Is the past a good indication of the future? It certainly is, believes Ben Latham, Client Services Director, at online marketing specialists Summit .
Developing an attractive product and proposition is only the start of retailing. Without a strong marketing strategy you cannot deliver a product successfully to market.
With a new product launching every three minutes around the globe, the world is now more competitive than ever. Whether you are selling through retailers or direct to consumers, understanding customer behaviour and predicting sales before they happen gives a competitive advantage that most businesses can only dream of.
“If you can’t understand the uncertainties in your marketplace, you will lose out to those who do”, says Professor David Wooff, Durham University.
Digital plays an increasingly important role in the overall channel mix for customer acquisition, especially as a ‘direct to consumer’ channel. However, the rising cost of digital means it’s a more competitive landscape than ever before. The key challenge is determining how to create cut-through and raise awareness whilst delivering against tight product margins to bring profitability back into the business.
The evolution of digital marketing has created new expectations in terms of measurable return; businesses can work smarter with the data they collect and embrace new optimisation techniques to deliver efficiency and review return on investment.
Typically, digital marketers can effectively measure today what happened yesterday and, through robust and rigorous analysis, understand what interventions drove changes in performance. In a sense we’re reliable historians. However, accurately predicting what will happen in the future, both tomorrow and longer term, largely remains the holy grail of marketing which is required to plan and optimise performance-based marketing campaigns effectively.
The Opportunity
More and more digital ‘fingerprints’ indicating customer behaviour are available to digital marketers in the form of data. By understanding the relationships in this data it is possible to predict how customers are likely to behave and take advantage of this insight when planning marketing and deciding where and when to invest budget. The ability to identify patterns and relationships in the available data allows for the reliable prediction of what’s likely to happen tomorrow, next week, next month and beyond.
From this knowledge, it’s possible to understand not only how many people will search for a product, but also how many people will purchase that product. Decisions can be made about where to advertise and how much to pay.
Assessing the real impact of additional digital marketing spend using a traditional ‘bottom up’ forecasting method, using click through rate and conversion, always proved wildly inaccurate in the long run and were of little commercial value to businesses which operate on slender product margins and depend on ‘certainty’ from their investment.
However, the market has moved on. Accurate budgeting and forecasting requires sophisticated predictive analytical models to provide recommendations of marketing performance, both at a channel level (PPC, PLA and display) and keyword level.
Accurate forecasting relies on being able to make predictions about performance with certainty and understand the relationships between important factors that affect the way customers buy. External factors that impact marketing such as seasonality, weather, promotions or TV advertising have been found to impact demand in a profound way. No surprise there, you may say. However, it’s more complex to understand the changing level of impact at a given time and dynamically adjust decision making accordingly.
Let’s have a look at some specific impacts around the relationship of product sales with both weather and TV advertising.
How does temperature affect sales? When looking at the UK retail market as a whole against temperature variations, it doesn’t show much noticeable change.
However, when looking at specific product types or categories, such as coats, shorts, electric blankets, lawn mowers, carpets, fans or paddling pools to name a few, there’s a strong correlation with weather, as you might expect. This correlation exists outside of any common seasonal trends, but is directly related to specific changes in weather conditions regardless of season. However the real value lies in understanding the temperature and weather thresholds that trigger a distinct change in behaviour. For paddling pools in the UK it’s 21°C that sends people out into the garden in Speedos, with a hosepipe and an inflatable swimming pool.
So, how can retailers capitalise on this knowledge to maximise product sales?
Let’s take electric blankets as an example. In order to understand exactly what happens in different weather conditions, the sales performance of electric blankets over a 4 year period against historical weather data was analysed and what we found was commercially invaluable.
A five degree change in temperature against the average can affect the conversion rate of electric blankets by 100%. Using this insight, we are now able to better anticipate product sales by analysing a ten day weather forecast each day. Real-time adjustments can now be made to budgets and the optimisation strategy based on automated weather forecasts ensuring we respond and capitalise on changes in customer buying behaviour.
How does TV drive online behaviour? With the UK watching 455 billion hours of TV a year, it still plays a significant role in advertising. Add to this the phenomena of ‘dual screening’, a behaviour that’s now become a normal activity in the home. Capitalising on this behaviour can also bring great rewards.
There is a clear uplift in online marketing activity during a product’s TV advertising campaign. In fact, paid search and TV correlates stronger than any other online marketing channel when reviewing online channel KPIs against television ratings.
This is probably no surprise, however, being able to understand the impact a campaign has on product sales by mapping online uplift in terms of traffic and sales against time of day, ad length and programme content enables you to understand whether you are efficiently buying TV advertising or not.
This in itself isn’t very actionable and doesn’t allow you to react real time to what is happening. However, when analysing the campaign by minute against the TV ad to understand length of impact for any given advert, an understanding of how you need to react in the digital space is gained; the greatest uplift can be seen up to 15 minutes after the TV ad has aired.
Forecaster, Summit’s performance marketing platform, is able to capitalise on this insight and customer behaviour by taking real-time inputs from all major television channels, looking for opportunities to launch and modify search marketing and display campaigns in real time synchronisation with television events such as adverts. Forecaster predicts the likely uplift from a TV advert slot and automatically responds to this event in seconds, uploading changes to Google in almost real time. New and modified search or display adverts are live before the advert has even finished. This gives businesses the chance to ‘hijack’ the uplift available from competitor TV advertising as well as brand advertising. In trials, Forecaster has delivered a 15-20% revenue uplift from keywords triggered against TV advertising. Of course this opens up a huge range of possibilities to synchronise a wider range of digital marketing channels with TV events.
Predictive analytics and understanding factors that influence consumer’s buying behaviour offer businesses a significant competitive advantage over current methods used by the majority of digital marketers and agencies.
Using these techniques have seen consistent uplift in revenue (up to 900%) with the average increase at over 35%; results have shown that the greater the daily spend, the higher the opportunity for greater uplift.
The future of marketing optimisation is to predict future performance with certainty, the answers exist in the data, understanding external factors will make your predictions even more accurate.
Let’s look to the future to understand the decisions we should make today.
