In Brief: 4 takeaways from Boston Consulting Group’s recent programmatic survey

Boston Consulting GroupIn their recent report on the profitability of programmatic advertising sales, Boston Consulting Group identified four ways in which the publishers are outperforming all rivals. The most successful publishers:

  • Use cross-channel, data-driven strategies
  • Segment and match inventory with the right buyers
  • Assemble the right technology
  • Build strong go-to-market and analytic capabilities.

The report argues that many publishers fail to incorporate programmatic sales as a core element of their strategy, or to properly manage their programmatic teams. In the worst cases, programmatic specialists are spending less than a quarter of their time creating value. As the programmatic market continues to grow, these firms risk eroding their market share, thus revenue, thus long-term profitability.

According to BCG, programmatic advertising is a $9 billion market and growing rapidly. Programmatic buying automates the process of identifying where you want to advertise (based on consumer traffic, browsing preferences and network reach) and buying advertising space at auction (up to a set maximum bid).

It’s the optimal way to keep pace when opportunities to advertise and consumer behavioural trends are emerging and being sold in real time, and shifts some of the data management burden from human staff onto automated systems. However, the systems need to be properly selected, constructed and deployed in order to achieve their full potential. BCG identified the following key elements in doing so..

Cross-channel, data-driven strategy


Successful publishers have a cohesive and well-formed data strategy. Their management and monetisation of first-party (i.e. data about their own audiences) governs the strategic choices these publishers make. Marketing strategies and pricing structures are based on concrete successes evidenced and supported by data.

A data-centric approach demands rigorous and clear planning, establishing the extent to which programmatic advertising will supplement or replace direct sales. Take eBay’s recent ‘Programmatic Only Week’ for instance. The experiment—where 100% of the online giant’s UK advertising was booked and executed programmatically— was opened to all media buying outlets and allowed the team to collect vital data for future campaigns.

Anna Stoyanova, head of programmatic EMEA for Essence said:

eBay’s ‘programmatic-only week’ focused on buying through auction. During the week we saw a massive increase in supply relevant to demand, but we haven’t seen a devaluation of the inventory, which is really encouraging for publishers and brands.

This is the kind of data that only comes through planning and testing.

Right inventory, right buyer

Building a body of reliable data also means that publishers can match advertisers more accurately with the audiences they want to reach. A publisher who knows their advertisers’ needs in great detail, and who can refresh and extend their inventory when an opportunity arises, can provide accurate targeting of consumers.

Staying close to buyers and customising programmatic sales of targeted inventory can yield six times the CPM of conventional direct sales.

Assembling the technology

The technology involved in programmatic buying requires careful choices to create the right ‘stack’ of software and user choices. The stack is based on ad-serving technology which delivers the sales. Decisions in which sales to deliver are supported by programmatic demand sources and tools. Both delivery and decision-making software must be configured to maximise CPM and fill rates, and these configurations must be specific to each channel and platform in which the publisher operates; one size does not fit all.

Finally, the stack is topped off by data management/output tools which check the effectiveness of the stack as a whole. In general, discrete stacks will need to be built for each area of a publisher’s inventory, although some ad servers and programmatic buying tools work across platforms and provide integrated input and output for different stacks.

Strong capabilities

Technology is useless without the skills to configure it, assess its output and make the decisions it recommends. Hiring programmatic sales specialists and data scientists is a start; integrating them into an existing team is better, since it gives the specialists access to your existing client relationships and thus offers more developed insights into your advertisers’ needs. The human factor is vital: data will not analyse itself, nor will new methods sell themselves to advertisers.

The bottom line

As trends emerge and platforms proliferate, automated identification and selling of advertising opportunities will become essential to keep track of opportunities and trends, and therefore compete in the market. The gap between direct and programmatic sales will gradually close, and publishers who want to stay in the game will be well-positioned for the predicted explosion in programmatic sales – up to 83% of the online market over the next two years.

The Basics: Audience Buying

Last week we looked at how publishers can take control of their own first party data in order to keep control of the sales process, extend their audiences, and offer additional value to marketers..

This week, let’s take a step back and revisit the path from buying media as a proxy for an audience, to being able to buy that audience itself.

Audience by proxy

RocketFuel's Audience Extension
Graphic from RocketFuel’s Audience Extension report.

Traditionally, advertisers were only able to target their audience indirectly, by trying to estimate who would be interested in a particular product or service based on a generalised demographic. This has supposedly helped advertisers to determine when and where to advertise—which publications, which venues, and what kinds of ads.

For example, want to advertise house cleaning products? Advertise on daytime TV or in women’s lifestyle magazines. Want to sell a Rolex? Target a wealthy and ‘masculine’ demographic in publications such as Forbes, or via outdoor on Wall Street and in the City of London. These may be somewhat dated examples but what they exemplify, apart from the reinforcement of age-old stereotypes, is a form of advertising based on certain assumptions about different kinds of people.

The idea is that, since you can’t directly buy the attention of a particular individual, you buy ad space where that audience is statistically more likely to be. And in many people’s minds, that is Advertising 101.

The rise of audience buying

The promise of digital for marketers has always been the increased ability to focus advertising messages to reach ‘the right person at the right time’, and over time this has indeed become more precise. But for a long time, digital marketing never fully managed to live up to this promise. Advertisers have still had to rely on publishers to reach audience segments that they deem statistically more likely to buy.

In recent years, though, things have started to change. We’ve seen the rise of ‘audience buying’: purchasing access to these audience segments directly rather than buying proxies for audiences. Thanks to the shift to digital and the increasingly sophisticated ability to track and store user behaviour, we have the ability to detect users’ activities, and to predict their interests based on patterns of behaviour.

In theory, this fulfils the promise of right time, right place, right person, right everything. The hyper-specific targeting of audience buying is more precise and more cost-effective, ensuring that consumers see ads that are far more relevant to them, and reducing wastage on ad spend for marketers. It also allows for a far more accurate attribution of ad spend, by making it clear which ad placements are truly driving engagement. In other words, it gives marketers full transparency, letting them see where their spend is going and with whom their message is resonating.

Fragmented industry

Audience TargetsThis shift has turned a hundred-year-old advertising industry on its head. It means that advertisers aren’t necessarily reliant on measurement and ratings companies to present them with their approximate audience segments, because they can create their own, collecting data from multiple sources and advertising anywhere that users can be detected and tracked. As we explored with the advent of publisher trading desks, publishers are also striving to reframe their own worth in this new context by exploiting their own rich user data, which they can sell within the audience buying model.

In some ways, though, this has led to such a fractured advertising landscape that we might wish for simpler times. The biggest problem is that the new and old models aren’t necessarily compatible, and yet many advertisers still try to use the old one and think audience buying can simply be tacked on.

For example, traditional TV advertising still mainly follows the old model while digital TV allows for much more direct audience targeting based on metrics like impressions and time spent with ads.

“The cost of using such different methodologies of buying selling and measuring TV and video is catching up to media companies that straddle both worlds,” says Lorne Brown, CEO of Operative. Brown continues:

Buyers and sellers often have separate media planning teams, technology stacks, sales teams, contracts and billing processes for the two worlds. As digital grows, it will not be sustainable to have two parallel processes. It also won’t be effective for advertisers, which will increase their expectations of measurable and effective cross-screen campaigns.

A cohesive model

At the same time, it might not be wise to ditch contextual targeting altogether because it can serve a different purpose. While audience buying is focused on the pre-existing behaviours of users whose data and habits have been tracked, marketers may still want to throw a wider net for growth on a larger scale.

Malcolm Coles of The Telegraph (UK) puts this in terms of ‘driving scale or becoming niche’:

a publisher has to choose between wanting to drive general engagement with their site and their brand, seeking growth over time, versus attending to the pre-existing audiences they have for, say, Telegraph Sport, or even a dedicated Arsenal FC page.

According to a white paper from Xaxis titled Harnessing Big Data for Audience Buying, a cohesive advertising model should have several ‘audience tiers’, ranging from ‘demonstrated affinity’ at one end (‘the consumers who have demonstrated through action a predisposition for the brand’), to ‘inferred affinity’ at the other end (‘the consumers who have not yet expressed affinity through direct action, but may gravitate towards the brand based on broader data such as demographics’).

The stronger the affinity with a brand, the smaller the size of a group, which means that targeting only users with proven affinity—in other words, only direct audience buying—may be insufficient for growth on its own. A good marketing plan, therefore, needs a cohesive model that encompasses both a hyper-targeted audience and a broader contextual reach.

The Basics: Publisher Trading Desks

In the traditional advertising model advertisers and their agencies buy advertising space from media owners and networks. They do this in order to reach an audience they think will be interested in buying their products and services. Simple. Increasingly, however, these relationships are becoming increasingly murky as advertisers morph into media owners and media owners become buyers. One way the latter is taking shape is through the formation of so-called Publisher Trading Desks.

As the advertising and marketing market shifts from trading on media properties as a proxy for audience and toward a more audience-centric trading paradigm, publishers are finding new ways to leverage first party data. By doing so publishers can capitalise on their own audiences, to give them the competitive edge. The premise is simple: publishers can leverage the rich data from their own pre-existing audiences to advertise more effectively, and not just on their own sites, but on third-party sites too.

As such, they can charge a premium for the use of their proprietary data, something to which most advertising companies simply don’t have access. So clients come directly to the publisher and pay the premium to have their product or service advertised across the publisher’s trusted affiliate network.

So how does it work?

PTD Simplified Model
On the left is the traditional buying model. On the right, the publisher moves up the chain, closer to the buyer, to trade the publisher’s own audience across both their own sites as well as on the open exchanges.

A publisher sets up a trading desk using what was traditionaly buy-side technology like MediaMath, DataXu, AppNexus or one of the other Demand Side Platforms (DSPs). This tech allows them to better manage and monetise their audience data. Previously, this would have required advertising middlemen, who would have taken a chunk of the revenue, but the publisher having their own trading desk makes this unnecessary. The actual purchasing of inventory is still done through real-time bidding, but the publisher leverages their proprietary data to target their ads.

This works particularly well for vertical or trade publications. For example, a publisher such as The Guardian, which has its own trading desk, will have rich data on its own readership, segmented into different interest groups. They will be able to look at the data for their travel section or their business section, and can analyse user behaviour to determine which campaigns their readership has been attracted to, which devices they tend to use and other demographic data.

This data then informs the publisher’s advertising on third-party sites—determining the nature of the ads, who to target, and on which of these third-party sites it would be most appropriate to advertise. It can also help to determine which devices to optimise for, such as smartphones and tablets.

But it doesn’t end there. When the publisher buys inventory on these third-party sites, these advertisements then collect their own data through cookies, which in aggregate creates an even richer pool of data.

A better proposition

This audience extension creates a much better advertising proposition overall, by not limiting publishers to the ad space of their own websites, and by allowing them to build on their already rich datasets. It also allows publishers to build mobile advertising even if they have no mobile inventory of their own, or video advertising if they have no video inventory.

For advertisers, there’s an element of brand safety guaranteed, as publishers can bank on their own reputations and those of trusted affiliates. Publishers themselves have the opportunity to work with their internal marketing teams to promote to their own audience beyond their website and across their network, increasing the reach of campaigns and promotional events.

However, a publisher trading desk is not without its challenges. The publisher will first have to prove their value to advertisers and other buyers. Publishers will be required to take on all the usual responsibilities of advertisers, which means that they will need to be savvy with their data, build in-house trading expertise where there is none, and source and train on new platforms where they have limited experience.

In other words, while there are many apparent advantages to publishers setting up their own trading desks in order to regain control of their own advertising and boost their own revenue, publishers will still require a thorough, fully-baked data strategy for it to work effectively.