Lean customer-journey analytics on a shoestring

Posted on Posted in Customer Journey, Lean customer-journey analytics

In the coming weeks, I’m going to publish a series of more in-depth, technical articles with tips and example scrips for lean customer-journey analytics. Keep an eye on the blog or follow us on LinkedIn or Twitter.

Lean Customer-Journey Analytics

The way customers interact with brands is changing rapidly. The customers’ paths-to-purchase consist of multiple interactions with your brand across many channels and everywhere they leave digital traces. Already, some brands are using these digital traces to better understand the customer and to improve and personalise their experience. Soon this will become the norm, and brands that stay behind will see their customers leave for the competition. Customer-journey analytics is great way to understand your customer’s behaviour and to improve their performance. In this blog post I will explain what customer-journey analytics actually is and how you can introduce it in your organization.

Journey mapping versus journey analytics

Customer-journey maps are a visual representation of all the steps that customers take on their path to purchase. Typically, it’s created based on interviews with stakeholders within the company and sometimes a small sample of customers. Journey analytics also produces visual reports, but differs from the simple map in a number of significant aspects.

Real, not imaginary journeys. Journey maps all too often describe journeys imagined by marketers and CX experts, not necessarily the journeys that really take place. Journey analytics is data driven and distills the paths-to-purchase from the billions of digital traces your customers leave behind. Journey analytics shows you what really happens and can uncover pain points that no CX expert could imagine.

Customer segments. Simple journey maps show a single, average customer journey. More sophisticated maps will break the customer base into a number of personas (customer segments). But how many customers of each type are there? Are there segments that you are missing? Good customer-journey analytics tools use machine-learning algorithms to automatically discover customer segments in the journeys. The segments than can be correlated with various KPIs, such as conversion rate, lifetime value, churn rate, etc. That allows brands to prioritise resources when improving customer experience: start with the most valuable customer segments.

Dynamic picture. While traditional journey maps are static, journey analytics always provides an up-to-date picture, as your customers’ journeys are changing. This is particularly important for business undergoing digital transformation.

Levels of detail. A traditional journey map is a single picture, whereas interactive reports produced in a journey analytics process will allow you too zoom in on important and interesting segments.

Use case: journey analytics for a bank

A traditional journey mapping exercise will uncover that a mortgage customer first fills out an online form on a website and then has two appointments at the bank. Journey analytics can answer more detailed and complex questions and uncover unexpected insights:

  • Only a small percentage of the customers take this path. Most of them visit the website several times before the first appointment, and contact the call centre a couple of times during the process.
  • Customers that in the end purchased a mortgage have a different behaviour on the website that customers that did not purchase.
  • The questions of a large group of customers could be answered by the FAQ section on the website, but it proves hard to find which in turn generates many calls to the callcenter. Improving the website could save a lot of cost.
  • A certain group of customers always require contact on the phone. These are the ‘complicated’ cases. With this insight, they can be discovered earlier in the process and redirected to specialised department which will service them faster without many unnecessary call to the ‘generic’ call centre.

The insights uncovered in the journey analytics process will allow the bank to fix various pain points in their process, tailor their service for different groups of customers and save cost. While introducing these improvements, the customer journeys will be constantly monitored to make sure that the improvements are yielding the expected results.

How to introduce journey analytics in your organisation

Proper journey analytics needs input from many sources: website analytics, mail campaign manager, CRM system, etc., etc. Big companies like Adobe or IBM offer full software suites that cover many of these functionalities and automatically connect data points from various channels. To use them, however, you will need to replace all your existing systems, not to mention the very high price tag. Also, once you start working with such an all-in-one vendor, you will run into trouble if you need a functionality they don’t provide. You’re stuck with what they have.

Lean journey analytics

Another approach is to start small by integrating data from your existing systems and keep complete ownership of your data. A certain amount of custom work will be necessary and you will need some analytical/programming skills, either yourself or you’ll need to hire it externally. After all, no two organisations have exactly the same needs and combination of software installed. However, the cost and duration of such projects is a couple of times lower than with all-in-one software suites.

Step 1: do a checkup of your data

For a customer-journey analysis, you need data at the individual level, i.e. you need to know what each individual person did. For example, for your website, you need a history of clicks from each visitor, rather than total number of visitors for each page. If you use Google Analytics to track your website visitors, you will need the Premium version to get out the individual level data. Fortunately, there are many other web analytics tools, that can also work alongside GA, for example Piwik. You can also use your web server logs (all web servers produce logs, so does yours).

Second, you need to be able to connect people across different data sources. This doesn’t mean that you have to purchase a Data Management Platform right away. There are tricks that can be used to connect the data sources during the analysis. For example, asking people to log in on your website will allow you to connect their web histories to your CRM system. A loyalty card is a great way of connecting online and offline purchase data.

Step 2: define a small pilot project

Start with a small pilot project. Choose two or three channels that you think are most important in your journey and run a quick analysis on the data from these channels to find out if there are interesting insights at first glance. In the example bank case above, only the website and CRM system needed to be analysed.

Define what is the endpoint of your customer journey. Is it a conversion? If so, how do you define it? For some companies it will be simply an online transaction, for others, it will be a sequence of events, like repeat purchase, or a download followed by a purchase followed by a recommendation.

To analyse your data, you don’t necessarily need expensive analytics and visualisation tools. There is a lot of open-source and low-cost software that can be used: from Python, SQL databases, to various visualisation tools and libraries. Look for tools that offer flexibility in defining what your customer journeys is (i.e. what is the endpoint) and that are data-source agnostic, i.e. can work with all types of data sources. Generic data-analytics languages, such as SQL, R, Python and Rax fall in this category.

You will also need someone with data-analytics and some programming skills. If you have a data-scientist on your team, that’s great, otherwise, you can hire an agency. Don’t be scared of the cost of the latter. A good agency can complete a pilot project within 2-3 weeks.

Step 3: expand on the pilot

The pilot analysis will in most cases already yield actionable insights. But often additional questions arise. You might find out that:

  • Your data quality is not high enough, you need to improve your data collection.
  • You’d like to connect data from more channels.
  • There are certain interesting behavioural patterns that require further analysis.You can then dig deeper into the data. Also, a follow-up interview with the individuals displaying these patterns might be useful.

Step 4: monitor the performance

Journey analytics yield actionable insights that you can start implementing in your organisation: improving your website, personalising email campaign, improve your call centre scripts, etc. etc. Monitoring the changes in customer behaviour is very important to make sure that your improvements have the desired effect. The analysis has to be easily repeatable to closely monitor the improvements. The use of scripting languages like R, Python or Rax will make it easy.

Example journey-analytics report.

Conclusion

Journey analytics is a great tool to improve your customer experience and operational efficiency.  However, it requires combining data from many, often disparate systems. Some big brands offer software suites that solve this problem, but their price tags are scaring many organisations from experimenting with journey analytics.

Luckily, a couple of smart people armed with open-source and low cost software can solve this problem too. You can have your first data-driven journey map for just a few thousand Euros.

Keep an eye on this blog for more practical tips, including example scripts, on lean customer-journey analytics.

This article was originally published in Dutch on Frankwatching.