Schibsted reduces cost-per-action with programmatic subscription offer

By Thea Kristin Hansen

Schibsted Norway

Oslo, Norway

Connect      

By Siri Holstad Johannessen

Schibsted Norway

Oslo, Norway

Connect      

Programmatic is just one grape in the cluster of technological media innovations — with varying life expectancy — that marketers have witnessed. Despite this, the method as we know it today survived for a decade. Congratulations, programmatic!

In the world of media, 10 years can be compared with a young child. Little did we know what challenges would arise during the juvenile years when programmatic was born. One thing is for sure: Fortunately, adolescence has been a bumpy ride as smooth rides often put spokes in the learning wheel.

The attention paradox

Programmatic marketing is an automated way to buy ad placements based on criteria like willingness to spend, segments with which you want to communicate, where you want to reach people, and specific KPIs. Direct ad placements used to be bought based on a gut feeling, but you can now buy and optimise ads live with Artificial Intelligence and data.

This opens the possibility of serving personalised and relevant ad messages on an individual level. It also likely means every receiver is bombarded with relevant messaging in the digital sphere — in an ocean of providers trying to reach specific audiences.

In other words, reaching relevant target groups at the right time and place has never been easier. But generating the attention this method offers has never been more complex.

This is a paradox in today’s attention economy, where the human attention span is increasingly short and the battle for attracting interest is growing. Giving receivers something valuable and making them feel something has never been so important. The need to create emotional impact in the noise of ads is no longer a nice-to-do factor but a need-to-do one.

What we expected

This paradox is something we experienced in our upbringing of the programmatic child. When we first started doing programmatic marketing at Schibsted, we saw a unique possibility to increase volume by promoting sales of our quality newspapers in every channel to every target group in every funnel mode. This was revolutionary; sales were expected to come rumbling in and we would reach sensational numbers of new buyers!

On that note, we built campaigns with a three-way strategy that reflected the overly optimistic expectations we had for programmatic:

  • Phase 1: Broad audience targeting. Message: sales promotion.
  • Phase 2: Retargeting those who visited a subscription form but changed their minds (“form abandonment”). Message: sales promotion.
  • Phase 3: Broad audience targeting plus retargeting segment. Message: “last chance” sales promotion.

Our experience

You know how some stories are too good to be true? We’re afraid our journey with programmatic is no exception. Early on, we felt the last-click measurement tools slap their thighs, laughing at our expectations when they revealed the harsh truth: 16 direct sales in three weeks.

Today, it’s a known truth that last-click attribution is somewhat worthless. Programmatic measurement tools that actually took into account the attribution model showed far more positive numbers. We could see direct sales were low, but the model indicated the sales from prospects that had seen or clicked on a programmatic message were far higher than first expected.

Conversions took place in numerous channels where we promoted campaign offers simultaneously with programmatic campaigns — electronic direct mailings, native ads, paywalls, telemarketing etc. – that happened in our own environment and in our own channels.

Because very few people bought directly from programmatic ads and the plausibility they were exposed to the campaign offers in our own channels was minimal, was it a waste to use the programmatic budget to only show off our campaign offers? What if we used programmatic to expose our exclusive content, to lower prospects’ threshold to buy our product in our other channels?

With our portfolio of subscriber-driven newspapers, we offer quality content we take pride in, and we would be fooling ourselves thinking the price in itself would be the only factor increasing the number of subscribers. Programmatic is not a channel in itself, so shouldn’t we stop treating it accordingly?

Schibsted broke its programmatic campaign into three phases: paid, owned, and earned.
Schibsted broke its programmatic campaign into three phases: paid, owned, and earned.

Closing the gap between expectations versus reality

Using this as backdrop, along with this paid/owned strategy from the ad trinity, we chose to alter our three phases in the following campaigns: 

  • Phase 1: Broad audience targeting. Message: editorial messages with editorial articles/content behind paywall as a landing page. Attract, expand, and nurture.
  • Phase 2: Retargeting those who visited via a subscription form (i.e. considering filling out the form from article paywall) but changed their minds (“form abandonment”). Message: sales promotion or yet another editorial article behind paywall. Attract and convert.
  • Phase 3: Broad audience targeting plus retargeting segment. Message: “last chance” sales promotion. Expand, attract, and convert.

By altering this, only prospects in the right mode — those who took an interest in our products’ content — would be exposed to the campaigns’ tactical part.

Did it drive results?

We compared the last 20 campaigns of our top four brands (Aftenposten, Bergens Tidende, Stavanger Aftenblad, and Fædrelandsvennen) over the last year. This was a good mix of campaigns containing both the old and new strategies.

What we found was quite exciting. By comparing campaigns with the old strategy (using programmatic to show sales promotions) to those with the new strategy (exposing great content and minimal price messages), we saw a significant decrease in cost per action (CPA) (conversion, in our case).

  • Aftenposten: 38% decrease in CPA.
  • Bergens Tidende: 31.3% decrease in CPA.
  • Stavanger Aftenblad: 57% decrease in CPA.
  • Fædrelandsvennen: 28.4% decrease in CPA.

When it comes to programmatic, you can add as many buts and ifs you want, and free yourself from drawing any conclusions. There are obviously other factors at play such as intuitive order forms, call-to-action (CTA) buttons, formats, complementary channels, banner attention, price offers in the retargeting phase, time of year, target groups, and more.

But, based on the 20 campaigns we ran over the last year, we see significant indications that our strategy has been an influencing factor in lowering the CPA.

To illustrate the mentioned campaigns with the new strategy, here are two case studies.

Case 1: Fædrelandsvennen — “Byen vår” (our city)

Fædrelandsvennen timed its campaign to the release of a great editorial investigative article. The article revealed social inequality in different parts of the geographical catchment of its business and target group.

Phase 1 teased key findings from the investigative article.
Phase 1 teased key findings from the investigative article.

  • Phase 1: Rotating, dynamic banners disclosing some key findings from the investigative journalism with a “read more” CTA button. Landing page: matrix promoting all the articles related to “Byen vår” behind a paywall.
  • Phase 2: Retargeting. Rotating, dynamic banners disclosing more key findings from the investigating journalism, with a new CTA promoting the name of the investigative journalistic disclosure. Landing page: matrix promoting all the articles related to “Byen vår” behind a paywall. Audience: Those who visited a subscription form in phase 1 but didn’t finish.
  • Phase 3: Retargeting. “Last chance.” Rotating, dynamic banners promoting a sales offer. Retargeted to those who visited an ordering form in phases 1 or 2 but didn’t finish.
Phase 3 of this campaign promoted a "last chance" offer.
Phase 3 of this campaign promoted a "last chance" offer.

We see some weaknesses. For instance, the phase 3 banner isn’t aesthetically coherent with phase 1 and 2. However, the campaign gave some great results. Compared to other Fædrelandsvennen campaigns, it generated a 38% decrease in CPA, making it the lowest CPA we’ve ever had on a FVN campaign. It also had the biggest share of post-click conversions, which can indicate that more people were interested in signing up for a subscription after clicking on the banners and seeing the matrix of “Byen vår” articles behind paywall. The percentage share of post-click conversions was twice as much as the post-click share on the other campaigns.

It was a great success.

Case 2: Stavanger Aftenblad — Editorial retargeting

In this campaign, we chose to promote great journalistic craft. One part of the campaign promoted a great feature article, “The Neglected,” behind a paywall. Prospects who showed interest in the article but were not already subscribers were retargeted with more investigative articles about the health system and criminal journalism, also behind a paywall.

The other part of the campaign promoted editorial sports articles and used a segment of people interested in sports based on Schibsted’s data. Phase 1 linked to a matrix of articles about general sports in Stavanger. The people who visited these articles but did not buy a subscription were targeted with sports banners and an editorial page about Viking, the local soccer team in the area.

Visitors in phase 1 saw an editorial page about a local soccer team.
Visitors in phase 1 saw an editorial page about a local soccer team.

We completely dropped the focus on price in phase 2 and did not start the retargeting until phase 3, where there was a “last chance” message during the campaign’s last week. In this phase, we reached those who had taken an interest in our two editorial phases but hadn’t yet converted into subscribers.

This campaign gave us the lowest CPA we have ever reached at any of our four newspapers, with a 57% decreased CPA compared to the campaigns that an emphasised price gave us.

Our take on the programmatic future

As mentioned, we mistakenly expected programmatic to be the gold mine of direct sales. We cannot claim we’ve found this gold mine, but we are confident we’ve at least found the shovel that gives us the opportunity to slowly dig toward it.

But to dig a straight tunnel, it is important to watch our step and let conclusions be temporary. We have to keep focusing on building our brands by showing our content, but we also need to test our strategies regularly. We must have both the acceptance and budget to make — and learn from — mistakes.

We also want to draw attention toward the last corner in the trinity related to nurturing. We will do this by including existing subscribers in our brand-building phases and not exclude them in the whole campaign as we’ve done up to this point. This way we can feed them with reminders about the quality of our subscriber-exclusive content.

Practical implications based on findings

This case can be transmissible to other businesses and newspaper vendors. Some companies measure results only by conversions and forget to build their brands. This either/or philosophy makes a false dichotomy. In light of this, we measuring brand-building editorial campaigns by CPA is somewhat ironic. But this case showed it is possible to do both simultaneously.

Your company can utilise this strategy by doing tests promoting your broad product portfolio, quality products, written content, or whatever is relevant for your business, instead of promoting low prices to your audience, regardless of where they are located in the sales funnel.

We know it can be tough to defend high budgets and low conversion rates to different stakeholders and departments (analysts, financiers, external customers, or the company’s objectives and key results). But, as we’ve seen, building brand decreases the CPA both on a short-term basis and likely in the long run.

Programmatic is 10 years old this year. We know the challenges will be tougher in the years to come. As the method enters its digital teens, a bunch of hormonal facets will emerge such as disregard, shorter attention span, and tracking, among other things.