The Problem
Who this is for
Supporting Materials - Help Doc
First Contact: Teaching New Users to Log Activity in a CRM
CRM rollouts don’t usually fail because the software is hard. They fail because reps log their first call, don’t set a follow-up, and three weeks later the pipeline is a list of names nobody has spoken to. Managers see stale data and start asking for updates in meetings instead, which is exactly the behavior the CRM was bought to replace.
Most onboarding content I looked at covers the feature. Here’s the Activity panel, here are the field types. That teaches the interface and skips the habit. I wanted the smallest possible task that actually changes what someone does on day two.
These learners are customers, not employees. Nobody assigns this to them. There’s no completion deadline and no manager checking a report. If it isn’t worth their time in the first fifteen seconds, they close the tab and file a support ticket later.
That constraint drove most of the design. The task is one workflow, not a tour. The guidance is dismissible. And the whole thing runs in the product rather than in a course player, because switching contexts to learn a thing you’re trying to do right now is its own kind of friction.
Two modes, one task
Kestrel is a concept CRM I built to design against. Sales teams use it to track contacts and move deals through a pipeline, and this is the onboarding piece for people opening it for the first time.
A software simulation with two modes: watch the workflow once, then do it yourself in the live interface. The task is small on purpose. Log one call, set one follow-up. It's the habit that decides whether a CRM gets adopted or quietly abandoned.
Audience: New CRM users during their first week of onboarding, at a company that just purchased the software (concept scenario)
Responsibilities: Instructional design, interaction design, prototype development, guided-layer logic, seed data design
Tools Used: Claude, Claude Design
Watch mode plays the workflow through with captions explaining each decision. Try mode gives the learner the same task and lets them do it.
I built both because they answer different questions. Watch mode answers “what am I supposed to do here,” which is a question you can satisfy with a video. Try mode answers “can I do it,” which you can’t. Someone who has seen the follow-up checkbox and someone who has clicked it are not in the same place.
The first version only had Try mode. Testing it cold the next morning, I found I was hunting for the Log Activity button myself, on an interface I built. Watch mode exists because the confidence to try comes from having seen it once.
I wrote a help-center article for the same workflow. Not a summary of the simulation, a companion to it.
The two formats answer different needs. Someone learning the product for the first time benefits from being walked through it. Someone who has done this before and forgot which step reveals the date field does not want a guided walkthrough, they want to scan and leave. Making the second person sit through the first person’s experience is a good way to get a support ticket anyway. So the article is built for scanning. Seven numbered steps, bolded UI labels, a troubleshooting section written from the actual failure states in the prototype, including the one the simulation’s wrong path creates.
It also carries something the simulation can’t. There’s a short section on how to write a note another person can use, with the difference between “Discussed pricing” and “Wants pricing for 40 seats.” That’s judgment, not procedure, and nobody absorbs judgment while clicking through a highlighted button. It works in text someone reads at their own pace.
Supporting Materials - Infographic
The third piece is a one-page job aid. The whole workflow on a single sheet, meant to be glanced at rather than read.
This covers a moment the other two formats miss. A rep is about to log a call, mostly remembers how, and needs a two-second check rather than a walkthrough or an article. Opening help for that feels like too much, so they guess, or they skip the follow-up, which is the failure this project is about.
A job aid works there because it's spatial. All seven steps are visible at once, so you aren't searching sequentially for the one you forgot. It also lets me place the follow-up step visually inside the same action rather than tacking it on at the end.
What it leaves out is the reasoning. There's no room, and there shouldn't be. The article explains why a note should lead with the outcome. The job aid shows the field and moves on. Someone reaching for this already decided to do the task.
Three formats, one workflow. The simulation builds the habit, the article answers the question, the job aid catches the hesitation.