Intelligent CIO Europe Issue 57 - Page 81

INDUSTRY EXPERT
Overall , if you can get more efficiency with less investment , and if customers get the experience they ’ re looking for , that ’ s a good thing . That ’ s why personalisation , when you get into these more advanced data use cases , can be so powerful .
How can prediction models help businesses make informed decisions to achieve their future goals ?
Predictive models can often get complicated , with data becoming hard to understand or maintain . For us , the first step is making data very easy to understand . Once it is made descriptive , as opposed to cryptic column names , you can then start to build machines that learn algorithms to predict things like churn or propensity to buy .
You can start to integrate your acquisition platforms ahead of that information coming along with records of past leads . It helps you be more efficient in knowing who you should engage with first and what messages engage people , but also gives you flexibility to start custom-building your own systems .
We ’ re now seeing the desire not only for simple predictions like churn , but also other events that you can engage customers with , and look at behaviours to predict and inform other decisions .
We built Snowplow with the mission of not only supporting third-party AI / ML platforms , but also providing increasingly sophisticated teams with the ability to use their own Machine Learning technologies .
Prediction is estimating the probability of an event happening based on all the features and behavioural data points you have created throughout a customer ’ s journey . You begin to delight customers by predicting their intention , as well as their outcome . If businesses can be exactly where the customers are while they ’ re asking a question , it makes the brand more relevant and the perception of their service increases .
How can organisations begin to utilise their data creation more effectively ?
Many businesses invest in a tool that isn ’ t quite right for them and work backwards , attempting to shoehorn their needs into it . Snowplow has designed a data creation platform to customise the entire approach and build a data structure that ’ s right for you and explains what your customers are doing .
Each business ’ customer journey is unique and you want to be able to bring it together in a way that ’ s descriptive of that journey . The first action should be to map that out and understand your model , customers and the needs for your platform .
The next consideration is where you are collecting that data from and what policies you want to enforce . You might be obtaining data from an IoT device , for example , so you ’ d need to think about your creation point and then apply policy by considering data sovereignty and GDPR . Regulations such as these can land businesses in real trouble if they get it wrong but , if they get them right , it ’ s a competitive advantage because companies are often too nervous to approach it .
Businesses should also remove things that may introduce bias – consider removing gender completely from your data , because there ’ s no gender for any of the models to use , room for gender bias can be significantly reduced .
Finally , think about the desired outcomes . For example , optimising your ad spend is one example which can save money and which you can drive via conversion . Once you ’ ve got that data , what is the activation you want to start with ? Once you ’ ve got the philosophy , you can continue to add more activations on top , ensuring consistently credible data .
What are some of the shortcomings of CDPs and what should an efficient and effective CDP offer ?
Handling customer data is always a very sensitive activity . Businesses should look for CDPs or a platform that isn ’ t just relating data from point A to point B but is creating data in a way that ’ s well understood , compliant and moves it into a data structure that represents the business .
Moving data around in a structure that doesn ’ t quite fit a business creates more work , more room for errors and more complexity .
Third-party data creation and data collection is becoming more regulated and more companies such as Apple are specifically removing the ability for third parties to capture this data . At Snowplow , we deliver all data first-party which means that with Snowplow you are the custodian of your data , it actually sits within your infrastructure , removing your dependency on third parties to provide these insights .
As custodian of our customers ’ data , you need to make sure that we do the right thing for your customers , ensuring that data is protected all while ensuring the data is customised for your businesses use cases . That ’ s a powerful feature for a CDP . p www . intelligentcio . com INTELLIGENTCIO EUROPE
81