Intranets Reimagined: A Study On What Employees Really Want

Intranets Reimagined: A Study On What Employees Really Want

Co-Author Tyler Sauerteig

“More than half of employees do not use the company intranet daily, even when the intranet is their homepage.” – Wall Street Journal

Lack of intranet usage is a reality faced by many organizations. You — the intranet, communication and human resource teams — know the value intranets provide. If only more employees and executives used them!?!

The common response to this conundrum is an intranet redesign. Unfortunately, intranet redesigns cost money and do not guarantee success. If only there was a way to know what intranet improvements have the biggest positive impact on your user’s perception, and in turn, drive their continued usage. Well, now there is a way.


Let the Intranet Nerdiness Begin

In partnership with the Worldwide Intranet Challenge, we’ve applied econometric principles to the results of over 200 intranet surveys and … statistically pinpointed which intranet attributes most positively impact a user’s perception of their intranet.

Through our seven week series of data analysis you will learn information on what employees are really looking for.

These results may surprise you! If you’re ready to see intranets reimagined, watch for our next blog post in regards to the statements that employees made about the look and feel of their intranet and how it affected their valuation of their intranet in The Worldwide Intranet Challenge.

What Your Employees Want From Your IntranetFree Download: What Your Employees Want From Your Intranet

In this download, you will find the following information:

  • Ways to increase your employees’ rating of your company’s intranet
  • How to invest your intranet redesign money to get the biggest return
  • Key intranet fundamentals to ensure employee engagement

How We Used the Worldwide Intranet Challenge Survey Data

The Worldwide Intranet Challenge (WIC) provides companies with a web-based survey to assess their intranets and anonymously benchmark them in less than 2 weeks. Their Intranet Benchmark Report, released on April 11, 2014, compared 210 surveys from 187 organizations offering a system of rating intranets from all over the world from 2009 to 2014. This survey consisted of eight categories of questions including:

  • Rate the following statements about the look of the intranet
  • Intranets ReimaginedRate the following statements about finding information on the intranet
  • How often do you use the intranet to:
  • Rate the effectiveness of the intranet in providing:
  • Rate the following statements about how the intranet is maintained
  • Rate the following statements about the performance of the intranet
  • Rate the following statement about the value of the intranet (Dependent Variable)
  • To what extent does the intranet need improvement in:*


The company was then ranked for each survey question in comparison to other participating organizations. This analysis showed how advanced employees believe the intranet is in each category and for each question. Then WIC sends all of the information to the company and how they rank in regard to their competition. We have used their information to take it a step further.

The dependent variable used within the study was the participants’ answers to the valuation statement “In general, I would rate the intranet as.” This shows the overall thought and valuation that an intranet user places on the tool. The independent variables within the study were the other questions that had the impact on the valuation of the intranet and what employees thought of its uses. We saw this as a way to help companies find what those with the highest value of their intranet really look for in their company’s intranet capabilities. For each variable we have taken a two-tailed test approach, meaning that we are going in with no prior implications of what people actually look for in the intranet valuation. Because there are so many schools of thought out there on the topic, we wanted to test all of the variables to analyze the true impact and confidently give your company statistically backed information to help your intranet thrive using the WIC data.


Behind the Numbers: How We Applied Econometric Principles

Econometrics, or an advanced statistical analysis, was run on the numbers for each company. We used Ordinary Least Squares (OLS) to estimate our data and checked for the classical assumptions to use OLS and remove bias. We wanted to test the importance of each question and category to users of the intranet in hopes of statistically concluding things to focus on the most within your intranet build or creation.

Throughout our study we checked for the econometric issues of specification error, serial correlation, and heteroskedasticity. An issue within most of our data is that many of our variables move in a fairly close relation with one another, known in econometrics as multicollinearity. However, in this study we have chosen to disregard the multicollinearity, which in econometric studies is allowed, because we realize the close correlation between many variables in rating their intranet, especially within a certain category of your intranet. Multicollinearity causes our standard errors to increase, which biases our t-stats downward (in other words, causes us to possibly conclude insignificance in a variable that may still be significant). Specification error generally occurs when there is an omitted variable or irrelevant variable within our data. We saw all variables as relevant within Intranets Reimaginedthe study and did not see that we were omitting anything based on the study and the questions WIC chose to ask. However, we still struggled with specification error within the data, which resulted in playing with, and omitting some, of the variables until specification error was not present and we had cut down on some of the multicollinearity. To ensure that specification error was not present within our data we ran the Ramsey RESET Test with a critical point of 5% up to 3 fitted terms. If we had a P-Value greater than .05, then we could conclude that there was no specification error in our study.

Serial correlation in our numbers would mean that the observations of the error term would be correlated with one another. This violates one of the classical assumptions and in turn biases our standard errors used to determine our p-values. This means that we have unreliable P-Values. The test that we used to determine if serial correlation was present was the Breusch-Godfrey LM Test with 2 lags. We again used a critical point of 5%, so if the P-Value was greater than .05 we could conclude that there was no serial correlation within our study.

Heteroskedasticity is an issue in which the error term does not have a constant variance. This violates another one of the classical assumptions and in biases our standard errors resulting in unreliable P-Values. The test that we used to test for Heteroskedasticty is the White Test. We used a critical point of 5%, so again if the P-Value was greater than .05 we could conclude that there was no heteroskedasticity in our study.



*Note: We decided not to utilize this category in our study because it related to all of the categories, the categories were the different statements of improvement so they were even more multicollinear, closely related, picking up the same impact, etc.