This is the second installment of our programmatic ‘ask the experts’. The other two are ‘How best to track offline impact of programmatic spend?‘ and How to integrate your programmatic and TV ad strategy?”
Let’s get on with it…
There’s a constant debate about targeting – what data sources are proving most valuable?
Sameer Modha, head of data at DigitasLBi:
Most of the third-party data vendors are just regurgitating ad-carrying site visitation, with some bits of statistical mystery-meat lobbed in for good measure, by over modelling from small data sources. In many cases it’s homeopathic targeting – massively diluted and not very potent – but that points to the bigger issue which is that no-one is measuring anything properly.
Clients generally do impressions and clicks, and some direct online sales, and forget about audience and brand, which means there is no quality signal in the marketplace for data beyond last-click and that just locks in a dire race to the bottom of the funnel where we chase efficiency but lose any broader business impact.
Daniel Gilbert, CEO, BrainLabs:
Almost any programmatic campaign we run tends to involve a combination of first, second and third-party data.
The more first-party data the better, but many businesses can’t rely on this alone. Second-party data is more bespoke and tends to be reliable, but it’s expensive and often too granular to achieve the reach most advertisers need.
So third-party data tends to be necessary in most campaigns. There’s a hell of a lot of it available, but at the moment it’s impossible to be certain about its value before you actually use it.
A valuable data source is something you have to work out through trial and error. Having a robust process of testing and learning in place is the key. In the early stages of a campaign, you just need to keep running audience composition reports, comparing the audience you targeted with the audience you reached, and work it out from there.
When GDPR comes into effect next year, advertisers who rely on third-party data can look forward to much greater transparency, which will hopefully create a clearer picture of the true value of a dataset. For the time being, test and test again!
Joel Livesey, director of partnerships for EMEA, The Trade Desk:
In the world of data, there’s no right or wrong for the entire market – it’s up to each brand to pick the right data. First, second and third-party data all have different advantages, and are valuable when used at different times.
First-party is incredibly beneficial for its accuracy and robust nature whereas third-party data is fantastic for reaching economies of scale. And reaching this scale is far more important, for example, for a haircare brand marketing shampoo than for an auto company selling a sports car, who would need to be far more precise with targeting.
Third-party data can enable economies of scale (important for shampoo, but probably not sports cars)
Jim Hawker, owner, Threepipe:
We follow data rather than let it follow us. What I mean by this is that there are situations when location is the most valuable and times when traditional audience targeting prevails, with the near limitless targeting opportunities in between, the key is to not go in with a bias so that you can judge what is working rather than dictate it.
This doesn’t mean you should test everything, go in with a logical approach to initial selection so you can start facing in the right direction. Keep in mind that what works best will be specific to the market and client and won’t last forever so stay true to the data.
Alessandra Di Lorenzo, chief commercial officer, Media and Partnerships, lastminute.com group:
The most valuable data sources are those that are closest to buying intent – so, in travel, that would be booking data, for example. But it can be difficult to derive valuable insights from the mountain of data available to us today. We know that complicated doesn’t always equate to good, so we like to turn our data and insights into sizeable chunks to make them actionable. After all, simple information such as destination of travel can do wonders for the performance of a campaign.
The future will be about predictive capabilities, but keeping it simple will still be key.
Kristina Kalpokaite, head of paid media, Summit:
Utilising first-party data is an obvious choice – these data sets are unique to our clients and ultimately bring a competitive advantage others do not have access to.
Applying a layer of third-party data on the first-party segment can further enrich the data – albeit the advertisers need to check which audience segment they wil be paying for so that you don’t inflate the cost unnecessarily – choosing the right DSP (demand-side platform) with a suitable targeting mechanism helps in addressing this.
What are the most important current innovations in direct response?
Jack Glanville, programmatic analyst at Journey Further:
The arrival of Google Attribution in Q1 2018 is really promising for programmatic as a channel. Set to be a more mainstream solution, it should give small and medium businesses the opportunity to invest more confidently in programmatic, starting small and proving impact before scaling. In addition, Facebook Advanced Measurement will measure impression data from DoubleClick, adding another data source to the mix.
Other innovations revolve around attribution modelling itself. It’s concerning how many agencies and publishers still attribute 100% of post-view and post-click conversions to an impression which a user may have seen up to 30 days ago. We often experiment with time-decay attribution models, which weight prospecting and remarketing activity based on the length of time since the initial ad impression.
Following on from the targeting debate, most of the important stuff is in the measurement space. The only way to move beyond direct response ‘Just So Stories’ which are made-up examples of superficially plausible targeting, is to measure properly (multiple tests, control-exposed, with particular scrutiny of the synthetic controls, as there are dodgy ones around).
The measurement kit from Google and Facebook is generally ahead on this and will become more so over the coming months.
Blockchain validation of publishers is something we are watching closely and would be one of the most anticipated innovations for us. This has the potential to radically reduce fraud and increase transparency within the supply chain which will mean less wastage, cleaner data and therefore better signals.
Taken together this application of blockchain would have an overwhelmingly positive affect for our clients and is something we are eagerly investigating.
It’s not necessarily an innovation, but a big step forward for the industry has been advertisers finally accepting that they can move away from the click as a way of understanding the success of a campaign. Recent studies have shown that click-through rates can be misleading and skew data the wrong way. Take Natural Born Clickers for example, which revealed that 8% of the internet’s audience is responsible for 84% of the clicks, meaning brands had been disregarding 92% of the internet when pinning success on clicks.
Attribution modelling and post-impression attribution, as well as using cross-device data providers, are much more effective and give advertisers the ability to gain more insight into their direct response campaigns. Now, we can see the impact that one medium is having on another – think desktop versus mobile – and how the customer journey connects with each.
Dynamic video is becoming more widely adopted, allowing personalisation to be applied to an asset that was previously very expensive to adjust for different target audiences.
Adtech providers are starting to share user data (See: Appnexus Summit earlier this year), thus aiming to rival the user ID reach of Google and Facebook, with potential to dent the market duopoly.
Cross-device retargeting and omnichannel attribution for store visits and revenue are becoming more widely adopted – moving closer to closing the loop between media investment and revenue sources.
Getting to grips with media planning? Why not take Econsultancy’s Digital Media Planning training course?