Algorithms can predict which products we are going to buy based on our Marketing DNA. These online buying signals can tell marketers which prospect is a great target and which one is a waste of time and money. Here is an example of one company that improved marketing performance 6X with Customer Intelligence.
Everybody needs toothpaste—but with other products, specifically B2B, it is not that clear-cut. Fortunately, with more complex purchases, it’s actually possible to find signals online that can predict which prospects are a good fit for a product. Moreover, if you have a portfolio of products, these same signals can tell you which of these products a prospect is likely to buy.
Matching the right prospect with the right product was a huge challenge to one of Mintigo’s clients, a leading developer of QA tools. The company was trying to segment the market for two products: The first was a quality assurance platform for developers and the second product provided website monitoring.
The company had a very successful lead generation program drawing a very large amount of inbound leads and had a large lead database. They were also very good at producing high quality content that clients find very valuable. However, the challenge with inbound leads is lack of data in order to match product interests to leads.
Customers today, whether B2B or B2C, have little patience for reading promotional materials or interacting with sales reps. Therefore, it is important to offer prospects exactly what they need the first time, rather than embark on a fishing expedition to find whether you can provide them with something they need.
Segmenting buyer personas
To boost the client’s marketing effectiveness, it was important to find a predictor that would be able to analyze prospects and say which product they are likely to buy—before the first interaction with the sales rep. Furthermore, the goal was to expand and find a lot more potential customers for each product based on their characteristics. In addition, armed with the right segmentation, the marketing team would be able to produce and send highly relevant content to cater to the needs of every buyer persona.
How do you know what people are going to buy (sometimes even before they do)? The secret is big-data predictive algorithms. Mintigo takes all available data from the Web and builds a predictive model—similar to college statistics’ regression analysis, but a lot more complex.
First, most of the data on the Web is unstructured, meaning that it comes in the form of tweets, blog posts, code on the website and profiles on social networks. To handle this data, Mintigo had to create the technology that could make sense out of this mountain of data.
Distinguish pattern from noise
The second part is to distinguish the pattern from the noise. The aim is to see which indicators influence propensity to buy and which ones are simply noise. Then focus on finding only those indicators that matter.
We all make simple exercises like this in our heads. For example, we can all guess that software companies are more likely to buy QA products. However, predictive algorithms can actually prove this link using math and probability, and in addition, find thousands of other indicators that are a lot less obvious and can hardly be detected using reasoning alone.
To distinguish between the persona that buys the QA product and the persona that buys the website monitoring product, Mintigo analyzed two sets of actual customers, one for each product. The goal was to identify similarities and create a “cluster” of attributes—that are common among product users and distinguish it from other product users.
There were several characteristics that distinguished QA from web monitoring prospects:
- Job Title Semantics: Prospects for web monitoring had titles related to Web and ecommerce such as merchandising, ecommerce, advertising, consumer and brand. QA prospects had titles related to development such as technical, software, product and quality assurance.
- Website semantics: The company website of web monitoring prospects had words like shop, online, price and holiday. QA prospects’ company websites, on the other hand, had higher density from keywords such as technology, support, solution and software.
- Technology: Use of technology is also strong buying signal. Website monitoring typically used online payment, SSL and pay-per-click advertising. Prospects who are more likely to buy QA tools used Google API, Salesforce.com or Oracle.
While these tidbits of data make sense on their own, it’s the combination of thousands of different indicators that create real powerful personas. The more indicators you add, the more robust and powerful your personas are going to be.
This analysis had very concrete results for the marketing team. At the end of the day, Mintigo can take any prospect—inside and outside the CRM—and tell whether it’s a good fit for their web monitoring or QA product and which content or sales pitch will perform best.
While Marketing Automation scores leads based on their history, new leads without any history get no score. Mintigo, on the other hand, relies on data from the Web and therefore is able to score any prospect even before it had any contact with the company.
The client had a database of thousands of leads with demographic data. Mintigo was able to complete much of this data needed from public and private databases and data from the web. Following the segmentation and persona identification, Mintigo was able to understand which product each one is more likely to buy. The team used these insights to create content that resonates specifically with each audience.
6X performance boost
Results were phenomenal. Before the segmentation, an average eBook would get around 1.5% Click-Through Rate. After the personas were identified, performance jumped to 8.8% Click-Through Rate for the web monitoring eBook and 4.7% for the QA eBook—a 577% and 176% improvement respectively.
Customer Intelligence relies on big-data to generate powerful segmentations and personas that marketers can target effectively. We leave buying signals online every day—science can now help marketers take advantage of them.