Tuesday 14 November 2017

Jonathan Seal – Mando Agency

In most cases AI used is narrow or weak AI – choosing a narrow domain, inputting data and training it to respond to that.  This form of AI is being used in a number of services: traffic lights, checking bank balance, estimating time of arrival/delivery, Facebook auto tagging etc.  Marketing departments in companies use AI very little – Tailoring promotions 15%, improving media buying 16%. 

Jonathan talked about how; if the service is personalised people will go back to it even if it’s hard to use.  Services that he mentioned included Netflix, Disney’s magic band, and Google search prompts.  They all pick up data from the users habits and then tailor the screen, promotion, and search towards those habits.  E.g. Netflix gather data on what you watch, when, how long for, when you stop watching a series, and with google search, depending on your search habits it will tailor your search to see what is relevant to you (example given: grand theft auto for a police man will prompt crime, for a gamer would prompt the online game). 

AI has good opportunities for marketing professionals and they can use major companies to help them understand their clients more. 

There are threats involved too:

The examples Jonathan gave included the tweets from @TayandYou which Microsoft released via twitter last year; trying to train AI to be human, which caused a number of offensive tweets.  Another example was Google categories that identified two people as gorillas because those training the data were white men and the material provided was not enough, as well as gender bias in some languages.

Why AI fails:

AI Black Box – it is hard to explain why some decisions were made by AI, this cannot always be easily tested by programmers, which can cause problems when trying to be compliant with GDPR.

Benefits of AI:

AI can be used to simplify office systems for example room booking, done by AI, or beer fridge report – in Jonathans example they added personality into the AI so it could respond to questions like “How are you?”

Online experience is becoming more and more human centred, personalisation at scale is key and it’s increasing improvements within the use of marketing.  Getting started is easy as there are companies you can partner with to support in this.

AI isn’t perfect!

Camila George – Deloitte

Camila talked about Deloitte’s journey into marketing automation and gave some tips.

She explained that they defined a strategy to evolve their digital presence – a way to engage with customers. They started in June 2016; although the journey started a few years back, with the desire to better understand their clients and delivery the service better.  Their main focus was client value.

Camila explained that with there being so many messages out there for people to absorb they wanted to ensure they got the message out there at the right time to be seen by the right people – “cut through the noise and be different”, by building trust, relevance, respect and accessibility.

Camila explained that depending on what the end result is, it is best to be clear on the strategy, getting buy in and defining business requirements.  Then it’s important to complete platform assessments (there are so many platforms out there).  It’s also good to do pilot campaigns, this also gets buy in further.  Getting marketing automation does create internal change and internal change management is required.  Most importantly, it’s key to build the right team, they don’t need to be experienced in marketing automation, but they need to understand the value, strategy and the business well.

Deloitte use Marketo as their platform and at present are delivering all campaigns through this with seamless integration between Marketo and their CRM (not all platforms sync with CRM, and some companies may wish not to).  They have had to increase training with internal stakeholders to improve the client engagement and have designed super templates for programs and emails.  They have designed and implemented a preference centre, which is great for GDPR giving customers the chance to change preference.  With this preference centre they can also work out when people are changing their preference and find out information they might not have known in regards to their interests.

Camila states how important it is to ensuring there is sufficient testing and a dry run data migration flow is completed before go-live, it’s important to ensure certain data isn’t falling out and that the sync works as it should.  The data will impact AI.

The use of marketing automation has given Deloitte a more client centric view, the content is client led, using insightful data.  There’s more of a bridge between marketing and BD and the data is more compliant.  It’s a great tool to use to review the data for GDPR and can amend what is sent to different people depending on their engagement behaviour.  It also can link what prompted the user to request more information.

To get buy in use a really narrow use case and then build from that.  The first few months of data wouldn’t be much, maybe show engagement level but data later on (usually a year) can look at what’s working well, what the views are etc.

By Nehar Ullah, Marketing Executive at Acuigen