Resources

Article

Bridging the gap between content creation and data analysis

Article

Bridging the gap between content creation and data analysis

Article

Bridging the gap between content creation and data analysis

Business insights

Article

Bridging the gap between content creation and data analysis

Business insights

<iframe width="1116" height="628" src="https://www.youtube.com/embed/I5-pDt_6ZTA" title="Michael Sants - Web page value project (s)" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen></iframe>

“In Visma, we have thousands of web pages with lots of great content that we hope and believe will serve its purpose and generate good results. But what if we could use data to predict what content will do well?” says Michael Sants, one of this year’s Management Trainees.

<iframe width="1116" height="628" src="https://www.youtube.com/embed/I5-pDt_6ZTA" title="Michael Sants - Web page value project (s)" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen></iframe>

—This project allows us to bridge the gap between content creation and data analysis

In this nine weeks-long project, Michael has analysed the content on our existing web pages. By looking at the conversion goals we have set for different pages and how well these pages convert on those measures, he’s been able to create a page value score. Using this value score, we can better indicate what content is performing well–and what is not.

This will not only help us improve our content creation strategy but also optimise our existing content because we will get a better understanding of what page elements work well.

“This project allows for bridging the gap between content creation and data analysis. The two might seem contrasting at first glance. However, they can be complementary to one another in today’s age of big data,” he continues.

So, how did he go about setting a value score?

Setting a page value score to determine high-performing and low-performing pages

“We produce a lot of content in Visma: We have thousands of pages with different content. Could we change or remove the pages that do not serve their purpose (produce value)? Can we improve our existing web pages?” These were the first, initial questions that sparked the Management Trainee project that Michael set out to solve.

Also, how can we enable better collaboration between content creators, web editors and data analysts in Visma to make sure that we create and optimise for high-quality content that generates value?

Michael explains that to assess whether a page brings value, we needed a way to quantify it through a value score. This value score aims to help our Visma companies improve their web pages by providing clear calls-to-actions for taking more data-driven decisions.

This will also get the content creators, web editors, and the data team involved in the understanding of: Why do we create this page, what do we hope to achieve—and are we achieving it?

You might also be interested in: Three important skills of a Management Trainee

Creating an overview of today’s success metrics and relevant variables for a value score formula

The project started with creating an overview of how we measure the success of our different content pages types today. What are the metrics we use to measure whether a product page is performing well or not?

What are the metrics we use to measure the performance of our content on other page types or our other platforms such as Visma Community and our blogs? Based on this insight, Michael and the project group looked at what variables would be relevant for a value score formula for product pages specifically.

Different weights were then given to the variables depending on how important we deem them to be for the page to be successful:

Content KPI identification master sheet

Based on this data, a value score formula was set to assess the performance of a product page:

Value score formula to assess the performance of a product page: page views score equals 10%, engaged time score equals 20%, CTA clicks score equals 20%, form submit score equals 50% = value score of 100%.

This way, we could get an answer to the question: which of our product pages on visma.xx performs well?

Illustration of different value score formulas for different page types: web, blog, Community pages.

Also read: Say hello to the Management Trainees of 2020/21!

From spreadsheets to the Visma Data Lake

A computer SQL query was then made to extract the relevant data from the web pages and from the Visma Data Lake.

A SQL query is a domain-specific language used in programming, which is designed for managing data that is held in a relational database. The Visma Data Lake is our internal system, or centralised repository, where we store all structured and unstructured data.

A step in this process was to create a code that runs on every page to determine if the page is still active. This way, only active pages are included in the overview of scored pages.

“Data Activation is the concept of unlocking value in data through the development of insights and turning those insights into action,” says Michael.

The scored pages can now be found in a Tableau report where the user can filter on domain and scoring. This means the user can easily get an overview of pages that should be improved, pages that could potentially be removed, and high-performing pages to get inspired by.

Results of the project

As a result of this project, we are now able to begin to see what product pages in our different national websites are performing well, and which pages are not. Based on these insights, we need to start looking into what these pages have in common.

“Which features on the page drive a high or low-value score? It can be things like content length, the use of images, the colours on the page, the words used, sentence structure. Here, it will be beneficial to include both UX designers and content creators to analyse the features,” says Michael.

Illustration of value score based on content features on a page versus the page content itself and metadata.

Once we have these insights, we can start running A/B tests on specific page elements, page structure, and the page content itself.

Here is one example of what some high-performing product pages have in common:

Illustration of what some high-performing pages have in common: data-driven team, clear page structure that is continuously tested, and assertive (the team creating these pages are not afraid to kill old pages or acknowledge improvements).

This is, naturally, not representative of all the product pages that live on Visma.xx domains as we still need to gather more data to get a better scoring outcome.

“Some results we already see from this project is that we need to improve the data quality. We have lots of long-existing web pages that aren’t tagged with necessary metadata that we need to gather enough data, or that are incorrectly tagged.

We also need to collect missing data: What data do we need to have that we don’t already have today? Are there any inconsistencies in the data?

Lastly, we need to continue to inspire people to work data-driven. “This can be done through showcasing the value of data gathering and the impact it has on decision-making abilities when creating new content pages,” says Michael.

Related content