Emma Wilson is Managing Director at Harvest Digital. Here she discusses the agency's trading desk, why ad networks are still important to the eco-system, the evolution of the agency model and competing with the big agency networks.

Harvest set-up its own agency trading desk late last year. What was the motivation behind this? Why would an independent agency, like Harvest Digital, needs its own in-house capabilities?

Harvest is at heart a performance agency and we're very familiar with auction-models from our large involvement in search marketing. So setting up our own trading desk has been a very natural step.

We announced last year that we were working with Appnexus, but we actually started working with Right Media back in 2008, so we are not at all new to the space.

How does any agency in-house the media optimisation function over the space of three months - when for the last 15 years it has been outsourced to third party buyers?

We've been buying and optimising digital media for ten years - which of course is longer than most networks have been in existence. Our trafficking has always been in-house, and at the heart of our approach to media is an ongoing optimisation process once a buy is in place. We have never walked away from a plan once it's bought, although obviously you do hear about that happening in the market.

So whilst optimisation may be done by a publisher or network, we will more often than not be driving this process, suggesting improvements, asking for changes. Some technical aspects of publisher-side operations are of course new to us, but we are very familiar with the basic principles.

Has this had a knock-on effect on the amount of ad networks you work currently work with?

No, not really. We are big fans of ad networks, so long as they are adding value, through technology or unique inventory or whatever. And actually, it would be arrogant of us to say that simply because we can buy through our own trading desk, that means that we can deliver every aspect of what networks can do.

However, if you took the example of a network that is simply buying inventory on the exchange, marking it up and packaging it up to us, well yes, that's a business model that is vulnerable. But we wouldn't be working with a network like that in the first place.

As a more general point, whilst we value the contribution of networks, we prefer to work directly with media owners as this gives us more granular control. Of course, that is more work for us - but our clients see the results.

Now that you can in-house functions like retargeting, what will you be looking for in terms of third party buyers? Will prospecting be a key criteria going forward?

We don't have fixed principles about this - clearly retargeted media can perform exceptionally well, but equally we appreciate that we need to share data with some partners in order to make tools like lookalike models work effectively.

Customer acquisition is the core of what we do, but actually we expect that to change. As we get access to richer client data, we can start to look at optimising to lifetime customer value or customer reactivation.

With independent agencies moving into ad trading, is the pure-play DR ad network model one step closer to extinction?

Our view is that ultimately whoever is offering the best performance to clients will win. So long as a network is adding value, it will survive and thrive.

Turning the question around, this is also a new world in which specialist digital agencies play an important role. Clients simply aren't going to share confidential sales or targeting data with half of Soho - agencies can act as trusted intermediaries.

What kind of skill set is an agency like Harvest now looking for in its new hires?

Same as ever - we hire bright numerate people who can think for themselves and have an entrepreneurial spirit.

Those who appreciate restrictive processes and suffocating structures will much prefer the pace of life at network agencies.

How important will the data strategy be in making automated buying work for the agency? Will Harvest be looking to build a DMP to execute campaigns through its white-labeled DSP solution?

With our search experience, we're very comfortable with highly data-rich environments. And one of the lessons we've learned from search is that behavioural targeting is almost always a much stronger signal than demographic targeting.

With this in mind, the challenge for us is how we use clickstream and purchase behaviour to influence our buying strategies. Instinctively we think that first-party client data is going to be a better thing to focus on than third-party data and we have some interesting ideas on how we can make this work.

As a full-service agency, we're bound to say that actually one of the most valuable sources of data which is often ignored is good old-fashioned web analytics. We see our goal as delivering the full value of digital, and hooking media up to web analytics lets us optimise to research and engagement as well as purchase.

How will you try to differentiate yourself from other agency trading desks on the market? And how will independents like Harvest compete with the holding companies?

The big networks have definitely been much faster than us in producing slick sales decks - and we're not alone in thinking that the Powerpoint has often preceded the product. To paraphrase a recent blog post from Chris O'Hara, an industry that depends on under-paying 22-year-olds to manage multi-million pound budgets is going to struggle in this brave new world.

If you look at the UK search industry, according to the last NMA league table 25 of the top 30 search agencies are independent. Independent agencies are consistently delivering better search performance than network agencies (ask Google!) and we confidently expect the same to be true of RTB.