Donald Trumpista tuli Yhdysvaltojen presidentti ja hänen on siitä kiittäminen sosiaalisen median markkinointia. Ok, tämä on liioittelua, mutta ei kuitenkaan kovinkaan runsasta! Vaalikampanjassaan Trump ohitti valtavirran mediat käyttäen Twitteriä ja Facebookia tavoittaakseen ja tehdäkseen suuren vaikutuksen äänestävään yleisöön. Tämä kampanja oli erittäin hyvä osoitus sosiaalisen median markkinoinnin arvosta. Sosiaalinen media on selvästi jättämässä traditionaalisen median varjoonsa vaikuttavuusvoimallaan. Sosiaalisen median markkinointi toimi Trumpille, ja oikein tehtynä se toimii myös sinunkin yrityksellesi.
Donald Trump became the President of the United States of America and he did it thanks to the power of social media marketing. Okay, so this is an overstatement, but not by much! During his election campaign, Donald Trump bypassed the mainstream media and used Twitter and Facebook to reach out and make a huge impact on the voting public. It was the best, most fantastic display of the value of social media marketing that anyone has ever seen, believe me, its true! Clearly, social media is eclipsing traditional media in its impacting power. Social media marketing has worked for him and if done properly, it will work for your business too.
Use Return On Marketing Investment as Your Compass
How do I know if I am doing social media marketing properly? The answer is to carefully measure, interpret and then maximize your Return On Marketing Investment (ROMI). Measuring the impact of his social media marketing campaign was relatively easy for president Trump – he won the election! For the average business, this is a little more complex. First of all, to make good measurements, one needs good data. In the past, when any data, let alone good data was difficult to come by, this was one of the major challenges in measuring ROMI.
We are now well and truly in the age of “Big Data” so the challenge is one of picking out the valuable data from the huge stream of data that flows in at a rapid pace. Another significant challenge is more fundamental to the nature of marketing. In general, ROMI is due to short-term effects and long-term effects. The short-term effect is the immediate (i.e. within days, weeks and months) effect that marketing actions have on the sales of a product for example. The long-term effect is, as the name suggests, the impact that happens over the time scale of years or even decades. For example, consider a famous car manufacturer such as Mercedes. The short-term effect of a particular Facebook ad campaign for example, would be, the immediate increase or decrease in sales caused by that campaign.
The long-term effect would be for example, the impact on the Mercedes brand image-which is much harder to measure. When talking about social media marketing measurement, it is also a mistake to treat it’s impact by separating online and offline. Your customers do not live in a purely online or purely offline world, sure they may see your ad on Facebook, but they may also see your and your competitor’s ads on TV or on a billboard. So any sensible ROMI measurement must account for interactions between online and offline as well.
In terms of handling the aforementioned challenge of picking out the “good data”, the past practice of a having human experts deciding on what was good data and what was bad data, is no longer feasible. The volume of data and their complexity is simply too vast. The answer fortunately is within reach. The exponential growth of Data Science/Machine Learning/Artificial Intelligence over the past few years has produced an array of exciting tools which are at our disposal. Related to this, Annalect Finland has recently debuted a machine learning model based web-analytics pilot program (article in Finnish) with Tallink Silja, that offers unprecedented predictive accuracy.
Marketing Mix Modeling Tells You the Building Blocks of Your Total Sales
Another popular method of extracting ROMI is to do Marketing Mix Modeling (MMM). MMM is an econometric modelling method which has been around for a while now. Broadly speaking, it is a way of modelling a KPI of interest as the sum of a base component and marketing-action based components. MMM was originally applied to traditional marketing mix data and has been successfully refined and expanded to include digital media marketing data as well (e.g. social media marketing, search engine marketing etc..). A prime example of this is how we at Annalect, use MMM (which we also refer to as “sales modelling”) to help McDonald’s Finland optimize their marketing effectiveness.
Another example is the case of Specsavers Finland. With Specsavers, we used sales modelling to clarify how their online marketing impact was affected by traditional media marketing. MMM continues to develop with techniques such as Vector Auto Regression (VAR) models, Bayesian networks being developed to make the models ever more accurate and able to incorporate the previously mentioned “long-term effects”.
In this regard, Facebook in particular has taken the lead and announced an MMM partnership program which offers exciting new possibilities. When partnered with MMM providers, Facebook now offers more flexibility in creating customized ROMI measurement solutions including geo-location based measurement options, the ability to do randomized statistical testing (e.g. create randomized experiments involving a treatment group to whom your ad is shown and a control group to whom your ad. is not shown) and allowing for data from 3rd studies to be included.
Attribution Modeling Helps to Recognize the Most Crucial Customer Touch Points
Another methodology of ROMI measurement is attribution modelling. Attribution modelling is a methodology which divides up the credit for a conversion, to the various touch points along the customer’s path to conversion. By knowing which touch points are most (or least effective), one optimally allocate one’s marketing budget. Attribution modelling is particularly desirable for digital media due to the easy accessibility of customer data (e.g. – organic search, paid search, direct etc..) via google analytics for example.
However, this field is still in it’s infancy, and bad attribution models such as “last click” and “first click” attribution are still being used. The last click model assigns all the credit for a conversion to the last touchpoint prior to the conversion whereas the first click model assigns all the credit to the first touch point. To use a hockey analogy, this is akin to saying that the person who scores a goal is the only person who deserves credit for the entire play. Clearly, even the greatest goal scorer needs a great team behind him. And each team will have its own particular characteristics.
The take home message is that attribution modelling is a team sport and each business must have its own attribution model because one size does not fit all. The correct way forward is to use algorithmic attribution models that learn from data to do the attribution. Looking even further, the holy grail of attribution modelling will be to integrate offline data such as point-of-sales data, with online data and to also combine MMM with attribution. All of this will allow for ever more potent social media marketing for businesses which utilize these opportunities.