Tuesday, 16 August 2016

Adblocking is asymmetric warfare and publishers can't win

For those of us who like to keep an eye on developments in web advertising blocking (adblocking), there was a flurry of excitement last week as Facebook announced that it had designed and released a way to defeat the blockers. Adverts inserted into Facebook's news feeds have been made harder for an adblocker to tell apart from regular posts by your friends. If the blocker can't spot the ad, it can't remove it.

This has been coming. Various companies have been trialling experiments ranging from asking people nicely not to block ads (The Guardian) to attempting to completely prevent access to a site for those running adblockers (CityAM).

The response to Facebook from Adblock Plus has also been coming. Its Open Source community defeated Facebook's new tech in less than 48 hours.

On Thursday, Facebook announced another change.

And Adblock Plus killed that one the following day.

I've been plugging away on Twitter for a while with the opinion that big publishers can't win an arms race with adblockers. In this post, I'd like to explain why.

Adblocking is asymmetric warfare. On one side, you have large companies - publishers and ad-tech suppliers - who deliver adverts. On the other you have a growing population of people, estimated at 12m (pdf) in the UK, who want to block ads and a series of smaller software projects which allow them to do that.

Effective asymmetric or guerrilla warfare is all about harassment, persistence and the ability to adapt quickly against a large and powerful, but much slower opponent. Guerrilla combatants don't win a definitive victory, but eventually their opponent is forced to give up and make a deal, because they're burning through resources and not achieving very much.

In order to deploy its new solution, Facebook had to develop the code, test it and then deploy it onto a system serving 1.2 billion users, without breaking anything. Adblock Plus took less than 48 hours to nullify that solution - twice - and did so for free.

The blockers haven't even really had to get sophisticated yet to remain effective. All most adblockers do, is check against a list of server addresses that deliver adverts and web page elements which contain them and remove those before the page is shown to a user.

For a newspaper or Facebook to show an ad, a bit of code is inserted in the web page that effectively says:

[start advert]
Call ad-server and request an ad!
[end advert]

Facebook made that code more difficult to spot, but apparently not difficult enough.

Let's say Facebook got really clever with hiding their ads and mixing them up so the code for their containers and servers looked exactly the same as your friends' posts. What could blockers do in response?

1. Compare the posts in your news feed to your Facebook friends list and zap anything that doesn't match. It requires more coding than just killing advertising flags on a web page, but it's not really difficult... There are already plugins that have this functionality.

2. Use the legal requirement that adverts are clearly labelled as being adverts, as a way for blockers to find and remove them. You've got to write "ad" in the corner of an advert to obey the law. This stuff really isn't that hard.

3. Start training machine learning models to use the content of adverts to identify and remove them. A Gmail spam filter for the whole internet. Are you sure you want to start this fight?

None of those even touch on the idea of weaponised adblock, which it's not unusual to see suggested on tech forums. Currently, adblockers are passive; they prevent an advert from loading. But they don't have to be. Adblockers could deliberately load and click on ads in the background, thousands and thousands of times - without showing them to the user - to screw up the web's revenue model. I'm absolutely not saying they should, but they could, and the idea isn't new.

Are you really sure you want to start this fight?

This is why the advertising industry must acknowledge - whatever its position on the morality of adblocking - that it can't win an arms race with the blockers. Adblocking counter-measures are slow and expensive to roll out and industry wide, will need the cooperation of a huge number of participants.

Blockers can rely on a passionate community of skilled programmers who iterate and deploy new solutions quickly. The blockers don't have to win every time or force a dramatic surrender, they just have to make life difficult and expensive for the ad industry until it eventually gives up.

Advertisers, agencies and publishers must acknowledge three key facts:

1. Adblock penetration is growing rapidly and this is a problem

2. Publishers cannot win an arms race with adblockers

3. A large and growing section section of the population is on the side of the adblockers and so is the law.

It is imperative that the advertising industry identifies the reasons for the rapid growth of adblocking, in order to work towards an alternative to the current situation, where over 20% of people in the UK consume all non-paywalled internet content entirely for free.

This alternative is likely to encompass a model where adverts are no longer a vector for malware, are significantly less irritating to users and provide for much better privacy.

(no, this chart doesn't follow my data vis best practice)

As I've written previously, our new world may well be a much nicer place for large publishers delivering quality journalism. Their audiences won't be trackable onto thousands of tiny websites, so large publishers will be in a much better position to charge a premium for advertising space.

For better or worse, this new world is coming. If you target younger, tech savvy audiences, it is already here. It can't be fought so we must work out how to be effective within it.

Wednesday, 22 June 2016

Enthralled by tech we don't understand

The advertising world has discovered this week that an app purporting to crowd-source spotting migrant boats in trouble in the Mediterranean did no such thing.

It rendered the same static image for every user and when they 'spotted' the boat in it, asked for their contact details.

On the plus side, it did update with the weather forecast for Libya.

The app just won a Bronze award at Cannes.

Apple has pulled it from their app store and an investigation has started.

Whatever the morality of producing the app in the first place - which is pretty shocking - what the episode brings home to me, is the need for senior managers making decisions about tech, to actually understand the technology. Not to be programmers themselves necessarily, but to have a good idea of what is possible, what is easy and what is difficult.

A bunch of judges at Cannes decided that the migrant spotting app was worthy of an award.

None of them apparently had the nous to say "hang on, where are they getting their live satellite images?"

Of course the app can't exist. Marketing creative agencies don't have access to live satellite video streams of the Mediterranean. This isn't 24 and Jack Bauer's not an agency staffer.

Google Earth isn't a live stream.

You can get live images, if by 'live' you mean one per day. And it's not cloudy.

At the very least when it comes to technology, if we don't know, then we need to find an expert and ask. There are charlatans out there in the world. Some of them are software vendors, some do digital advertising and some make apps. If we're not to be taken in, the level of tech savvy in our industry needs to increase and quickly.

Thursday, 14 January 2016

Six tricks to make your data visualisations look better

I saw a tweet yesterday from @colinttrainor and couldn't agree more with this idea.

(tweet shown split into two for clarity)

Whatever software you're using, there's simply no excuse for accepting the defaults and not trying to make your charts look more professional.

But where do you start? It's easy to look at great design and agree that it's great, but when you're looking at an Excel default, inspiration is much harder to come by. What do you change first?

These are a few tips, which I've picked up as I tried to make my own output look better and hopefully a few of you might find them useful too.

1. Make yourself a colour palette.

Your company very likely has a corporate colour palette that you're expected to use for PowerPoint, and whether it's good or bad, it doesn't half help to simplify your choices. Rather than selecting from the entire spectrum of colours, you'll have six or eight to work with that (hopefully) are designed to complement each other.

You can make your own colour palette really easily and use it to give your charts, your blog and anything else you build a consistent, professional feel.

Go to Adobe Color and have a play. Find some visualisation examples that you like, which will help you to decide on the feel of your palette. Do you like striking contrasts, or more subdued tones? There isn't a right answer, but what you choose will have a dramatic impact on the impression that your charts create.

2. Simplify, simplify, simplify.

If I had to reduce my Effective Data Visualisation for Marketers presentation down to a single tip, it would be this (paraphrasing Tufte): If it isn't essential to your chart, get rid of it.

Turn off gridlines, reduce colours, turn off legends and axes if you can,  Look at Colin's tweeted example above. Most of the difference between the two, comes from turning off gridlines and axes, leaving only data.

3. Black text looks amateur.

Have a look at virtually any professionally produced visualisation, infographic, or presentation document. Is the text black? Look closely, is the text jet black, rgb(0,0,0)? I bet it isn't.

On a white background, changing all of your text to a very dark charcoal shade of grey, just works. I'm not a designer so I don't know why it works, but it does. Just do it.

4. Are you sure you want a white background?

The answer might well be yes, but this one is worth thinking about. Clean, white space is a very good thing but a shaded background can lift your visualisations and give them extra pop.

You probably don't always want black, but it's worth keeping in mind to make bright colours jump off the page.

If not black or white, then light greys can look good. When you look closely, it's surprising how many visualisations that you might initially assume have a white background, actually don't.

Here's one of my football match single-page dashboards. The background is a very light, warm grey. On the white background of this blog post, it's easy to see, but against Twitter's dark image borders (which is where I use these), it's not nearly so obvious.

Bright colours definitely aren't for backgrounds and one other thing you absolutely must do if you change the background, is to change everything that was white background, into your new colour. Miss a spot and it will look awful.

5. Times New Roman? Yuk.

Like colour palettes, this is something that most companies do, but few individuals take the time to set up. It makes a difference.

Take the time to pick a font. Here are a couple of resources to help.

By the way, there's a reason that the example fonts above are a picture and I didn't just change the font in this blog post to type them. I want to show you exactly the fonts I chose and not every web browser will render all fonts correctly. If your computer doesn't have a particular font installed and a website tries to use it, then you'll get something else.

Have a read about web safe fonts.

Web safe fonts aren't important if you're drawing charts in Excel and publishing pictures of them, but they are if you're building Tableau dashboards to publish online. If a font isn't available, Tableau falls back onto Times New Roman and makes your dashboards look horrible. Some of my dashboards do this and I really need to fix them.

In Tableau, it's a good idea to use a very common font that's close enough to what you really wanted, than to try to use something beautiful, which nobody else will be able to see.

6. Different is good

There's one more overarching theme which I'd like to mention and it comes from the world of advertising: Different is good.

Different is eye catching, and it will set your work apart.

Even if Excel, Tableau, R and other programs' default charts were beautiful (which they aren't), there'd still be huge value in changing them. If your work looks like everybody else's work, then it's not eye catching, will blend in with everybody else's work and not get noticed.

Find your own visualisation tone of voice and your charts will stand out much more. It's not always easy to break away from the defaults and try to make something better, but it really is worthwhile.

Finally, remember...

Good artists copy, but great artists steal.

Picasso (maybe)

Google image search is your friend. Good luck!