Making speaking up at work worth it

Feedback that went nowhere

Companies collected feedback. Little of it turned into change.

One of our customers had seen average employee turnover rise 28% in two years, and 65% of their exit interviews named “unaddressed workplace frustrations” as a reason for leaving. The feedback was there. It just never led anywhere.

The problem looked different depending on who you asked:

Three people, one broken loop. Employees spoke up, managers couldn’t tell what to fix first, and HR couldn’t prove the value of fixing it. So nothing moved.

What the research showed

Listening across companies, roles and levels

I led the research myself, starting at INGKA and then widening it to other industries:

• 24 contextual interviews at INGKA across departments and seniority levels. I ran 14 of them.
• 8 interviews at other organisations in tech, healthcare, finance and manufacturing.
• 16 sessions with HR and leadership stakeholders, all facilitated by me.
• A review of existing survey data and exit interviews, and a competitive analysis of the feedback tools on the market.

Many complaints, four causes

Three groups needed different things from the same loop. Employees wanted to be heard without extra work. Managers wanted to know what to fix first. HR partners wanted a clear story that drives decisions and proves impact.

One loop to close

Another survey tool wouldn’t fix this. People needed to see feedback the way employees experience work: as a journey of moments, each one good or bad for a reason.

Turn feedback into a map of the employee journey, so anyone can see which moments matter most, and why.

Four principles followed from that:

Choosing what to build first

With Fount’s decision makers, I ran a prioritisation vote on what was most feasible, most valuable and most likely to bring in revenue at our stage. Then I ran an ideation workshop with stakeholders on the journey and flow for each group. Eight features made the first plan, and four waited for later, including an AI agent. Within that plan, version one shipped the core: quick pulse surveys, and a journey view that turns the answers into priorities.

Act 1 · Fount: from feedback to a map of the journey

Fount had two halves that worked as one loop. Managers and HR send short pulse surveys at the moments that matter, and every answer lands on a map of the employee journey, so anyone can see which moments are working, which aren’t, and why.

Every new user lands on the same welcome screen. It shows both halves side by side, each with a preview of the real thing, so people know what they’re getting before they click.

1234

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1A welcome that says why. The first words explain what Fount is for, in one short paragraph, with docs one click away.
2. 2Two ways in, one loop. The two cards match the two halves of the product: surveys to collect feedback, and the dashboard to make sense of it.
3. 3Show, don’t describe. Each card previews the real screen, a survey question and a heatmap, so the choice is easy without reading.
4. 4One clear next step. Each card has a single Get Started button, so nobody wonders where to begin.

1. Pulse surveys in minutes

PrincipleQuick to give

Get Started on surveys opens straight into templates, so the product starts with the thing that brings in data.

Before

One long annual survey, with results promised months later.

Now

Short surveys built from templates, sent to the right people for a set time, and quick to answer on any device.

1Creating a surveyClick a tab to follow the flow, and see the decisions behind each step

Start from a templateBuild itChoose who and whenYour surveys

1234

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1A head start. Get Started opens on ready-made surveys, so the first one goes out in minutes.
2. 2Templates for real moments. Each template targets a moment where friction tends to hide, from a first day to a new policy.
3. 3You see the question before you pick. Every template shows a real question and its answer type.
4. 4Or start from scratch. The blank option is there, but it isn’t the default.

1234

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1See it as you build it. Every question has a live preview beside it, exactly as employees will see it.
2. 2Question types, not blank fields. Pick a mood scale or a text area, and the right options appear.
3. 3A recommended range. Five points is marked as recommended, so people don’t have to guess.
4. 4Required only when it matters. The mood question is required. The written one is optional, so a quick answer still counts.

123

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1A start and an end. Every pulse runs for a set window, so it never becomes another always-open form.
2. 2Only the right people. Each survey goes to a chosen audience, so people only get surveys meant for them.
3. 3Launch when it’s ready. Setting up and sending are separate, so nothing goes out by accident.

123

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1Named by purpose. Each survey says what it’s for, like finding friction for new joiners, so a team can tell them apart at a glance.
2. 2Everything in one place. Every survey a team runs lives here, ready to edit or reuse.
3. 3Always one click from a new one. The New survey button sits at the top of the list.

2What employees seeCheck the same survey on a phone and on a desktop

On a phoneOn a desktop

1234

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1Two questions, done fast. Short enough to answer between meetings.
2. 2Built for thumbs. The mood scale wraps onto two rows on a phone, so every option stays easy to tap.
3. 3A warm opening. The survey opens with a friendly line, not instructions.
4. 4Check both, then share. Switch between phone and desktop, and share the preview before launching.

123

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1Works anywhere. The same survey runs in any phone or desktop browser.
2. 2One row on a wide screen. On a desktop the whole scale fits in one row.
3. 3One clear action. A single button to submit, nothing else to decide.

2. The whole journey, at a glance

PrincipleSee what matters most

Answers are grouped into a journey, like Q4 new joiners, then into the moments inside it, like meeting the team or requesting time off.

Before

Feedback copied into spreadsheets by hand, themes guessed, and no way to tell which problem mattered most.

Now

One overview of the journey, then a heatmap and a key driver chart that show which moments to fix first.

1The overview

12345

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1One score to start from. The journey’s Net Promoter Score (NPS), with how it changed since last time.
2. 2What drives it, ranked. Key drivers are ranked by impact, so the first thing to fix is at the top.
3. 3Impact in plain words. High, mid and low impact, in colour, instead of a statistic to interpret.
4. 4The best and the worst, side by side. Top moments sit next to the weakest touchpoints, so wins and problems get equal attention.
5. 5Every number shows its sample size. A response count sits beside each score, so 30 answers aren’t read like 3,000.

2Finding the moments that matterClick a tab to see each view and the decisions behind it

HeatmapKey driversList viewChoose a moment

1234

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1Hot spots stand out. Colour shows where a moment is failing for a group, without reading a single number.
2. 2Compare any group. Switch between gender, job code and location to see who a problem affects.
3. 3Start broad, then narrow. All moments first, then pick one to go deeper.
4. 4Two views of the same data. Switch between the heatmap and a list.

123

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1Where to act first. Each moment is plotted by impact and satisfaction. High impact with low satisfaction is where to start.
2. 2Pick the measure. Run the analysis against NPS or another outcome.
3. 3Trends, not snapshots. Moments over time show whether a change is working.

123

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1Every score has a benchmark. A 65% means little alone. Next to a 72% benchmark, it’s a problem.
2. 2Up or down, in colour. The difference from the benchmark is red or green, so gaps jump out.
3. 3Which results matter. Impact significance shows which results are worth acting on.

12

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1Everyday names. Moments are named the way people live them, like receive laptop or meet team members.
2. 2One click to go deeper. Pick a moment and everything below it updates.

3. Inside one moment

PrincipleKnow why, not just what

Each moment breaks down into touchpoints, the people and things that shape it, and attributes, what people actually felt.

Before

A slide with a chart and “further investigation needed”. Leadership saw what was low, but never why.

Now

Every moment shows its trend, the touchpoints and attributes behind it, how each compares to the benchmark, and the comments behind the numbers.

1One moment, in depthClick a tab to see each view and the decisions behind it

The momentTouchpoints and attributes

123

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1The numbers that matter, in one line. Score, change, benchmark and responses sit together at the top.
2. 2How it’s moving. Satisfaction over time shows whether things are improving.
3. 3Straight to what people wrote. Analyze feedback opens the comments behind the numbers, so a score is never the end of the story.

123

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1Who shapes the moment. Touchpoints show which people and things made the moment good or bad.
2. 2What people felt. Attributes turn answers into plain statements, like helpful conversations or enough feedback.
3. 3The bright spots too. Colleagues score 9% above the benchmark, so teams can see what’s working as well as what’s broken.

A year later: what was still slow

Fount worked, and people used it. A year of real use also showed where the work was still slow. The feedback was coming in, but HR teams were drowning in it:

So I ran a second round of research over three weeks: 18 HR directors and VPs, 12 People Operations managers, 9 team leads and 4 C-suite executives.

“I feel like I’m always putting out fires instead of preventing them. By the time we see patterns in our data, the damage is already done.”

People Leader, manufacturing company

That was a real need, and a business opportunity, so we built Fount AI.

The journey, before and after Fount AISarah, a VP of People Operations, from spotting a problem to solving it. Switch between them with the tabs.

Before Fount AI After Fount AI

1234

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1It starts once the damage is done. The process only began when someone left or a complaint escalated. There was no early warning.
2. 2Most of the time went on gathering. HR pulled exit interviews, surveys and Slack together by hand, from six to eight tools. 80% of the time went on collecting, not analysing.
3. 3Weeks of debate. Spreadsheet analysis with no benchmark led to meetings where people argued over what the data meant.
4. 4Twelve weeks, and the mood never lifts. Sarah stays frustrated at every stage. Even at the end, she can’t measure whether the fix worked.

1234

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1Found in real time. Continuous monitoring spots patterns before they escalate, instead of waiting for an exit.
2. 2Answers with reasons. Each summary explains itself in plain language, with a confidence score and its reasoning, so Sarah can check it.
3. 3Options she can shape. Recommended solutions come with resource estimates and chances of success, and Sarah can adjust them.
4. 4About two weeks, start to finish. Progress and impact are tracked as the fix rolls out, so course corrections happen straight away.

Act 2 · Fount AI: a business partner for HR

Fount AI finds friction early across different sources of feedback and recommends what to do next, backed by employee data and proven practice. Three strategic calls shaped it:

4. Ask Fount AI

PrincipleA conversation, not another dashboard

HR asks in plain words. Fount AI reads the comments behind the moments that matter most, names the main problems, and suggests what to do.

Before

HR checked six dashboards a day and pulled data into Excel to find patterns.

Now

Ask a question and get the main problems, the comments behind them, and fixes with their sources.

1One conversationClick a tab to follow the conversation, and see the decisions behind each answer

The problemsThe fixes

1234

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1It says what it read. Before it answers, Fount AI says what it looked at: the comments on the top five moments.
2. 2Tied to a moment. Every answer is tied to a moment from the journey map, so it links back to the rest of Fount.
3. 3Problems, with proof. Each challenge shows how many comments back it, and opens to show them, so HR can check the source.
4. 4Fixes that match the problems. Recommendations sit right under the problems they answer. E-signatures speak straight to the slow approvals.

1234

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1It asks before it acts. Fount AI offers the next step and waits for a yes, so HR stays in charge.
2. 2Feedback plus proven practice. Each fix draws on what employees said and on approaches that have worked elsewhere.
3. 3Sources you can open. Links show where the ideas came from, so HR can read more before deciding.
4. 4The next question is one line away. Every answer ends at the message box, so following up feels like talking to a colleague.

5. What to fix first, and why

PrincipleHelp people decide

Knowing what to do isn’t enough when budgets are tight. HR has to know what to do first, and be able to defend it.

Before

Dashboards showed what was happening, but not what to do about it, or where to start.

Now

For each moment, Fount AI picks the fix with the highest return, shows the hours and money it saves, and explains why.

1The highest return first

1234

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1The question HR really has. Which fixes give the best return if done first, for every moment at once?
2. 2One fix, clearly ranked. The top recommendation for each moment is labelled Highest ROI, so the order is never in doubt.
3. 3A case leadership can fund. Hours and money saved a year turn a people problem into a business case.
4. 4The reasons behind it. Supporting reasons explain why it works, so HR can defend the choice to leadership.

Tested with the people who’d use it

I tested both products with the people they were for, at INGKA and beyond. The clearest request came in the AI round: customers wanted the biggest problems first, then the comments. That’s how Fount AI answers, starting with the top five moments.

Usability testing, both productsFount’s dashboard in August 2023, then the AI features with customers. Switch between them with the tabs.

Fount: findingsFount: matrixFount AI: setupFount AI: matrix

1234

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1A real task, not a tour. Testers got a real job to do with the data: find two or three stories in it. So we saw how they’d actually use it.
2. 2Colour and clarity landed. People loved the colours, and called the dashboard clear, simple and tidy, with fewer pages.
3. 3A finding we didn’t expect. Most people didn’t notice the benchmarks at all.
4. 4A decision, written down. We kept benchmarks as they were for now, because showing them over time would open a much bigger problem.

1234

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1Every tester, every assumption. Each row tests one assumption about a feature, with a result for each person.
2. 2The assumption, in plain words. Each one is written as what the user should get, like a quick grasp of the overall experience and how it changed over time.
3. 3Confirmed by all five. The overview passed with every tester.
4. 4Questions to design for next. Even when it passed, notes like “at which point does it flip from green to red?” showed what to make clearer.

1234

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1Tested on a working prototype. Testers clicked through a Figma prototype of the AI features, in its second iteration.
2. 2Customers, not colleagues. This round was with customers and external users, not our own team.
3. 3Names you’d know. Testers came from T-Mobile, UBS, UHG and Discover, among others.
4. 4Starting with the basics. First check: can people find the new buttons? All five got there. One noticed the change from last year first.

1234

⤢ Full size

Hover a decision to see it in the designTap a decision to see it in the design

1. 1Everyone got it. All five understood what the Action recommendations button and the highlighted text were for.
2. 2Useful, with a wish list. Most found highlighting useful. Others asked to see insights inside themes, and the single moments doing badly.
3. 3Faster analysis, mostly. Does it speed up reading free-text comments? Most said yes, and added ideas like exports and filters.
4. 4Top five first. Michelle at UBS asked for the top five pain points and opportunities first, then all the comments.

“It’s really great work. It’s beautiful too. It’s very clean… You’re not getting lost on the page.”

Trish Delude, Customer Success Manager

Feedback that finally led somewhere

Most employees took part every week, and feedback turned into action. These were shared results. My part was designing a loop people trusted enough to use.

Use stayed high: 91% of employees used Fount at least once a month, and 74% of managers checked their dashboard every week. We sold Fount AI on top of Fount, as an add-on for every enterprise customer, priced by contract.

What people said

Customer success worked closest with our clients, so their view mattered most.

“I’m so excited with what you guys are building… I think the foundations are really, really robust.”

Emilie Bastrup, Head of Customer Success

What this project shows about how I work

Where the work goes next

From the roadmap I proposed, based on ongoing feedback and usage data:

• The features that waited in the first vote: an action tracker, a feedback portal and performance management.
• Industry modules for industries with high friction, such as manufacturing.
• Integrations with performance management and learning platforms.
• Enterprise single sign-on for complex security requirements.

The goal behind all of it: nobody should ever wonder whether speaking up was worth it.