Courtney Vigo
All work Collaboration & pushback

Merlin

Turning engineering pushback into a phased plan, and a presentation maker that takes 45 minutes instead of most of a day.

Problem
Customer success teams spent about 6 hours building each customer presentation, and engineering pushed back on a redesign too big to ship in one release.
Solution
I mapped impact against effort with engineering, shipped the fixes users needed most first, and kept the full redesign as the north star for later phases.
Role
Product designer, end-to-end design
Company
LinkedIn
Team
Product, engineering, customer success
Users
~5,000 customer success team members, and the Insights team
  • 6h → 45maverage time to build one customer presentation
  • ~88%less time per presentation
  • ~26hsaved per person each month
  • ~8 weeksof work time back per person each year

Estimated, based on about 5 presentations per person per month.

  1. 01 Understand CSM needs and stakeholder goals
  2. 02 Sketch One screen for everything a CSM needs
  3. 03 Explore Two directions for each hard problem
  4. 04 Prioritize Impact vs. effort, with engineering
  5. 05 Learn Three lessons from early ideas
  6. 06 Prototype Figma Design and Figma Make
  7. 07 Ship P0 first, north star kept
How it came together. Jump to any step.

The problem

Customer success managers (CSMs) build presentations that show business customers how their LinkedIn products are performing.

Merlin has two audiences, and I designed for both. The Insights team builds the variables, and CSMs see those variables as the settings they fill out in step 2.

MerlinOne tool, two audiences
  1. Insights teamBuild the variables that show up in the step 2 settings
  2. Customer success managersFill out the settings in step 2 and build the presentationAbout 5,000 people
Two groups shape every deck: one builds what the settings can do, the other uses them to build the presentation.
Before
  • ~6 hoursto build one deck, on average
  • Frequent errorsin finished presentations
  • PowerPointfor the finishing touches, by hand
  • One sourceof presentation content at a time
Goals
Time to build a deck8h 2h
Tools used7 2
Goal
CSM satisfaction+80%
Goal, on feedback surveys
The original Merlin presentation maker: three collapsed steps for choosing slides, editing settings and setting up notifications, then a Generate Deck button
Before: three collapsed steps and a Generate button. Nobody saw their slides until they downloaded them.

Brainstorming

I started with one screen that held everything a CSM needed, then sketched in pairs with teammates.

Sample layout: customer switcher and deck name across the top, an objective picker, slide thumbnails on the left, a large slide preview in the center, and a right panel with search, modules, decks, dashboards, filters and content recommended by objective
The sample layout everything else started from.
  • ModulesChapters
  • Fixed side panelsExpandable panels
  • Every content type alikeDistinct looks per type
  • No explanationDescriptions for new reps
What the sessions changed.
Problem 1 of 2 · Adding content

How do CSMs find content to add?

Wireframe: an Add story topic menu offering customer objective, sales objective, stakeholder and product, above slide thumbnails, a preview and a content panel
Pick an objective, get recommendations

CSMs choose a story topic, such as a customer objective like hiring, and the content panel recommends decks and slides for it.

Explored, not the final direction
Problem 2 of 2 · Settings

How do settings apply across a deck?

Two wireframes: global settings for the whole presentation, then settings for just one slide after selecting it
Set the deck, then override a slide

Global settings apply to the whole presentation, and selecting a slide opens settings for that slide only. In the final design, deck settings apply to every slide unless a slide has its own.

Became the final design

The pushback

Small in the interface What people see and click Large on the back end What engineering has to build Interface Back end
Interactions that looked small in the interface were large changes on the back end, so engineering couldn't safely ship the whole redesign in one release.

Prioritizing with engineering

Instead of defending the design, I sat down with engineering, learned where the cost was, and we mapped every feature together.

Impact versus effort map Features plotted by how much they help users and how much engineering they need. High impact, low effort went into the first release; high impact, high effort became later phases. The list after the chart names each feature and why it landed there. SHIP FIRST · P0 LATER PHASE · P1 SMALL POLISH LEAVE OUT FOR NOW 1 Clone slides 2 Custom settings per slide 3 Notes that carry into PowerPoint 4 Live data preview 5 AI feature 6 Customer pages and research 7 Dashboard data source 8 Tutorial you can reopen 9 Clearer Next button 10 Resizable side panel 11 Blank slide Engineering effort → User impact → Low High

Shipped first · P0

  1. Clone slidesUsers needed to clone a slide and add their own data to it, to compare things like regional numbers side by side.
  2. Custom settings per slideCloning only worked if each slide could have its own settings. Both answered the same need, so they shipped together.
  3. Notes that carry into PowerPointNotes travel with the deck, so people can see what to do next without having to remember it.
  4. Live data previewShowing real data in step 2 let people see the actual slides before downloading.

Moved to a later phase · P1

  1. AI featureHeld for P1 so the first release could focus only on building presentations.
  2. Customer pages and researchThese phases of the project were also held for P1, for the same focus on presentation creation.
  3. Dashboard data sourceHeld for P1 until we learned more about how people used it.

Small polish

  1. Tutorial you can reopenA short intro eased CSMs from the old Merlin into the new studio, and they could open it again any time.
  2. Clearer Next buttonIn testing, CSMs weren't sure when to move on to step 2. Turning on the Next button once the required steps were done made the next move clear.
  3. Resizable side panelCSMs wanted a bigger settings panel but still needed to see their slides. A panel they can resize covers both.

Left out for now

  1. Blank slideA blank slide to customize in PowerPoint. Left out because few people needed it.
How we agreed what to build first. Each dot is a piece of the redesign, placed by user impact and build effort; positions are approximate. Tap or hover a feature to see why it landed there.

One of her standout strengths is her ability to prioritize effectively while maintaining clarity around the broader platform vision.

Ana Luiza Loures Pinto, insights manager, Merlin

What early ideas taught us

  1. Lesson 1 Show changes as they happen

    CSMs need to see their slides update as they change settings in step 2, so the middle pane had to stay visible throughout the experience.

    The idea: a shopping cart.CSMs would browse a library of content, pick and choose what they wanted, then start customizing. But browsing took over the screen and hid the slides.
  2. Lesson 2 Make the next step obvious

    Without a clear call to action, CSMs didn't know when to move on. Navigation moved into the right panel, and Next turns on once the required steps are done.

    The idea: a stepper you can jump around in.Instead of Back and Next, CSMs could click any step, so going from step 3 to step 1 took one click instead of two. But nothing showed when they were ready to move to step 2.
  3. Lesson 3 Let people make room

    Some settings needed more space than a side panel gives. Instead of a full-screen takeover, the right panel became resizable, so CSMs widen it only when they need to.

    The idea: full-screen settings.Step 2 settings can be long, and the right panel is small. My theory was that CSMs would check their changes in step 3, so settings could take over the whole screen while they filled them in. In practice, they wanted to see the slides update as they went.
Each lesson shaped the final design. Click a thumbnail to see the early idea.

I checked these fixes with a quick wireframe in front of the user group before polishing.

Watch it

Watch a CSM build a deck in the MVP, step by step.

A silent, 85-second walkthrough of the MVP prototype: what the CSM does at each step, and why. Customer names and data are examples.
Read the steps instead
  1. Search for content to add by typing Hiring Value Review. One panel searches slides, chapters, decks and dashboards.
  2. Show the results as modules.
  3. Select slide 4, so new slides land after it.
  4. Scroll the results and add a module.
  5. Add the selected slides after slide 4. Flag icons mark slides that may need updates in PowerPoint.
  6. Drag slide 4 below slide 5 to reorder them.
  7. Hover over slide 5 for quick actions.
  8. Duplicate slide 5 (P0). CSMs clone a slide and add their own data, to compare things like regional numbers.
  9. Add a note to slide 5: I want this one to show NAMER data (P0). Notes carry into PowerPoint, so CSMs see what to do next without having to remember it.
  10. Go to step 2: Settings.
  11. Drag the panel edge to make room for settings. Settings can be long, and the panel widens without hiding the slides.
  12. Fill in the deck settings. They apply to every slide unless a slide has its own.
  13. Select slide 5.
  14. Give slide 5 its own country (P0). Custom settings per slide make cloning useful.
  15. Slide 5 now shows Custom settings applied. Go to step 3.
  16. Check the preview, then generate (P0). CSMs see the actual slides before downloading.
  17. Open the Presentation library.
  18. The deck is generating in the Presentation library.
The handoff: one document as the single source of truth for engineering, plus weekly office hours for design questions.

The outcome

The original presentation maker: three collapsed steps and a Generate Deck button Presentation Studio: reorderable slide thumbnails, a large slide preview with notes, and a panel for adding content BeforeAfter
Drag to compare. Real data now shows in the slides before anything is downloaded.
Before6h a deck
30h
After45m a deck
3h 45m

26h back every month, per person

A month of decks for one CSM, drawn to scale. Each block is one presentation. Estimated at about 5 a month.

Average time per deck went from 6 hours to 45 minutes, about 88% faster and well past the 2-hour goal. Engineering got the time to build the complex features properly in later phases.

  1. P0 · Shipped Presentation creation
    • Clone slides
    • Custom settings per slide
    • Notes that carry into PowerPoint
    • Live data preview
    6h → 45m per deck
  2. After · Next An AI agent in Merlin
    • Customer research
    • Note-taking
    • Content suggestions
  3. North star · Target The full redesign
    • The complete design every phase works toward
Where it stands.

Pushback is information. Once I understood the why, we found a path that protected both the user experience and the engineering timeline.