Businesses of all sizes are now calling upon their marketers to act as scientists. Decisions about how to manage email campaigns, optimize websites, create content etc. are powered by data and defended by the scientific method.
Huge businesses such as Google, Facebook, Amazon etc. are engaging in more intensive research in fields such as artificial intelligence (AI) than major universities. (Consider Google’s development of a neural network that learned to recognize images of cats on its own.)
But what about small businesses? In an era when marketing science can involve anything from extracting emotions from video feeds to algorithmically derived customer segments, it’s easy to feel like this:
For some years now, I’ve worked in a data-driven marketing department where decisions must be defended with data and hypotheses. My research into business intelligence technologies is informed by this practical experience with marketing analytics.
In this article, I’ll guide you through the complex landscape of marketing analytics methodologies and technologies to help you usher in the era of marketing science at your small business.
I’ll cover the following topics:
The Core Elements of a Data-Driven Marketing Strategy
The first thing to understand when designing a data-driven marketing strategy for a small business is that even though you’ll be using some of the same tools and techniques as a huge business, you don’t simply want to follow in the footsteps of those marketing behemoths.
Being data-driven is all about:
- Starting with small experiments
- Figuring out what works
- Scaling from there
So, even when moving to a data-driven model, small businesses need to explore a scaled-down strategy.
In Gartner’s “Maturity Model for Data-Driven Marketing” (content available to Gartner clients), analysts Andrew Frank and Martin Kihn advise:
“The marketing organization cannot improve its data-driven capabilities only by investing in software and services. Marketers must build their capabilities over time by improving systems, skills, processes and other areas.”
Here are the key elements of such a strategy, all of which have been crucial to our own department’s massive bootstrapping successes:
The 4 Key Elements of a Data-Driven Marketing Strategy
Instead, data-driven marketing covers the analytical tasks that power decisions with data.
Software is crucial in executing such tasks, but people and organized methodologies are just as important. So, let’s take a quick look at each area of analytics in the above diagram before moving on to your software options.
A/B testing is a simple concept. Alter a crucial variable in a marketing asset (the formatting of an email template, the color of a call-to-action button on a landing page etc.), and show both versions of the asset to different user groups. Then, you can “scientifically” see which version is more successful.
A/B testing is a great jumping-off point for a data-driven marketing strategy, because it demands an experimental and methodical approach to business problems such as increasing conversion rates. Only one variable is altered between versions, and there are clear criteria for evaluating which version is more successful.
More complex approaches are known as multivariate testing, as more than one variable is altered between versions. However, multivariate testing is far more demanding both in terms of statistical analysis and experimental design. Small businesses need to start with A/B testing.
Landing page optimization is one of the key applications of A/B testing. In fact, we were able to substantially grow our own conversion rates through this method.
Gartner analyst Martin Kihn lists a number of other important use cases in “Use A/B and Multivariate Testing to Improve Marketing Programs” (content available to Gartner clients):
- Increasing completion rates for checkout and quote processes
- Selecting more engaging content, such as videos and text
- Designing more effective email subject lines and content
- Improving the impact of search and display advertising
In a marketing context, segmentation is the analysis of customer data in order to segment customers into groups. Marketing efforts can then be focused on the highest-value customer groups.
This sounds easy enough, and indeed, segmentation can be performed fairly easily in Excel or another spreadsheet tool using the output of a customer relationship management (CRM) system or a traditional system of record for business intelligence (BI). That is, if your customer data is properly organized in the first place.
Segmentation grows more complex as your data sources increase. Digital marketers will frequently need to use:
- CRM data
- Revenue data (from a BI or ERP system)
- Web analytics data (from Google Analytics, Adobe Analytics etc.)
- E-commerce data (from a shopping cart or order management platform)
- Point of sale (POS) data (from a hardware POS system)
- Social media data
Integrating all of these data types can be extremely challenging. Indeed, simply linking revenue data to web analytics data is a challenging feat, as we explain in our report on content marketing dashboards.
Small businesses should focus on linking customer records to revenue (if the CRM system or system of record doesn’t automatically do this), linking customer records to digital marketing efforts and linking digital marketing efforts to revenue. You’ll then be able to segment customers based on:
- Demographic traits (age, geography, gender, income level etc.)
- Buyer behavior (shopping cart abandonment rates, basket size etc.)
- Online behavior (source of web traffic, engagement metrics etc.)
Once you’ve derived your segments, you can work on turning them into personae. Essentially, the process of creating customer personae involves putting a human face to a segment derived from your analysis, i.e., the visitor who lands on your site via your blog, spends less than five minutes reading an article and then makes a high-dollar purchase.
The important thing to remember is that ultimately, the segment is what matters for your marketing efforts.
Segmentation should be at least a quarterly exercise in data-driven marketing departments.
Advanced segmentation leverages algorithms such as k-means, and typically requires the use of a tool for data discovery.
Siloed data is the bane of digital marketing departments. Data lives in so many places: CRM systems, spreadsheets on hard drives, spreadsheets in the cloud, operational databases, data warehouses—the list is potentially endless.
The key purpose of a KPI dashboard is to tie all of these data sources together and provide visibility into the most important indicators of marketing performance—for instance, total monthly revenue generated by digital marketing efforts.
The challenges involved in building such dashboards, however, are substantial. Siloed data results from unintegrated systems. For example, your web analytics system is most likely separate from the system that records revenue. Integrating these systems can be incredibly challenging and costly.
Thus you have two basic options:
- Use a BI system that can connect to all of your data sources, pull them into a common store and generate web-based dashboards.
- Integrate all of your data using spreadsheets.
- Outsource your analytics to a provider of reporting services.
If you go with option two, chances are that you’ll need to become very familiar with lookup tables in order to weave different types of data together across spreadsheets.
Excel is definitely cheaper than a BI system or outsourcing. However, you must also factor in the human hours spent on manual data integration when making this assessment, as well as the significant potential for error in manual data integration.
And ultimately, if your business continues to grow, you’ll hit a point where you have too many data sources to perform spreadsheet data integration efficiently.
We offer guides on how to create dashboards for the following purposes:
Opportunity forecasts can be fairly straightforward. For instance, a digital marketing company can look at three variables with respect to landing pages:
- Conversion rate
- Number of conversions
It could be the case, for example that page A has a higher number of conversions and gets more traffic than page B, but has a far lower conversion rate.
One could reasonably conclude that there’s an opportunity to drive more converting traffic to page B than A, since page B has the better conversion rate, and page A is already getting decent traffic.
Such basic opportunity forecasts were key to how we scaled our own digital marketing efforts in the early days. More advanced opportunity forecasts depend on predictive algorithms that leverage complex statistical techniques to extrapolate future trends from historical data.
Keep in mind, predictive analytics should be the endpoint of your journey to a data-driven business model, not the beginning.
There’s potential for devastating errors with predictive analytics if you build your own models. Models that work out of the box do exist, but they’re typically bundled into either BI suites or enterprise-grade business applications for CRM, ERP etc.
The Technology Stack for Small Business Marketing Analytics
Now that we’ve taken a look at the most essential analytical tasks in a data-driven marketing model, let’s take a brief look at the software that supports these tasks:
- Excel and other spreadsheet tools are key to segmentation, data integration and basic opportunity forecasting. KPI dashboards can also be created as spreadsheets.
- Web analytics platforms such as Google Analytics and Adobe Analytics are key to linking online behavior to customer records and revenue.
- Enterprise systems of record that support data warehousing are a key element of the technology stack at small businesses that have already moved to unified BI.
- Modern BI systems are key to integrating data at businesses without a central reporting system. These platforms can connect directly to a range of data sources, including relational databases used primarily to record transactions, and generate web-based dashboards.
- BI systems with advanced statistical modules represent the next level for small businesses. These platforms can take opportunity forecasts from an art to a science via sophisticated predictive algorithms.
- Self-service data integration platforms can be used to integrate particularly tricky siloed data sources.
If you have more specific questions about selecting technology to support marketing analytics at your business, you can contact me at email@example.com.