Really in-the-know

Learn the latest B2B marketing industry trends, techniques and news stories on our B2B blog to stay ahead of your competition

Boost your marketing campaign with these data modelling techniques

Posted by Jonny Drodge on 23 Oct 2019, 11:10:00

With the ever-growing complexity of B2B marketing, marketers are increasingly expected to move away from vanity metrics towards real actionable measurements for ever more transparent and trackable campaigns. They also need to prove real ROI and use evidence and data to aid in campaign decision-making. But, with all the different analytical models out there, how can marketers effectively put on their statistician hats and ensure they are using data effectively in their campaigns?

Using a B2B marketing agency with dedicated data and analytics teams can help your data work harder for you and ensure you’re always optimising your marketing decisions to get the most from your budget. If you’re starting to make data driven decisions yourselves, here are a few simple tips and models to get started.

Cohort analysis

One of the first steps in looking at your marketing database is to split it into segments, also known as cohorts. These can be groups of prospects in similar industries, job functions or geographical areas, or even groups who have experienced similar events previously, or share another characteristic.

Segmenting your data into cohorts allows you to compare, experiment, and find the most or least engaged or profitable groups. In the B2C space, this is helpful when comparing segments’ performance over time. In the B2B space, we can compare firmographic attributes of our prospects as we don’t tend to have a continuous flow of new prospects and customers all following the same journey.

Here are some tips for splitting your cohorts out:

  • 1. Try to keep your segments organised by just one main characteristic so that you can be sure what’s driving any differences.
  •  
  • 2. Whether in your CRM system, your marketing automation tools or other database storage, clearly separate your cohorts so that you can track how each pot performed when testing marketing strategies.
  •  
  • 3. Isolate your variables, if you test one thing at a time (messaging, subject lines, channel, timings etc) with your cohort analysis you can be confident of the reason for the change in results.
  •  
  • 4. In the B2B space we track the progression of prospects through to MQLs, SQLs, and conversions until sale, so at the most sophisticated level a cohort analysis would need to account for all these stages.

 

Propensity modelling

Propensity modelling (sometimes known as behavioural clustering) is a powerful tool in predicting the behaviour of a wider database based on known behaviours of a look-alike proportion. This is especially key in B2B campaigns where many of your appointments or sales may come from a known core segment of customers. But, what about your other wins in smaller niche segments, could you split them out for a targeted campaign or buy more data in that niche to maximise your results?

Propensity modelling can be complex, potentially looking at hundreds of behaviours across the B2B buying journey, and a number of attributes, so let’s look at an example:

Company A knows that 50% of their sales come from procurement contacts in the construction industry. But 50% of their target prospect database also sits in this industry, so the actual performance of the behaviour of “converting to a sale” is about average! (50%/50% = 1, where 1 is a perfectly average score). However, company A also has 5% of their lead database in the manufacturing industry generating 10% of their sales. When applying similar analysis (10%/5% = 2) we can tell that on average, a manufacturing lead is twice as likely to convert to a sale. Using this information company A can acquire more manufacturing data, perhaps create a bespoke messaging campaign to this group, and reap a better ROI on their marketing investment.

This is just a simple example, in the B2B space where buyer journeys can be long and complex, with several stakeholders and perhaps several behaviours you wish to track , much more in depth analytical models can be created.

Here are some tips for getting the best out of your propensity models:

  • 1. Ensure you can actually track the metrics for each stage of the funnel you care about, and integrate those into your data from the outset to avoid having an incomplete picture.
  •  
  • 2. Think about the stage of the funnel you’re looking to optimise. Is it the top of the funnel? If so, looking at behaviours such as propensity to engage, have a hot response from a lead, or convert to an MQL is key. Or is it your end of funnel? If so, looking at propensity to convert from SQL to sale or even customer spend will yield more insight.
  •  

Attribution modelling

Attribution modelling is the process of assigning value from a conversion, lead, or sale to a specific marketing channel or activity in the customer journey. For example, the commonly used last touch attribution gives 100% of the credit to the very last touch point in the journey between you and the customer. While this is simple, it often doesn’t tell the complete story about the value from earlier touch points and channels in your funnel. Alternatively, other models like first touch can assign value at the start of the funnel (if all you care about is reach and growth). In reality, a good agency will work with you to determine the best model for your unique circumstances . The right model can help you determine where your return on marketing investment (ROMI) is coming from, and help you optimise your marketing budget to channels where it will produce maximum impact for you.

B2B marketers have historically lagged behind the B2C space in picking the correct model for their campaigns due to the complexity and number of people involved in the typical B2B journey. In fact, 57% of marketers aren’t confident they’re using the correct attribution model.  So when running complex multichannel campaigns, its more important than ever to get your attribution nailed to ensure you can put your budget where it’s needed most.

Implementing a new attribution model from scratch can be a big task, but to get started consider the following tips:

  • 1. Take the time to fully understand your campaigns’ customer journey and fill in any “missing touch point” tracking.

  • 2. Ensuring that your marketing database is clean and that you have processes to track every touchpoint will make reporting on customer journeys simpler.

  • 3. There are plenty of vendors popping up in recent years offering full attribution software. While pricey and with lengthy onboarding processes, these vendors could offer a full solution. But, make sure you know what your needs are before committing to expensive software.

 

Bringing it all together

These are just some of the modelling techniques to getting more value out of your marketing campaigns and this is by no means an exhaustive list. Good data practices, savvy implementation, and statistical rigor are key to getting the best out of these techniques.

At Really B2B, we have a dedicated team of B2B marketing analytics experts on hand to answer any questions you might have.

Get in contact by giving us a call today on 02392 314498.

 

Topics: B2B Data Analysis, b2b attribution modelling, b2b propensity modelling, b2b marketing analytics, b2b cohort analysis