Elli Cares app design: a DesignRush awards finalist
The Elli App is a finalist in DesignRush’s June 2026 Best App Design Awards, Health & Wellness. What the jury noticed, and what health apps can borrow.
5 Nov 2026
Workflow automation UX design: build the log first
Workflow automation UX design people trust: why users switch automations off, and how status, previews, run history and undo keep them on.
Short answer
Good workflow automation UX design makes the automation visible. People keep an automation switched on when they can see what it will do, what it did and how to undo it. Give every automation a clear status, a preview, a dry run, a readable run history and an undo. Then design the log before the builder.
Here’s an illustrative scene. An operations lead sets up an automation that tags new leads and sends a welcome email. Two weeks later a customer replies to an email they should never have received. Nobody can say which rule sent it, or how many others went out with it. By lunch the automation is off, and it stays off.
The appetite is there. In Zapier’s 2021 survey of 2,000 US knowledge workers at small and medium businesses, 94% said they do repetitive, time-consuming tasks in their role. People want the busywork gone. They just don’t want a machine acting behind their back.
Most automation products pour their love into the builder and leave the history for later. That’s the wrong way round.
What builds trust in AI at work, and how many workers can’t see how it’s governed. Source: Salesforce, June 2024.
Why do people switch off workflow automation?
People switch an automation off after a surprise they can’t explain, so workflow automation UX design has to start with what the user can see. It tends to go the same way every time:
A surprise.
The automation emailed the wrong list, closed the wrong ticket or ran twice.No visibility.
Nobody can find out what happened. There’s a green toggle and nothing behind it.No way back.
Undoing it means a support ticket, a spreadsheet and a lost afternoon.
One surprise is forgivable. A surprise with no explanation and no undo is how the feature that starred in your demo ends up switched off in the account. Usage drops, and the renewal call gets a bit quiet.
Trust follows understanding. In Salesforce’s 2024 survey of nearly 6,000 knowledge workers, 54% said they didn’t know how AI is implemented or governed at their workplace. Those who did know were five times more likely to expect to trust autonomous AI within two years. It’s about AI, but it fits any rule acting on someone’s behalf.
The five screens that keep an automation switched on, with run history designed first. Framework by Dtail Studio.
What does an automation people trust show?
Five things, each answering a question the user will ask sooner or later. Here’s our checklist:
Status.
On, paused or failing, in plain words, right on the list of automations. “Failing since Tuesday, 14 runs skipped” beats a red dot.A preview.
Before it goes live, say what it will do in one sentence: “When a deal moves to Won, send the onboarding email to the contact and create a task for Sam.”A dry run.
Run it against last week’s real data without acting, and show the result: “This would have sent 37 emails. Here they are.”Run history.
Every run, what triggered it, what it changed and for whom, readable by someone who didn’t build it.Undo.
Reverse one run or a batch from the history. If a step can’t be undone, like a sent email, say so before it runs.
None of this is rocket science. It’s visibility of system status, the first of Jakob Nielsen’s usability heuristics, applied to software that works while nobody’s watching. Most of it is words, which is why UX writing matters more here than almost anywhere.
Why design the log before the builder?
Because the log is where trust is won or lost, and it shapes everything else. The builder gets used once, by one person, on a good day. The run history gets read by everyone, usually on a bad day, often by someone who didn’t build the automation and wants to know why a customer is cross.
So we’d design the run history first. Decide how a single run reads in plain words. Then design the builder so everything it makes reads well in that log. If a rule can’t be explained in one sentence in the history, it’s too clever for the builder. (We say this as designers who have, in our time, built some very clever rules.)
It helps sales too. A buyer who sees a clear history in the demo stops asking “what if it goes wrong?” because the answer is already on screen. Products that feel simpler as they grow usually got the boring screens right first.
How much should an automation do on its own?
Less at first, and more once it’s earned it. Let a new automation suggest before it acts, then ask for a quick confirm, then run on its own once the user has watched a few good runs. Think of it as a new hire’s probation period, minus the awkward review.
Keeping people involved pays off. In McKinsey’s 2020 survey of 1,179 respondents, 61% said their companies met their automation targets, and the successful ones were much more likely to use “human in the loop” approaches, where people train the system over time. In the Salesforce research, 63% of workers said more human involvement would build their trust in AI.
For a founder, this is a churn question as much as a design one. Automations that run quietly and correctly are a reason customers stay. Automations switched off after a bad week turn into seats you’re asked to remove.
How we’d approach it
We’d start with the run history, not the builder. For a workflow automation product, that usually means:
Audit the off switches.
Pull the automations customers paused or deleted, and the support tickets around them. Each one is a surprise someone couldn’t explain.Write one run in words.
The plain sentence for a single run: what triggered it, what it did, who it touched and how to undo it.Design the status and run history screens.
Readable by someone who didn’t build the automation, with undo wherever possible.Add the preview and dry run to the builder.
So people see real results before anything goes out.Test with someone who didn’t build it.
Hand a teammate the history after a bad run and time how long they take to explain it.
Bring your list of paused automations, a few grumpy support tickets and your current builder. If you also report on runs in a dashboard, our SaaS dashboard audit checklist is a good first pass. More on how we work.
FAQ
Do we need run history in the MVP?
Yes, even a simple one. It costs less than the support tickets you’ll get without it.
What if an action can’t be undone?
Say so before it runs, and add a confirm step or a dry run for that action. Sent emails and payments are the usual suspects.
Should AI automations work differently?
Same rules, more need for them. If the system decides something, the history should say what it decided and why, in a sentence a customer could repeat to their boss.
How do we know if trust is the problem?
Look at how many automations get paused or deleted, and how soon after they’re created. A high pause rate in the first weeks usually points to a surprise, not a missing feature.
Try this today
Open your product as a customer and pick an automation that ran yesterday. Without asking an engineer, answer three questions: what triggered it, what did it change, and how would you undo it? Time yourself. Any answer that takes longer than a minute is the next screen to design.
If you’d like a second pair of eyes on your automation screens, let’s talk.
About the author: Rori Thompson is a Content Writer at Dtail Studio and writes about product design, UX and the details that make an app easy to use.
Sources: Zapier, The 2021 state of business automation (April 2021), Salesforce, 77% of workers trust an autonomous AI future (June 2024), McKinsey, The imperatives for automation success (August 2020), Nielsen Norman Group, Visibility of system status (Aurora Harley, June 2018)
Digital health UX agency: 7 US studios to consider
Choosing a digital health UX agency in the US: 7 studios with real health work, the criteria we used and questions to ask on the first call.
Wellness brand design agency guide: clinics and consumers
A wellness brand design agency view on selling to clinics and consumers: what changes in tone, proof and packaging, and what must stay the same.
