Ad Blocking Experiments

Summary

In an interesting coming-together of serious and whimsical, FT Labs worked on some Ad Blocking ideas to help the Ads team address this thorny issue on FT.com. The prototypes we built helped trigger some excellent internal debate, teasing out many of the issues and subtleties. One of the prototypes (‘missing words’) formed part of a multivariate test on Registered (non subs-paying) users. The media response to the test was generally positive. Sadly, the outcome of the test leads us to think that a simple block on displaying article content, when a user is Ad Blocking, leads to the better outcome. But the debates were good…

Details

Ads and Ad Blocking

For decades, advertising has been a major source of revenue for news organisations and publishers around the world. Recently, revenue from printed advertising has declined sharply and, for a variety of reasons, advertising from online sources hasn’t been able to make up the difference. To add to this, users are now able to easily block ads from their web experience. We understand people don’t want to be followed and hounded across the web, but we’re not like that. Ultimately, good quality journalism, such as the FT offers, needs revenue to allow us to continue the work we do and, since the FT operates a mixed revenue model, some of that revenue comes from advertising to our users.

What can we do about Ad Blocking?

Ultimately, a suite of ad-block solutions was built and deployed to a cohort of FT.com users (registered, but not paying subs), based on the much larger set of candidate solutions dreamed up in FT Labs and the wider business.

If we detected ad-blocking on our site, we would do one of the following

  1. Remove a number of words from the article that roughly represented a percentage value of the amount of revenue the FT loses to ad-blockers (with an explanation of why…).
  2. Explain about the problems caused by Ad blocking and request that the users allow the site, but leave the article viewable
  3. Simply block access to the article whilst ad blocking was in place
  4. Do nothing

It was felt by Labs and the Ads team that this idea was a good middle ground between punishing our users, and letting them continue with ad-blocking unchecked, but some of the other ideas we had along the way were a pleasure to explore, safe in the knowledge that they would never see the light of day.

Ad Blocking Experiments

The Labs department was asked to partner with the Ads team and find creative, effective solutions to our ad-blocking problem. We aren’t the first publisher to have a look at solving this problem. In recent times, publishers have created myriad solutions that aim to coerce or convince users into adding their websites into the allowlist of their ad-blockers, or disabling them altogether. They have been humourous, blunt, subtle, and some have ended up drawing the ire of social media users. Here are few instances we found interesting, enjoyed, and/or had a measurable effect on the number of users ad-blocking.

So what could we do that was different? Initially, the Labs and Ads teams sat together and had some brainstorming sessions in which we came up with a handful of ideas, and afterwards we elicited ideas from other members of staff.

We prototyped these ideas on the then beta version of the FT.com and deployed them behind a flag, so that only FT staff members could see the demos. Once the our demos had been enabled, a UI would appear at the top of each article page that would allow the user to flick through and view all of the ideas we had had.

Screen Shot 2017-02-22 at 15.36.22

Broadly, our ideas fell into one or more of these categories:

  1. Punish our users

    2. No ads, no allowlist, no service.

  2. Amuse our users…

    3. So they look more charitably upon us.

  3. Utilise our users.

    4. If we can’t get revenue from them, can we make them work for us?

  4. Explanation

    5. Level with our users and let them know why ads are important

  5. Passive aggressive

    6. We’re not mad that you’re blocking ads, we’re just disappointed

We did not worry too much, at this stage, about the risk of offending users, since we were mainly interested in provoking internal discussions around Ad Blocking, and exploring the possibilities.

It should be noted that the following were considered as experiments only, and most were never intended for public consumption.

  1. No Vowels

    2. This was our first idea. If we detected a user was blocking ads, we would punish them by removing all of the vowels in the article, making it difficult to read. We thought we were being really creative with this, but it turns out to be a common idea over the years that’s come to be known as ‘Disemvoweling’

  2. Bart

    3. When Bart Simpson misbehaves, he’s punished by being made to write out the things he should not do over and over again on a blackboard. Much in the same guise, when an ad-blocker tries to view an article, we overwrite said article with the thing they should not do – I will not block ads.

  3. Revenue

    4. Remove an amount of words from each article roughly representative of the revenue lost from blocked ads.

  4. Questions

    5. We would ask users what reasons they had for blocking ads, and responding to answers with counterarguments.

  5. Ad Articles

    6. The most passive-aggressive proposal in this list. Between every second and third paragraph in an article that had its ads blocked, a modal was inserted that linked to an FT article about ad-blocking.

  6. Delay Page

    7. The longer ads were blocked on our site, the longer we would make the user wait to view each article. Our reasoning was that eventually, user frustration with the delay would outweigh any perceived benefits of blocking the ads.

  7. Ad Guidelines

    8. A simple modal dialog that tried to educate our users about why our adverts are better/more responsibly sourced than adverts from other publishers

  8. Word Shuffle

    9. Much like No Ads, No Vowels, we would adjust the content of the article, but instead of removing characters we would rearrange them into an almost intelligible gibberish.

  9. Archive OCR

    10. If we can’t get users to view ads, maybe we can benefit from their labour instead. In order to read the article with an adblocker, a user would have to transcribe an image of an unidentified phrase fragment from our historical archives.

As part of the demos, a small questionnaire was included for our colleagues to give us feedback on how they felt users would respond to each idea, and any ideas of their own that they may have.

After giving FT staff two weeks to play with and mull over the options we’d presented to them, and having received feedback on all of those demos, the Ads team selected and deployed one of the ideas as part of an A/B test to measure the effect. Ultimately, we decided to go ahead and test the revenues idea on a portion of our ad-blocking users ad revenues idea. Now, if we detected ad-blocking on our site, we would remove a number of words from the article that roughly represented a percentage value of the amount of revenue the FT loses to ad-blockers.

The users we decided to test our approach on we classify as ‘registered’ users. These are users who have an FT.com account, but have never had payment details or a subscription added to that account. The only revenue we gain from these users is through advertising, so if they’re blocking ads, we need to address that.

We were also very keen to not experiment on our paying customers with a potentially highly contentious experience.

Results

The experiment ran for 30 days, and the media response was generally positive.
e.g.

Once the A/B test had run its course, the results were as follows:

When presented with our message and word removal, 46.59% of users opted to allow or disable their ad-blockers to view the FT.

Our page view actually increased by 9.41% for the period of time that our test was in place. Though not statistically significant, it certainly is interesting. We believe people clicking on extra pages to make sure the words had returned may be the cause of this result.

Subsequent visits to the site were unaffected, so asking people to disable their ad blockers hasn’t had a negative effect.

We’ve established via longer term testing that simply blocking access (for registered but not subscribed users), with an explanation of why, leads to a higher proportion of users actually allowing FT.com and reading more articles.

It is with some sadness that we will park Bart and his ilk in this blog.