Instead of simply buying media for marketers, agencies must offer guidance on whether their clients are ready to adopt programmatic and, if not, provide help on how to do so.

This should include data strategy and operations setup, says Vincent Niou, Essence's Asia-Pacific senior programmatic and partnerships director, who warns that agencies that fail to offer these services will provide opportunities for consulting firms to jump in.

In this Q&A with ExchangeWire, Niou further explains how agencies can facilitate the development of programmatic amongst marketers and publishers in Asia-Pacific, as well as answer calls for greater transparency.

ExchangeWire: How would you describe the level of programmatic knowledge among Asia-Pacific marketers, and publishers, today?

Vincent Niou: It varies by market. Some like Australia are fairly advanced, while others like India are still in relatively nascent stages. Overall adoption is a factor, but we should also examine how marketers think of programmatic within their channel mix. Marketers in nascent markets tend to approach programmatic as a DR (direct response) driver, accessing mostly remnant inventory; whereas those in more advanced markets see its value for brand campaigns, also.

Likewise for publishers, those in more advanced markets position programmatic as a premium proposition within their sales waterfall. With the onset of header bidding, some even have programmatic competing directly with reservation inventory. On the other side, less developed markets still view programmatic primarily as a means to move cheap or unsold inventory.

Another factor is sales reps in certain markets aren't all incentivised to trade programmatically yet, thus, presenting barriers for prospective buyers. Markets such as China and Korea have large walled gardens limiting open access to inventory and data, as well as ops challenges including limited third-party ad serving.

There's also the question of inventory quality, fraud, and ad verification tools by market. It's all generally improving and moving in the right direction, but at different speeds and starting points.

What are some key challenges they continue to face in their adoption of programmatic?

I've touched on many of these above and they can be broken down into three main areas:

Vincent Niou, APAC Senior Programmatic & Partnerships Director, Essence

Buy-side
- Adding programmatic without first building the necessary foundation. These include having a validated data foundation, organised optimisation approach, and sound measurement and analytics. Without these, marketers could miss the full benefits from programmatic buying and the ability to properly measure success. This, in turn, could lead to under- or mis-evaluation of the channel.

- Treating programmatic as direct-response only, thus, losing out on its potential for brand initiatives.

- Lacking the ops or technical expertise to troubleshoot issues and work through challenges, especially in developing markets or getting the practice off the ground.

Sell-side
- Suppliers that are hesitant to make their inventory available on exchanges or only positioning programmatic at the bottom of their waterfall. If publishers see programmatic as a means only to move low-quality remnant inventory, its potential is greatly diminished.

- Sales reps not being incentivised to sell media programmatically or not receiving commission in the same way they would when selling direct media. This creates obvious barriers for prospective buyers.

- And even when reps are commissioned to sell programmatic, they may lack the technical or subject matter knowledge to properly answer questions or field requests from buyers.

Marketplace & ecosystem
- Poor inventory and data quality. This is directly tied to sell-side challenges. The higher the quality, as well as availability of inventory and data, the faster the rate of adoption.

- All markets have walled gardens, but the extent of monopolisation varies. Generally speaking, the more suppliers with strong inventory and data that make their offering available, the higher the adoption.

Why is it still tough to ensure data quality and accurate cross-device measurement today?

First, we should define what we regard as quality. Is it measured by what the data says it is? Or is it based on use case and performance? Segmentation also plays a factor and it's difficult to make broad generalisations. For example, while first-party data is generally a marketer's most prized dataset, there's no guarantee 100% of it will be useful. Conversely, while the provenance of third-party data can at times be questionable, certain segments can be useful, if leveraged correctly.

In working with publishers, I'd encountered instances when I would ask for the full menu of data, but they would not always provide. Perhaps this is because they are afraid the data may not deliver as labelled.

With regards to cross-device measurement, on a high level, there are two methods for tracking users across devices: deterministic and probabilistic. Deterministic matching leverages known user data – generally, a consistent login across devices such as email. Players that use this include the likes of Facebook, Google, Twitter, and BAT (Baidu, Alibaba, Tencent) in China. Probabilistic vendors use an algorithm drawing from a variety of signals including device type, location, operating system, and so on. They use these to create a device graph to infer statistical connections between devices. These vendors typically use a seed set of deterministic data to train their model over time.

The challenge is that deterministic players, whilst offering both accuracy and scale, operate as walled gardens. While powerful, their tools can only be used within their ecosystem. On the other side, probabilistic solutions can cover the rest of a marketer's activity, but device graphs are often blackboxes. This makes it difficult to ascertain accuracy and precision between different vendors.

There's also the question of privacy and ethics. Technology can now identify individual users across devices, but where do we draw the line? How granular should we get? Is the answer black-and-white, or does it depend on consumer attitudes, which also can vary by market? And what happens as the Internet of Things (IoT) continues to expand and more device types are added?

Data quality and walled gardens – which would you say posed a greater barrier to the progression of programmatic, and why?

The two are not mutually exclusive. Walled gardens exist because these players recognise the value of their data. It's no secret that data is the new oil and it makes sense for it to be heavily guarded. Given this, it's important that publishers and vendors outside the walls provide marketers with quality alternatives. Above all, they should leverage data judiciously and responsibly to respect the user privacy and experience.

Where do you see agencies playing a role in facilitating the adoption and development of programmatic amongst Asia-Pacific marketers, as well as publishers?

Upstream as consultants, and downstream as media operators. Regarding the former, a hot topic today is the entry of consulting firms into digital marketing. I'd argue the opportunity exists because it's a service that agencies have not traditionally provided clients.

A media agency's typical bread-and-butter is tactical planning and buying. In today's ecosystem, however, agencies need go beyond to provide upstream services akin to strategy and consulting. It's a larger conversation, but, in this context, agencies should evaluate whether a client is even ready to add programmatic buying to their channel mix. Do the clients fully understand their target audience and the lifetime value of each customer? Are their sites and apps fully tagged for measurability and data collection? Do they have the right attribution model in place to properly assign credit to channels? These are just a few questions that agencies should help marketers answer. If the answers are largely no, then let's start there. If the answers are largely yes, then we move downstream and provide expertise as DSP operators and programmatic media planners.

For publishers, agencies should communicate client requests and provide feedback to help shape product. This is especially relevant in developing markets where publishers are still developing their programmatic offering. On a more technical level, agencies should also call out instances of impression fraud or undeclared parties in the supply-waterfall. A specific example would be if reporting had an unusually high investment going to a URL known to have limited scale. This could be an instance of a middle-party using a wrapper to label unknown impressions as belonging to said publisher and reselling at a premium. These are fake impressions and we can call that out to the publisher to help solve the issue.

There've been calls for greater transparency in the industry, including agencies. Where do you see room for improvement here on the agency's part?

Clients need to know where their dollars are going. Whether it pertains to inventory or data, they deserve peace of mind that their investment is going where they think it is. If there are other costs, these should be disclosed as well.

As marketers grow more savvy, they are less satisfied with traditionally opaque trading desk solutions. According to a report from the World Federation of Advertisers, nearly two-thirds of marketers are uncomfortable with the conflicts created by trading through agency trading desks. For our part, Essence has operated a fully transparent media model from day one, and the trend has been playing out for some time. For example, GroupM now offers a transparent alternative with mPlatform and Havas recently launched their fully transparent model.

What kind of processes and governance can agencies adopt internally to help ease concerns about transparency?

As discussed above, a fully disclosed model that explicitly calls out all costs including data, inventory, and service, would go a long way in easing concerns surrounding transparency. In synergy, clients should also be asking and expecting this from their agencies. As these asks have increased in recent years, so have the number of transparent programmatic models.

How should agencies ensure they remain relevant amid the increasing presence of tech vendors and consulting firms, such as Oracle and McKinsey, in the ad tech space?

Agencies must provide value beyond just media buying. Within an ad tech context, they should provide clients guidance whether they are ready to add programmatic to their channel mix rather than moving to activation without first looking. If the client is deemed unready, agencies should act as consultants to educate and equip to move the client towards readiness. This can include ops setup, data strategy, audience segmentation, and platform selection. If agencies don't move to provide these services, opportunities are ripe to be snatched by the consulting firms.

What challenges do you face today as a digital media planner and programmatic advisor for marketers in Asia-Pacific?

With the differing levels of programmatic knowledge and adoption by market in this region, a central challenge for agencies is calibrating their strategy based on the state of each market. For example, it might make sense for a planner to provide a recommendation largely comprised by programmatic channels in a market like Australia, given the available inventory and data in that market. However, the same client in Thailand may require a higher percentage of reservation media as the inventory in question may not be available programmatically yet. A one-size-fits-all approach does not work in Asia-Pacific.

Where do you think the next generation of programmatic systems need to deliver to better address issues that remain today?

Two challenges that have been present since the dawn of programmatic are access to inventory for buyers and maximising yield for sellers. Header bidding has helped with both; but has also created a number of new issues as publishers have increased the number of demand sources in their tech stack. As a result, exchanges are sending more impressions per second and DSPs must evaluate more queries than ever. This, in turn, strains the ad tech infrastructure leading to increased costs.

In addition, publishers are still leaving money on the table as multiple second-price auctions can't bid against one another. On the other side, advertisers are losing inventory opportunity because their bid on one platform is enough to win an impression, but the second price sent to the publisher is beat by another buyer.

To solve this, some in the industry advocate a holistic unified auction to increase market liquidity and decrease strain resulting from too many demand sources. I agree with this assessment. We're still a ways off from a complete solution, but many companies are already offering partial solutions as the next phase of progress.