Using data to predict purchase intent
Marketers globally are facing a dilemma. Despite the continued growth of digital advertising, the increasing sophistication of programmatic technology and more customer insight than they have ever had, the effectiveness of their digital campaigns has stagnated.
Although the recent Advertising Expenditure Forecast from Zenith Global predicted that digital ad investment will grow 13% this year, taking it over US$205 billion (and beyond TV’s US$192 billion), click-through rates have remained consistently poor at around 0.15% over the past five years or more, according to Google-Doubleclick.
So how can marketers kick start ad response and conversion to deliver more value for brands from their digital ad spend?
The log jam lies in the current siloed approach to customer data, the lack of transparency, duplication of data, the latency between collection and activation and the methodology of who dictates what user falls into what segments is failing to give marketers the data required to engage efficiently and effectively with consumers at the point when they have an interest or intent to buy.
Fabrizio di Martino, director of digital marketing, Europe, for Intercontinental Hotels Group, raised the issue of duplication, saying that sometimes he thinks it is just the old Wild West.
The solution demands a shift in mindset and a change of tack.
Here are 7 ways that marketers can use data to predict purchase intent.
1. Climb out of the silo
Currently, each data supplier only see’s part of the consumer journey, making it difficult for marketers to understand where they are on the route from engagement to purchase. This leaves knowledge gaps that mean brands simply don’t get a true understanding of how consumers are making a buying decision. This makes targeting advertising at them, at the key point of intent, a hit and miss affair.
Marketers need to look at new ways of understanding data to complete the jigsaw picture of the consumer journey and where they are in the purchase funnel. This is the optimum time they should communicate as this is most likely to drive conversion.
2. Connect with the real world
The rise of programmatic advertising and the ability to deliver mass digital campaigns seems to have led to a ‘more is more’ approach by brands and a lack of sophistication in terms of targeting. Although it’s true that targeting has become something of an obsession, it views consumers as homogenised segments rather than real people. Segments — say new mums — are targeted en masse, with little effort made to understand the mindset that an individual might be in on being served an ad in terms of their intention to buy. Advertisers can serve ads to the right audience, but if the individuals targeted aren’t in the right frame of mind, they are simply wasting their time.
Marketers need to understand the data, interpret people’s behaviour, look at key values, signals they are giving off that indicate an intention or interest to buy. Place yourself in the shoes of the consumer, what key signals do you give off when you are on the path to purchase?
3. Define a clear goal
Before you can predict consumers’ intent to purchase, you need to set a clear and specific objective as this will help to narrow down the data you need and also create something by which you can measure the results to define success.
Setting a vague goal, such as “Retail purchases will increase” will not work as it is too broad. Look at your business plan for the next three to six months and pull out a key piece of insight that will help to drive success, such as “identify which customers will buy a barbecue within the next 21 days with 80% accuracy”. This gives a defined focus to your work.
4. Collect the right data
Now you have your goal, you need to specify the data that you’ll need to achieve it. To start with consider working backwards and collecting insight on purchase history, ensure you obtain data from all of your communication channels, search, social, programmatic. Investigate the path to purchase both positive and negative, what signals of intent did the consumer provide, is there a pattern or trend that you can find, what content did they read, what was the corpus of the pages they where reading, what was the trigger to start them on the path to purchase, how long did it take and does the pattern change per product and per country
The most important predictors are usually the most informative, especially when it comes to understanding the true meaning of the content and the key intent signals it gives off but not in an isolated case, there has to be a process of stitching these signals together .
There’s no clear rule for the amount of data you will need, and it’s worth discussing this with your data analytics and data modelling partner. It could be two years worth of data, down to a single month.
5. Model data intelligently
Once you’ve collected your data from multiple sources it’s time to bring it all together and work with your data analytics and data modelling partner to apply intelligent techniques to build a more complete picture of the consumer journey and be better equipped to gauge intent.
Intelligent data modelling moves brands closer to consumers and reverses the current targeting methods. Starting with the “right mindset, right intent signals”, then establishing the right place and right advert drives performance and provides a better understanding of what a consumer is going to buy, guiding which product and price point marketers should communicate.
It delivers a greater understanding of the consumer’s intent, so that advertising can be used not simply to drive a purchase, but also to guide consumers along their buying journey, providing the information they need when they need it to ultimately make an informed purchase.
Ii is best to start with a simple model, because these can be turned around quickly, and this is important when building predictive models because it’s an iterative process and the biggest gains can be made quickly. Complex models are time consuming to improve and the results more difficult to interpret. Before applying to live data, it’s important to test the model on a sample dataset to see how it performs.
6. Delineate active buyers from prospects
Advertisers should develop a different strategy for targeting “the interested buyer” that are looking for a product or service compared to ‘Intent audiences that are ready to purchase. As an example a mobile provider will know a lot about their customer from name, gender, what type of phone, when their phone is up for renewal to how much they spend on data each month. However, how do they target new consumers with the right message and strengthen loyalty to their brand?
Understanding what the customer’s current, real-time focus is on is the key, What if they knew that in the last hour the customer has shown multiple “intent” signals from their digital behaviour that indicate their intention to book a holiday right now and buy a phone, this would be an ideal opportunity to change the marketing message in real-time to, say, “No data roaming charges when abroad this summer”, thereby becoming powerfully relevant to the customer.
The result would be more efficient and effective advertising trading strategies that would drive up campaign performance levels. This also holds the key to closing the current dangerous gap between digital advertising growth and ad effectiveness.
7. Know when not to target people
What’s even more exciting is how the mobile provider can use the intelligence of understanding “intent signals” from data modelling efficiently and effectively when targeting their own customers. This includes knowing when NOT to target their own customers — not everyone is going on holiday at the same time, even if it is summer!
How many times do we see a product we have just bought or already own, asking us to buy again!!
Conclusion
Intelligent data modelling applied effectively to the right data set offers advertisers unparalleled access to their target consumers at the most important moment — when they are interested or intending to buy. It enables advertisers to present ads that are more relevant and interesting, therefore driving engagement, response and conversion, and also improves the consumer experience as they are not being bombarded with irrelevant and annoying marketing messages, significantly reducing the need for ad blocking.
Harvey Sarjant, Managing Director, Sirdata Limited
