Finding the optimal target for your marketing campaigns requires a smart, data-driven approach to CRM that should not be influenced by clichés or gut-feel. Expertise of creating effective campaigns is key to attaining those coveted results that targeted marketing alludes to: successful campaign stats (open, click and conversion rates), a flawless customer experience, great brand perception and of course, high campaign revenue.
Let’s look at a few examples of messages that could be used in targeted campaigns:
- A sports material retailer who wishes to promote a new range of sunglasses (a product with a high margin but which doesn’t correspond with the core activity of the brand);
- A retailer specializing in audio-visual and household equipment who wishes to clear a stock of Kitchen Aid’s (but whose price is 10 times higher than that of the average product on the site);
- A distributor of cultural and electronic goods, who wishes to operate a trade-marketing promotion for a new mobile phone, with the approach of the festive season in mind.
Similarities between these scenarios resides in the fact that it is worthless to send targeted messages to your entire contact base. Niche sales, flash sales, seasonal trends – each campaign is only going to appeal to a specific fraction of your audience, hence the question is not “what to sell”, as there is a business opportunity present in all three cases. Instead the real question is: “to whom to sell?” Or more specifically: “how do I distinguish the fraction of my customers who will be interested in a specific campaign, from those who will not?”.
Traditionally, there are three primary approaches for setting up targeted marketing: socio-demographic targeting, RFM targeting, and affinity targeting, also known as retargeting. Whilst these methods form a base for targeted marketing, they comprise severe imperfections and fail to leverage the latest innovations, such as Artificial Intelligence, which can significantly increase the performance of marketing campaigns.
Tinyclues created their own predictive solution, aimed at bringing a new level of sophisticated targeting into the hands of the marketer. Based on four fundamental criteria, Tinyclues suggests that. if targeting is going to be pertinent and optimal, it needs to be:
- Subtle; via automatic learning of past purchases, on a product by product basis. Remove common sense or reliance on clichés and manual categorization. For instance, when comparing two mobile phone models, such as Wiko and iPhone, with prices ranging from 2 to 4 digits, it is obvious that there is never going to be only one unique segment for “phones”, even for “smartphones”. The same applies for music in general: it’s wrong to assume that you can send the same campaign to fans of jazz as to classical music.
- Exhaustive; via the use of a multitude of available criteria and data, even those which are unintelligible for humans. Why limit yourself to measuring just one aspect of your data, when we can take all of them into account? Socio-demographic data, purchasing records, navigation records, clicking records, etc. Each table and each column in that table is potentially exploitable and should be exploited.
- Objective; via the detection of tiny clues and indirect causalities; does the fact that a pseudonym is “firstname.surname@” and not “killer_1988@” mean that a customer will be more inclined to buy a lawn mower for $1000, or a particular genre of novel or a popular brand of sneakers? If you clicked on a newsletter for summer dresses last month, is there a higher or lower probability of you buying a pair of sneakers? And what about an Android digital tablet? All these actions may seem insignificant in isolation but accumulate to build a multi-faceted view of each customer.
- Automatically updated; via the continuous readjustment of targets according to the evolution of behaviour, trends and latest events. Artificial Intelligence is the lever which allows us to make sense of “big data”, and we are in an era where automatized learning gives powerful momentum to daily marketing actions, without resorting to fastidious manual work.
Based on the above criteria, Lookalike models can be built by metabolizing the entirety of your available customer declarative and behavioral data, detecting and consolidating low frequency patterns and implicit signals, connecting users and products in a magnitude of different ways, and ultimately obtaining the optimal target for any campaign (click here to learn more about how Tinyclues’ SaaS solution works)
Tinyclues artificial intelligence SaaS technology can help you identify, in just a few clicks, top-performing targetings and help you promote the diversity of your product catalogue to the customers who are more likely to buy from you. You can run multiple campaigns simultaneously to different targets in your database, without causing customer fatigue. Not only will it help you achieve greater ROI and incremental revenue, but your brand will benefit too.
Interested? Ask for a demo!