Native advertising’s main metric is arguably engagement, often measured as active read time. For native articles published on Bonnier News’ own content management system (cms) we use Burt’s measurement for this. This is not to be confused with “average time on page in Google Analytics” where reading times are always far longer. We could indeed have made life easier for ourselves by simply adapting this more advantageous metric but we really believe that the active read time provides a much fairer value.
In the chart below we can see the spread in dwell times across our differens news sites. On the vertical axis we have the active read time in milliseconds* (no less). The shape of the figures actually say a lot about how native advertising is consumed on the different sites. Di Digital, a daughter publication of Dagens industri focusing on the digital economy, only uses “continuous scroll”, meaning natives are automatically and intelligently loaded into the reader’s news feed. This increases reach but cuts the average dwell time significantly.
We can see that most articles have a dwell time of less than 25 seconds and that no articles are read for more than a minute. The figure for DN (Dagens Nyheter, Sweden’s main daily) looks entirely different; it is at its widest around the 1-minute-mark, but has a similar shape throughout. The figure is also quite tall.
Quick exit (QE) is exactly what it sounds like. In Bonnier News’ different reports and dashboards it is defined as a page view lasting less than 5 seconds. The quick exit metric obviously has a strong relationship with the average active read time since a high proportion of QE pulls the average down.
The biggest determinator for QE is the means of distribution (see below). But there are also inherent differences between the sites which we attribute to varying reading patterns. To a certain degree QE is also affected by the article post in the news feed or even headlines and image choice.
* 25 000 ms = 25 seconds
Native advertising is a relatively “narrow” product. Whereas display advertising is effective for conveying a simple message to many native excels at explaining and storytelling. That is why engagement is so important. At Bonnier News we have two main types of distribution for native advertising:
The classic way of distributing native content is via posts in the main news feed from where the reader clicks her way to the article. The native article can be published in the cms or externally, on a so-called campaign site. Campaigns thus distributed have - relatively speaking - fewer page views but longer read times.
A newer way of distributing native content is using continuous scroll, that is, automatically loading native articles after an editorial article to promote further reading. This is not available on all Bonnier News sites and not on all devices and platforms. It also works slightly different across our sites but the principle is the same. The purpose of continuous scroll is to maximise distribution and page views. These campaigns therefore generate many page views but the average engagement level is lower*.
* The average active read time is lower for continuous scroll campaigns because the share of readers who simply scroll past the native article is pretty high (up to 50% is not uncommon). However, when we look at the average reading times for those who have stayed for at least five seconds, the engagement levels are normally very good.
The index is our composite score for each article and is based on reach (35%), engagement (60%) and action (5%). If the article gets a score higher than 100 it is better than the average article.
There are significant differences from news site to news site, from one industry sector to another and depending on the type of distribution. When calculating the index, therefore, we always consider these factors.
We can also use our index to track how we improve over time, that we actually learn and create better content for our own channels. A skill that we use to help our customers. In the figure below we have sorted all native articles per quarter (all sites included). If you want, you can read more about this type of chart here.
All articles are sorted after index, from 0 to approx. 300.
You can write native articles in a million different ways. With valuable input from our editorial department we have arrived at a few features that we can use to write native articles as well as evaluating the result from the individual article.
The cookbook was created for our current and future customers in that they can learn from the collective experience regarding native content at Bonnier News. How do you write about insurance products for Expressen or travel for Dagens Nyheter? By turning to our cookbook you will find an answer. We do not only know what works best but also what does not work at all. The cookbook can therefore even be seen as risk minimising when venturing inte to exciting world of native advertising for the first time.
Product - you write about the product, e.g. a car
Context - you write about going on a car holiday with the family
Informative - you use facts
Emotional - you mainly rely on emotions to get your point across
Yes or no
Stock photo - the image is from an image database such as Shutterstock
Advertiser - using the advertiser’s own photos
Professional - using a photographer for the specific project
None - video
Advertiser - you as an advertiser explain using your own words
Case - another personer or company tells the story on your behalf
External expert - the text uses an expert or is based around a report or a study
None - the text is written like a newspaper article, “objectively”
There are several ways to make the article more interesting such as lists (listicle), fact boxes, quiz, celebrities and so on
Let’s use the car industry as an example. Here, we recommend that you write about the car, preferrably in a way that does not make it look too much like an ad. Furthermore, an informative angle if preferrable if you write for DN och Di. If you write for Expressen we recommend that you use emotional arguments. In any case you want to ignore agency photos, though this is hardly an issue in the automotive sector where glossy brochure photos are the norm. Across all site a neutral narrator works best, but feel free to use quotes from experts or even the advertiser (the head of design, say). DN readers enjoy fact boxes for cars whereas readers of Di and Expressen value test drives and reviews.
Using our cookbook when writing articles increases the likelihood of a successful article. But there are other very important factors such as the topic and good timing.
What works best varies from site to site and across different industries. What yields the best results for telecoms on Expressen does not necessarily hold true for cars on DN. These five areas can obviously be used individually, that video content works best for the health category, as an example. But can the choice of angle affect how well video content works?
We believe so and that’s why we prefer to look at the entire chain which gives us 64 different combinations. And the cookbook will give you the answer by industry and new site. Here’s an example of what the highest rated article on Di within Banking and finance consists of.
We mainly report on and evaluate native ad campaigns using various technical measurements that we combine in a dashboard for easy and fast access to all campaign details. But we also offer qualitative studies.
Measuring the effect of a native ad campaign is tricky for two reasons: the campaign itself is often limited in size and the campaign’s task is inherently slow-moving and possible even difficult to put your finger on, such as increasing knowledge or providing inspiration). At the same time as there is often some sort of call to action.
Having said that, we believe we have arrived at a measurement that works well. We also use a panel consisting of thousands of readers of DN.se, di.se, expressen.se, sydsvenskan.se and HD.se. The survey questions fall into two categories: the advertiser’s brand in relation to relevant competitors and the effect of the native content.
The details for each advertiser vary according to the campaigns main targets och how niched the sector or product advertised is. We often consider brand awareness, brand attitudes and brand preference as well as a few individual aspects. When evaluating the results comparing the answers of those who remember the campaign and those who do not we arrive at interesting and meaningful differences and patterns that indicate how the campaign was able to lift the brand, inspire purchase or other effect even though it may be difficult to prove causation. The number of interviews vary but we often have n in the region of 175-200.
A native campaign usually lasts for at least a month and contains several articles and/or video clips. In other words, there are plenty of numbers to stay on top of so rather than updating and e-mailing Excel files we conveniently collect all data in a dashboard.
All larger campaigns with Bonnier News receive a separate live dashboard where both the advertiser and the agency can follow the campaign every step of the way. The dasboard is indivudal and customisable but often contains the following:
Reach (page view totals, timeline, per week)
Page views per article
Quick exit and dwell time per article
Benchmarks for above metrics
Other: such as reach and engagement on Facebook if applicable
Normally we can trace clicks out from the native articles and this is even part of our index (see separate chapter). But clicks out are just that, clicks outbound from our sites. Det interesting bit is, of course, the clicks to the advertiser’s site and the continuing actions of the vistitor, such as further reading, subscribing to a newsletter, purchase etc). Our recommendation is to trace your exit links using utm tags and make you own analysis using your own web traffic tool, such as Google Analytics (GA).
Campaign data from your own GA can also be added to your dashboard. Contact us for more information.
Bonnier News Brand Studio
Contact: johan.wilberg@bonniernews.se