe and bid low.

- Ungraded oversupply. The online population increased, but volume of inventory increased more – and nobody was really sure what was good and what was bad inventory. The old adage “I know 50% of my marketing works, I just don't know which 50%...” could never be truer (although it might be more like 10% of it working in this case). Result, buy in bulk but buy cheap to be on the safe included:

- Protracted decisioning processes. The classic “ad call”. Either just straight poor decisioning or multiple linear decisions by different servers before it’s decided which advertiser gets the slot. Result, bid low, you have no idea what you are getting.

- Fragmented supply of good inventory. Even when supply is almost certainly good, it has been fragmented before it gets to market – premium and remnant – and is available on secondary and tertiary markets. Result, can be bought cheaper elsewhere.

- Investment in bidders. More recently advertisers did get much smarter. Real Time Bidders decision properly and take advantage of weak supply. Result, the Quants had an easy time beating price down with maths.

There are probably more, but it doesn't really matter now. Things are changing.

I am predicting a bull market in the “teenies”. The emergence of RTB as a protocol is a critical precursor to better communication, and with better communication comes greater efficiency. This does not mean the advertiser is simply going to be paying more – it means that both parties at each end of the transaction are making a better contribution and getting a better return.

It is better to view this in terms of the Real Time Market than RTB per se. The Real Time Market is a place where buyer meets seller on a per impression basis, communicates effectively and then shares in the decision making process. It couples perfectly with the blooming Data Management Platforms to add value at the event of transaction and provides a more acceptable level of transparency to all parties around price and quality.

The demand side has clearly led the way in the last few years and is by and large enabled to make real time bids now. At Switch we believe that as remotes of supply, publishers, enable themselves to trade effectively in real time we will see a reversal of fortunes for the inventory clearing price. This relationship between supply and demand is an enormously exciting one with huge scope for exploration and collaboration. It is now the turn of the publisher to build as much sophistication into the “ask” as the advertiser has into the “bid” - enter the Real Time Ask. And we're nurturing it.

To unlock this value, we at Switch, believe it will be necessary to have a customisable real time platform that enables publishers and networks of supply to holistically manage all of their inventory – all campaign types, all advertising media and all devices. Effectively, we built a real time bidder and an asker into the 'adserver' - so even direct campaigns are delivered in competition with real time bids. Direct premium sales can be managed alongside traditional network deals and, most importantly, real time asks for prices from any real time bidder – be it an SSP, an exchange, or directly to a DSP or ATD.

The choice of who to connect to is really down to the Publishers' own strategy and leverage – which will naturally evolve as they find their footing in this new marketplace. First and third party data can be overlaid to asks, direct delivery and reporting to add value to every impression. Yield management is fundamental – we've seen increases of over 30% to overall revenue where our engine has been employed even before going to the real time market. Forecasting is bespoke to the requirements of the inventory and client. A live overview of campaign performance is available at any time and campaign delivery pacing is programmable.

Fear of the Real Time Market is completely understandable, but with the right tools in-house, it is the route to a better quality of transaction and better returns for everyone. Like any new model, we need to engage gradually and sensibly – but a future where every impression is traded in real time looks to me like a future where everyone is getting more of what they want out of the market.

Dual screen behaviour, smart TVs and a rapidly onsetting fundamental interconnectedness of everything are all a reality – so inventory needs to be managed in the same way – holistically. The financial markets are using big data to analyse sentiment gathered from social media towards stock prices. How long until we are picking up consumer trends in the same way and targeting against them?

The delivery engine is critical – it must operate on a certainty principle rather than probability and priority. It is the platform that brings the capacity to act as an exchange to partners that use it and is compatible with everything. And each instance is standalone – you need deploy a new instance per customer. And each instance has to be complete with a big data factory and a local noSQL memory facility. Switch is now delivering these tools to help publishers realise the potential of the real-time market, and we believe this strategy will help bring us out of the trough of disillusionment.