Mobile programmatic IV.

Gábor Brindza

Managing Director

I’ve come to the end of my article series or at least I think so. What I definitely promise is that the finale will be like in tv serials, a tiny chance would be left to continue it.

In my previous post I’ve enlisted the difficulties and the situations we faced on the bumpy road of programmatic. This time I outline the development trends that have accompanied our programmatic campaigns. Most of the solutions could be improved, and many times we force topics which may not be commercially justified (e.g. viewability) but are vital to stay professional. The automated world is not perfect but knows many things better than the traditional media buying method. Here come – focusing on mobile – the topics we touch, and services we offer on the programmatic side.

  1. 30+exchange

Pretty much. Actually not the exact number is the most important on the first place, but the fact that any inventory is available in programmatic. Every marketplace is different regarding the formats, surcharged services, etc. We could state without concerns that almost any kind of content and advertising inventory is purchasable on mobile too.

  1. Location

I should say: location, location, location. Exciting topic I first came across wondering around in New York. For now, it holds rather the promise than the real business potential everybody expects to harness. Although it’s a typical trend in the digital world that market players grow by believing in things which are far from business potential, but carry the chance of it. So, maybe this scheme is valid for mobile and programmatic too, and especially for mobile programmatic.

So, location is a promise now. There is a lot more data in the USA, thanks to Yelp and also all restaurants, bars and hotels share their own data. Here, in (Central) Europe all we have is the data bases filled with location data coming from the telco sector and internet providers.

  •             By the way the situation is not so disastrous. We can target as deep as zip code level, and not only for users coming from wi-fi, but from 3G or 4G as well. The question is whether it’s good if the ad doesn’t appear when the audience is at that specific place.
  •             Then comes hiperlocal targeting. We give a 100 meters’ diameter circle and the ad appears for those who cross it and(!) consumes that particular content. It’s still not sure if enough people are looking at that content.

Two roads seem to realize in the future. One is the messaging apps (Viber, Whatsapp, Messenger, etc.). They see location too, and are able to send messages, but unfortunately they are not really good – let’s admit it’s not easy – at advertising. The other road may be if someone gathers all the apps in one platform, and the users after sharing their location with the app would receive push notifications through this platform, when they are around the given coordinates. Exciting things are on the horizon.

  1. Language

We had campaigns where we had to reach Hungarian audience abroad, or Slovakian Hungarians on Hungarian sites. Tricky one. The solution for the first case is to target on the language of the browser, because it’s usually the same as the nationality, and pair it with foreign location. For the second case Slovakian location and Hungarian content is the perfect match.

  1. Private deal

I have written about it earlier, and this’s one of the most promising topics. We’ve more and more of it, which means we could advertise for fixed price on Hungarian and foreign sites, for specified audiences. For that we need one – if we are lucky – code, in less lucky case a bit of prior consultation. It’s hard to tell whether private deals or non-guaranteed buying have a bigger share in our overall purchase, but it’s clear that the number of this type of deals is increasing.

  1. 3rd party tracking

I’m not the fan of putting another adserver code into the programmatic system, but it’s solvable. Thus, the main indicators could be followed from another adserver as well.

  1. Domain/subdomain target

My favourite solution. We can target on any specific content or url. The banner doesn’t have to be thrown into the whole big mess. The site list could be created mindfully based on target audience or reach. Another small step towards professionally acceptable campaigns.

We mostly serve our campaigns on list created in advance according to contextual goals or on the sites with the biggest reach, but in appropriate environment. Even more interesting and tighter targeting is achievable if we go further in an url. For example, it’s unnecessary to put a tech ad on the Guardian’s homepage, the tech column is far enough – plus mobile target. It’s not revolutionary, but not all the systems know the domain and the sub-domain target. The problem is that everyone focused their developments on reaching the most users, but not on reaching the most relevant ones.

  1. Viewability by Moat

We’ve already written about this big and deep topic separately. There is viewability measurement within the programmatic system, but we use an independent service provider. In our opinion Moat is currently the best at it, and helps a lot to evaluate our campaigns. It does not come cheap, but I think this is what it takes a programmatic campaign to be professionally founded.

I note where we saw a remarkable difference in the data, there typically were some glitches that after were fixed the viewablity data had improved. No surprise that in the western countries the CT-focused thinking is going into this direction.

And the above-the-fold targeting is only the icing on the cake, which is likely to drive better viewability data, even if this means shutting out many zones. This may not necessarily be the perfect solution, because the mid-article ad won’t appear above the fold, but probably works better than the ad in the top of the page zone otherwise suitable for the target.

  1. CPA optimisation

Back to the roots. Thanks to Google the use of automated systems was equal with the click-based solutions, and how to purchase an action for the cheapest price. The RTB based buying forces the potential advertisers to bid in real time. The base of the bid is appearance, thus only CPMs are purchasable. Who offers different system is likely to convert back the buying to CTs.

The point is there is a way that after a while – of course if our code is in the specific product pages – the system automatically let’s the appearances optimise according to purchase or conversion. The system learns what kind of ads on what sites would most likely to transform users into customers, and optimizes the appearance according to that. It’s worth it to let the machine do its job, it knows better than us.

  1. Rich media and HTML5 solutions

From the dawn of our operation we’ve separated special attention for rich media and HTML5 formats, and Celtra is still our reliable partner in creating them. We test every new solution whether is it working in automated systems, and if yes, how.

The problem is that the most exchanges subordinate the banners bigger than 50k. They win less bids against smaller banners. But in case of rich media banners their true size couldn’t be recognized by the systems. This is how we serve far bigger banners than 50k – thus, their quality is much better – and even video ads. This is our fastest developing department, and another contact point between the old and the new world.

  1. Retina/HD resolution

It’s a burning issue. First I awake in New York, when browsing on the big local sites low-grade and blurry ads came up. At the conference I attended was everything about location. In the afternoon workshop where they served beer I almost ask, what is the point of location if the banners are blur? I didn’t have the guts to ask it, the beer wasn’t enough, and probably they wouldn’t even understand the question. Only those understand it, who are in it. Recently I’ve run through the webpages of the top creative agencies in UK and US, and I was surprised by the non-mobile optimized versions. Mobile wasn’t taken into count during designing, or the page was created a long time ago, who knows. But the problem starts on this very basic level. Mobile web on most markets is catastrophic.

 The most depressing part is that advertisers are not aware how and in what quality their brand appears for the customers. Until they don’t realize it I’m not sure if it makes sense to think in strategy – especially not in mobile strategy. 

Anyway, we only serve razor sharp ads. We check them on many different devices, and the good news is the resolution is perfect on the programmatic channels too. We barely see this anywhere else. Really nobody does care about quality?

  1. 1 impression/page

The most common problem is the unlimited amount of ads on one page. We are able to avoid it but the assistance of the media owner is necessary. This setting has to be allowed in their SSP, and then it’s manageable, but only if the bid comes from the same system. If two separate hands do the bidding, it is inescapable. Big brand destruction may be prevented with this little setting.

  1. Brand safety

A great part of the above mentioned points improve brand safety either by specifying on which pages the banner appears or by increasing visibility. Besides that, in Adform other brand safety options are on hand. For example, we are not only able to tell where the ads were served, but we could tell that real people saw it, excluding the impression or CT generating robots.

  1. Heatmap

This’s the cherry on top. In the rich media systems used to be feature like this, so it was visible where the users clicked. It’s available now in the programmatic systems we use. It shows perfectly which part of the banner is the most clicked, and the accidental clicks could also be recognized. The gathered information could help a lot during the creative process how to build a mobile banner.

  1. Data

The latest big challenge. We are nowhere, but everyone is working on it with full blast. We have tested several partners searching for data sources. I hope we get ahead in the foreseeable future. The good news is we could get e.g. weather data from a database and target the ad to the given temperature. Not bad, but still a long way to go.

  1. Three years

I don’t know whether is it a lot or not. We have been working with programmatic for three years on mobile. Not much, because we haven’t found an answer for every question, too much, because we have run a huge amount of campaigns and have experienced a lot. The real reason it seems such a long time is because it was an intense period, we barely could keep up with the trends, not to mention going ahead them.

This is us and mobile programmatic. It has become a full round mini-series, and now I feel like I almost told everything that was in me. And who knows, maybe with a brilliant plot twist Bobby returns – as in Dallas – to continue the series.

December 27, 2016
2017 Rewind

It’s become a tradition now to summarize the best moments, do we have a lot to sum up this year again. Gábor has already shared his predictions for 2018 but for now let’s see what were we up to in 2017.

Technical development of the year: MCX

In 2017 we introduced our big hit, Madhouse Creative Exchange (MCX). This innovation allows even rich media formats to run on programmatic. Here are two …

Read it >
2016 Rewind

We have already listed our wishes for 2017, so now let’s kick off the new year with a compilation of 2016’s bests&peaks.

Read it >