How to run an employee demographics survey
Written by Clara Greo, Harry Vos, Adam Robertson and Martin Lugton.
Illustration by Mariana Gonzalez Vega.
A demographics survey is an effective first step towards becoming an equitable organisation. It can reveal the outcomes of your unfair recruitment. It can lay the foundations of how to make better services and products. It can be a great way to collect reliable, revealing and actionable data.
But, it’s easy to get that step wrong. It can also be a way to further exclude people. You can collect skewed data that sends you down the wrong path. Or it can be virtue signalling that doesn’t result in any change.
For the best chance of running your survey well, these are some principles to hold close:
- Understand why you are doing this work. Write down your aims, interrogate them and be transparent about them
- Do no harm. Create no further oppression. Protect and be respectful of marginalised, oppressed and excluded people and their data
- Listen to and believe what people tell you about themselves
- Collecting data is meaningless if it doesn’t lead to action and change. Find out why and fix it
- If you get it wrong (and you probably will), apologise and do better
- Collecting accurate data is less important than all of the above
We have faced lots of challenges running demographic surveys and learnt lots of lessons. This blog post outlines how we recommend you go about it. We’ll be writing a follow up post on some of the things that have gone wrong.
Let people know why you’re doing it
What your aims are and how you communicate them will impact the data and results of your survey.
Your colleagues will need to fill in your survey, so you’ll need to get their trust. You need to build a shared understanding of why they’re filling in your survey. Organisations already collect some demographic data in HR systems. Colleagues can be hesitant to fill out more information.
It’s important to speak with colleagues early on, share why you’re doing it and what you’re doing to keep the data secure. This will help underline the need for your survey.
If you’ve been working with your HR or people team on your survey, make that clear upfront. You need to make sure your colleagues are okay with this.
Ensure you do all this communication every time you run your survey so you continue to get buy-in. If this is not your first survey, make it clear what’s changed (if anything) since your last survey based on feedback.
On a practical note, you want to make it clear to colleagues what you’re asking them to do. For example, ‘We need you to fill out this survey once it launches. It will take between 5 and 10 minutes. It’s important that everyone fills it out’. Unless you’re on the executive board, you probably won’t be able to make the survey mandatory. But, we used firm messaging saying, “We need 100% of people in product management roles to complete the survey.”
Run your survey
While we surveyed communities within our organisation, you could also survey other groups of people. For example, attendees of an event or members of an organisation.
From the start, Clara designed a survey to make people pause and think about the privileged demographic groups they belong to. It was a conscious choice rather than optimising for speed. This choice also made it less confronting for people in marginalised groups.
If you’re surveying employees across multiple locations, you’ll ideally want to have separate surveys. This means you can compare your results to local populations rather than national or international.
But be aware that, if you expect a small number of people to complete your survey (around five or less) either in a particular location or in total, you'll need to take extra precautions to protect people’s privacy. You could combine locations to increase the number of people completing each survey. For example, surveying EU colleagues rather than surveying Denmark and Sweden individually. Or, you could remove questions about less visible and partially visible characteristics. We'd want to avoid colleagues guessing who might be masking their neurodivergence or a queer teammate who hasn’t yet come out to their colleagues.
As we were running the survey in the UK, you might need to make the questions more relevant to your location. Where possible, use your country's official statistics classifications, so you can make comparisons.
When launching, share a link to a start page with your colleagues, not your survey. This gives some context before your colleagues rush through your questions. This is especially important for people hearing about your survey for the first time. The start page also helps colleagues to find a privacy notice.
You could ask line managers to remind their line reports to complete your survey. Make sure you have enough time to answer any questions about your survey while it’s live. Run your survey for three weeks so that people can still complete it, even if they’ve taken a couple of weeks off.
You should be able to chase people to complete your survey, but you won't know who hasn't completed it. You can send out messages like “three people haven’t completed the survey”.
Analyse your results
Copy the numbers from your survey into an analysis spreadsheet. To compare the demographics of your employees, use local population statistics where possible. If not, use national statistics.
Some people might not finish your survey. As a result, some of the demographic group sizes will be inaccurate. You need to adjust for this. Consider skewing your results to make less-privileged groups smaller than they might actually be. This could lead to slightly more interventions in these communities. But we’d argue this is better than masking what actually might be an underrepresented group.
Once you've got the percentages of each demographic group, consider making the tallies private. This makes it harder for colleagues to try to identify people amongst small tallies of partially-visible characteristics.
Share your results to drive targeted action
When sharing your results, make sure everyone understands why this work is important. Show the connections to your organisation's diversity, equity or inclusion goals. This should show how your survey, and the resulting actions you’ll take, will support those goals.
Once you’ve identified some early actions to make improvements, share them with colleagues and management. Show them your initiatives that aim to make your organisation more representative of wider demographics. Highlighting early progress will put you in a better position to publicly share less representative demographics.
Try running a recruitment event aimed at the underrepresented groups you’ve identified. More privileged groups will already feel more welcomed by your organisation and have better access to job application tips. The event should be welcoming and help people understand your organisation. What does the work entail? How do you navigate the recruitment process etc?
If bosses will genuinely recognise them for doing it, ask your colleagues from minority groups to present. Presenters can talk about their own experiences and give people the best information to put in a competitive application. You can also offer follow-up 1:1 sessions for specific questions or support.
You could also focus on sharing job adverts with communities of your underrepresented groups. Or offer mentoring opportunities within your organisation to help people progress.
Let us know how you got on
Once you’ve blogged about the results and what you’re going to do about it, we want to know how you got on. Please send us your questions, feedback and links to what you publish.
Clara Greo is on Mastodon
Harry Vos is on LinkedIn
Adam Robertson is on LinkedIn
Martin Lugton is on LinkedIn
Learn more about equitable recruitment
Clara Greo recently shared her recruitment advice on a podcast with the UK Department for Education. Give it a listen!
In Copenhagen? Harry is looking for work
Photo on screen: Some rights reserved by Jorge Franganillo.
I’ve recently moved to Copenhagen, Denmark. As well as wanting more organisations to run demographics surveys, I'm also looking for my next product management role with opportunities for social or environmental impact. If you’re interested, please connect and let’s get kaffe!
Previous: The value of testing riskiest assumptions