How A.I. is changing the face of digital marketing

Artificial intelligence is an umbrella term for types of technology that enable machines to mimic human intelligence, for example the ability to understand and respond to the environment, problem solve and understand human speech.

It’s an exciting technology trend for digital marketing and advertising and with digital ad spending overtaking TV ad spending since the end of 2016, advertisers and marketers should be innovating and investing in digital advertising.

These advances have led to the development of our award-winning AI powered segmentation tool, the ATOM, which serves creative to the most relevant target audience, therefore fully optimising ad campaigns.

For many though, despite widespread media coverage, the specifics of AI are often lost, misunderstood, or even unreported.

Here, we attempt to clarify what AI is and present some digital marketing trends we should look out for as AI increasingly infiltrates our lives and approaches to business.

Bots

Bots are programmes that operate autonomously. One popular type of bot is a chatbot, which is able to use human language to communicate with humans. Bots are already used by many brands; in fact Microsoft CEO Satya Nadella claimed that ‘bots are the new apps’.

One way in which bots are being used currently is to predict the success and popularity of social media posts. The New York Times used Blossom, an intelligent bot within the messaging app Slack, to predict how articles or blog posts would perform on social media.

Facebook posts recommended by Blossom on average have been getting 120% more clicks than non Blossom-powered posts. In the future, more brands may take a technology driven approach to select and promote social posts and even adverts.

Ad personalisation

Ad personalisation is a huge aspect of digital advertising, with channels allowing for audience segmentation based on interests, demographics and online behaviour.

However, ad personalisation requires collecting, analysing and interpreting huge amounts of audience data, as well as having the understanding to apply it appropriately.

Brands and marketers can automate such processes by employing AI technologies to carry out the collection and analysis of audience data – it is likely that as computers get smarter, they will be able to interpret and understand such data too.

With these machines in place, there should be more time for creative brainstorming and high-level planning (until they can do that too!).

Our new specialist tool the ATOM, gets the best creative in front of the most responsive audience to take the guesswork out of targeting. It allows us to collect huge amounts of data on audience interests, demographics and purchase intents.

Using the ATOM, we can work out clever rules to apply the data to make the right advert content and creative to target the exact right group of people. The ATOM is just one example of using big data and applying it intelligently in social advertising.

Image recognition

One of the most exciting and possibly most important developments in AI is image recognition. If we want machines to be able to ‘think’ like us and mimic our ability to respond to our environment, such as in the case of self-driving cars, it is of vital importance that they are also able to ‘see’ like us.

Computers are now able to identify and recognise simple objects and scenarios. Although these abilities are negligible compared to human vision and perception, these building blocks of computer vision have enabled some important technological developments, and continue to do so.

For example, Pinterest use object recognition to identify which products appear in pins that are liked, pinned or repinned by users. They then use this technology to boost pins and recommend relevant pins and ads to individual users.

These examples are just a tiny glimpse into the potential AI has to change our business and marketing approaches, as well as our everyday lives.

Find out how our AI powered segmentation technology, the ATOM, can generate your perfect audience based on consumer intent through AI segmentation analysis so you can get the right messages to the right customers at the right time.

What is the best way to measure paid media success?

In the ever changing, multi-channel digital landscape, measurement of campaigns is a challenge that is increasing for brands and making the question of where to invest marketing budget more and more difficult.

This difficulty is only amplified in awareness and brand campaigns with brands facing the challenge of not being able to link visible and actionable reporting all the way through to the business impact.

The key to measuring the impact of any campaign is to have a clear understanding of what you are trying to achieve. From this, you can then explore the technology and products you can use to measure this. Accurate measurement will provide insight into the impact your paid media is having and allow you to track the effectiveness of your campaigns.

Outside of the typical attribution modeling, brands are increasingly asking about the profiles and behaviours of those who are engaging with their campaigns to ensure their spend is having an impact in the desired market segment. This need for greater audience insights and a tangible and intelligent link between brand and direct response  was one of the key drivers behind the development of our audience builder and insights technology, The ATOM.

The ATOM

Using the ATOM, brands and media buyers are able to pinpoint the exact audience profile that is engaging with each content piece and capture these to strengthen the link into direct response campaigns.

The capture of these intelligent audiences allows brands to draw users into their offering not just based on an interest in their brand, but also what particular value or USP they are interested in at first contact. Not only does this enable a personalised approach and add more value to the brand budget, but advertisers are able to gain a detailed audience profile and measure this audience of responders against the audience they set out to reach to better understand the impact of their branding campaign.

The above shows a snapshot overview of the responders to 2 key creative messages from a branding activation.

From understanding your audience at brand level, we will then need to look at how we draw them in and understand the impact of each channel in this process. Let’s walk through the basics of attributing paid media success.

The problem with the ‘last-click’ model of analysis

One of the main challenges in measuring success comes when you’re running sophisticated multi-channel campaigns.

How do you attribute business value to each of those channels and determine which is providing the most value?

The ‘last click’ model of analysis is commonly used to analyse the success of paid media. This looks at converted leads and establishes where the ‘last click’ from a sale came from.

For example, if a customer first clicked on a paid social advert which did not result in a conversion, then went directly to the site, which similarly failed to convert, before finally clicking on a display advert which did result in a conversion, under a ‘last click’ method of analysis the display advert would be 100% attributed to the sale.

But that’s neglecting the influence that the initial paid social ad did have.

The last click approach makes analysis relatively easy, but also gives inaccurate results. This is not helpful when you’re trying to decide where to focus your paid media strategy. The primary reason the ‘last click’ model is inaccurate is because, as we know, consumers are influenced by a whole host of marketing channels.

Large companies can run up to forty different channels simultaneously, which means the majority of converted leads will have been exposed to multiple channels along the way to their purchase (which is what we call the conversion path).

How to attribute converted leads differently

One way to do this is by building conversion paths with Multichannel Grouping using Google Analytics and assign rules to establish your attribution model.

So, if a consumer only had one interaction before conversion, then the last click could rightly be given 100% attribution.

However, if a consumer had multiple interactions and followed a ‘conversion path’, then the first click may receive 50% attribution, assuming the conversion occurs in the next 28 days, whilst further interactions with other marketing channels would receive a lesser attribution percentage.

This way, a more accurate picture of attribution can be painted, allowing for better informed marketing and budgeting decisions.

It might be helpful to think of this in terms of winning a race in a high performance car. There are so many factors that can be attributed the success of coming in first: position, the car mechanics, driver skill, the weather and the latest tyre technology. How much of the success is attributed to each of the different factors at play will vary depending on the circumstances.

Although setting rules is undoubtedly an improvement on the ‘last click’ model, the model is only as good as the rules you set.

Formulating an accurate attribution model

The key to a better attribution model is using a data-driven approach.

By feeding the data we collect (i.e which marketing channels our customer was exposed to) into algorithms, we can actually analyse each conversion path, identify the most effective channels, as well as the relationships between different channels too.

After clicking on a social media ad first, what is the most common method of conversion?

By allowing an algorithm to crunch the numbers, these types of questions can be answered, giving you valuable insights into the inter-connectivity and efficiency of your paid media campaign.

For each individual conversion path an algorithm can attribute accurate values to each marketing channel based upon the wealth of data it has been fed from previous conversion paths.

Clever stuff right?

Working with attribution models means you can identify paid media success with a far greater degree of accuracy than the conventional ‘last click’ model.

We can see how our marketing channels are working for us with these attribution values, and make decisions on budget and resources accordingly.

The result of this? A more effective, and efficient, marketing strategy.

Smart audience insights

Our unique ATOM audience insight technology builds the smartest audience insights, delivering incredibly fast results and improved cost-efficiency.

Get in touch to find out how ATOM can help supercharge your paid media campaigns.