VP & Head, Performance Marketing
Neo@Ogilvy is one of the premier media buying agencies in the world, under parent company GroupM / WPP. Can you share more on your role as VP & Head of Performance Marketing at Neo@Ogilvy and what it entails?
My role at Neo@Ogilvy is to partner with brands in helping understand and realize measurable business ROI from their digital investments. This primarily entails developing right digital KPIs and marrying them to business KPIs, building & optimizing a multi-channel investment model focussed primarily on biddable media and ensure we are delivering to target business goals be with Cost Per Acquisition, Return on Ad Spend etc.
You run all kinds of performance media across channels like search, display, video, and social. How do you think about cross-channel attribution, and assigning proper credit to upper-funnel tactics beyond search?
Cross-channel attribution is big challenge and is often a function of analytics infrastructure. Sophisticated brands have better tools in place to allow for cross-channel measurement while a lot of brands don’t have the right analytics platform to glean cross-channel insights. Having said that, my teams try and do the best with the information available. This includes analyzing and testing the conversion data against few attribution models such as Linear, Time Decay models from tools such as Google Analytics to get directional guidance on the role each channel plays in the customer purchase journey. The same is also supplemented by analyzing conversion path reports facilitated by ad servers (e.g. DCM) to understand which channels over-index in influencing the final purchase. Doing the same over the years has also helped build some historical truths that we then validate with real data. For instance, it is fairly common to see Social tactics playing a huge role in starting the customer journey (First Touch) even though they may not appear like a profitable channel from a last click (usual attribution model).
Paid Social is without a doubt a critical channel primarily because of the increasing time customers spend on these platforms. So, for brands looking for customer engagement these channels are critical. However, paid social as a performance marketing tactic often suffers from simplistic “last click” attribution model.
Twitter as a paid social tactic proves its usefulness primarily on specific campaigns such as product launches, webinar/whitepaper downloads due to the presence of fairly decent targeting options as well as the presence of informed consumers using the platform for consuming relevant content. However, Twitter for me has lacked in delivering a continuously evolving ad platform both in terms of last-mile (conversion focussed) ad formats as well as exposing its data hose in a sophisticated fashion.
LinkedIn as a paid social tactic is something I haven’t explored a lot from a B2C standpoint but it obviously lends itself very well from a B2B advertising standpoint with its unique targeting options (Job Title, Industry etc.) and some of the ad formats including InMails etc. I am probably not as informed as I should be but I have found LinkedIn.
Facebook on its part has done extremely well in exposing its user data set to advertisers as well as building some very interesting ad products (Facebook native Lead Generation format for instance). However, the challenge for me has been a lack of sophisticated third party integrations with measurement tools. This leaves us having to rely on Facebook platform data around conversions etc. which does not re-concile very well with third party measurement tools such as Ad Servers or Web Analytics platforms.
What are your thoughts on custom audiences for Facebook, and tailored audiences for Twitter? Is there look-a-like targeting superior due to the 1st party data they have on users?
Custom audiences have been a welcome addition to the mix and has given a user-friendly way to engage with existing prospects/customers. The results for that matter have been promising as well and we have seen anywhere between 25-30% reductions in CPAs for custom/tailored audience across Google, Facebook and Twitter. We haven’t seen significantly strong results from look-a-like based on custom audiences yet.
Speaking primarily from a regional perspective (India), these two platforms haven’t come alive yet in terms of advertising. However, I do see Pinterest doing really well from a performance marketing standpoint because of ‘buy worthy’ content and relevantly high intent.
There is a lot of display ad buying happening through programmatic channels. What do you see happening with publisher direct ad buying in the future? Do you think much of it will be done through private marketplaces, potentially even for mid-long tail publishers?
Publisher direct is already moving through Programmatic channels at a rapid pace especially in developed markets like North America. Adoption in APAC markets might be relevantly slow owing to relatively lower understanding of programmatic but it is only a matter of time. Publishers are stressed for margins and advertisers also would warm up to unifying their display buying. Private marketplaces would allow increased efficiencies on both publisher & advertiser side while potentially improving yields for publishers as well. Mid-long tail publishers might be slow to adopt as they grapple with the dilemma on increasing direct sales v/s the cost of the sales force but would eventually come around. The private marketplace itself will evolve and like provide more incentives to publishers.
What role do you see agency trading desks playing in the future? Typically, is all biddable or RTB inventory bought through an agency trade desk, and more custom executions done through the media planning teams?
Trading desks are already under a ton of scrutiny from a transparency standpoint. While I do not see trading desks losing their flavor for the next 5 years, the reality is they will continue to be the source of all biddable or RTB inventory as well as programmatic/publisher direct for standard ad formats. Custom executions such as Takeovers etc. would continue to be done through media planning teams but there is no saying that the DSPs wouldn’t evolve to allow for custom executions in the near future.
How do you approach view-through attribution, and what type of window do you typically use for clients in each vertical?
View-through attribution is a tricky one to get right and there is no specific attribution window that can be applied across clients across verticals or even within the same vertical. The general rule of thumb is to have an attribution window tied to the product sales cycle. For instance, for a vertical with low-ticket value consumer item the view through attribution could be as a low as 12 hours to 24 hours whereas for a long sales cycle vertical such as Auto or High-Tech, it could be as long as 30 to 45 days. The way to approach is to apply a data driven hypothesis and take a test & learn approach.
When it comes to performance-based programmatic buying, does placement transparency really matter, or is it more about hitting a clients’ CPA / CPL / CPS goal? How do programmatic platforms allow you to optimize towards these performance metrics?
This is the beauty of performance-based programmatic buying! When you are evaluating performance basis a transaction metric such as CPA, CPL the net result is that you end up serving inventory of ‘good’ placements only and the non-performing one’s get weeded out. Having said that, there are brand safety filters that we put in place to not show up on questionable inventory sources and most DSPs offer robust ways to put in brand filters. A lot of DSPs now also offer granular placement and creative level reporting to ensure campaign managers can manually weed out placements or inventory sources which are not helping hit the performance goals.
What types of 1st and 3rd party data are you able to leverage through programmatic buys? Which types of data have been most effective for your clients?
Access to brand specific DMPs allow us to leverage 1st party data including CRM data, website, mobile app data as well as in some cases offline sales data. In simplistic scenarios as well, website based 1st party data is immensely useful which also helps create look-a-like segments. In terms of 3rd party data, reputed players such as Lotame, eXelate etc. provide quite a few data segments that we are able to leverage. In terms of performance, 1st party data is obviously very effective and it is a matter of just increasing the volume as well as the quality of that data. With respect to 3rd party data, it differs a lot from Geography as well as client vertical. For instance, 3rd party data sources from the likes of Lotame etc. have delivered strong performance when I ran it for some of the North American clients. The same unfortunately does not work as well in say the Indian market which could be a function of the data being not as rich as it is say for the North American market.
It was announced earlier this year that Google launched ‘programmatic guaranteed’ buying through Doubleclick Bid Manager. They have a huge advantage due to the ubiquity of Doubleclick for Publishers amongst publishers, who can now list themselves in the Doubleclick Marketplace with the flick of a button. Do you think Google stands a chance of really dominating the direct ad buying market? Why, or why not?
No question about it! There is every reason for publishers to adopt it because as you rightly mentioned, they are already invested with the Doubleclick solution. It is now just a matter of switching a button. Having said that, I doubt Google will take a very aggressive approach with this approach and will likely roll it out in a slow & steady manner. With this move, Google stands to monopolizing the display advertising market (Google already dominates the search market landscape) and is likely to raise quite a few eyebrows. The other potential reason would be the revenue sharing model wherein Google is not known to offer the best rev share to publishers. Publishers would probably continue to shop around and work with partners who offer a better rev. share even if a unified Doubleclick approach seems more efficient.
Lastly, what is your take on the future of performance-based programmatic media buying?
It is here to stay and is only going to increase for good reasons. The massive increase in consumption channels (and the corresponding consumer data that is generated) coupled with increasingly personalized consumer want would call for a real-time data driven media buying approach and that is the entire premise programmatic media buying is built upon.