All case studies

Better business plans at EngineRoom

I turned a 32-page planning feature only our managing director could navigate into a guided plan advisors used with almost every customer.

The redesigned Plan performance page for a sample business, showing revenue, transaction yield and new customer cards against plan, a revenue versus leads chart, and a leads card broken down by channel.

Snapshot

Drove active plan adoption from 6% to 83% of the customer base by September 2021.

Role
Senior Product Designer and Project Lead
Company
EngineRoom, a roughly 25-person digital growth advisory with an in-house platform for its SME customers
Team
1 designer (me), 1 tech lead, 1 software engineer and 1 data scientist. The managing director and 3 advisors were core stakeholders.
Timeline
March 2020 to July 2021, in two phases (8 months, then 3 months)

Where we started

The original Plan feature, a dense page with a long sub-navigation, seven tabs, a stacked bar chart, a twelve-column table, data boxes, a donut chart and an area chart. Annotations count 32 pages, 69 charts, 42 tables and 58 data boxes.
Plan held real depth, but it was spread across 32 pages, 69 charts, 42 tables and 58 data boxes, and only the person who designed it could find the story.
The Plan product team smiling around a high table with their laptops.
Product team

Context and the real problem

Plan was understood by our managing director. It needed to be used by advisors.

The stakes

  • Plan was meant to be EngineRoom’s proof of value: the place that connected a customer’s revenue growth to the marketing work we delivered for them.
  • In practice, only 3 customers had an active plan, two of which were built by hand in Excel.
  • That undercut something EngineRoom actively sold.

The brief

  • “Make plans easy.”
  • Targets: 100% of customers with a completed plan, a 50% lift in weekly active users in Plans, and 80% advisor satisfaction.

How I reframed it

  • Our managing director’s judgement lived only in his head: what a good plan looks like, and how revenue breaks down into customers, leads and spend. The clunky interface grew out of that.
  • I won’t design for data I don’t understand. So I spent time with him unpacking how he read a business: which figures he looked at, in what order, and how he got from raw numbers to a decision.
  • That shifted the focus to making his reasoning visible on screen, so an advisor could follow it, trust it and explain it to a customer.

Key decisions

Decision 1

Show the work, not just the answer

Options
Calculate targets behind the scenes and show only the result, or expose every step.
What I chose
Every derived figure shows its formula inline. Revenue × % of customers from EngineRoom ÷ average transaction yield gives new customers needed; ÷ conversion rate gives leads needed.
Why
That chain was our managing director’s mental model, pulled out of him over many sessions. Advisors defaulted to other analytics or had to get time with him to create something for a customer.
Trade-off
More on screen in an already dense feature. I kept the formulas in dismissible tips (“Hide these tips”) so experienced advisors could clear them.
The customers step of the plan form. Beside the percentage input, a tip shows the working: new customer revenue times the share from EngineRoom, divided by average transaction yield, gives 1,126 new customers. Divided by the lead to sale conversion rate, that gives 4,853 leads required.
The tip sits beside the input it explains and reads in the order our managing director reasoned, from revenue to customers to leads.

Decision 2

One flowing page instead of a wizard

Options
A multi-step wizard, a step-by-step form, or a single page.
What I chose
One page that shows every input up front and reveals sections progressively, with a progress ring and “save and continue later”.
Why
A wizard makes the user do the navigating. A single page moves that weight onto the system, so people see the whole job and feel guided rather than processed.
Trade-off
A long page can look daunting. I offset it with reassuring copy (“Don’t worry about getting all the details”), visible progress and progressive reveal.
The targets step as one flowing page. A rail on the left lists Targets, Revenue and Average transaction yield. Revenue history, future targets and monthly targets with year-on-year growth fill in, ending in a calculation card of new customers required, while a progress ring rises and the plan saves automatically.
The section rail keeps the whole job in view. Progress and autosave make a long page feel doable rather than daunting.

Decision 3

Don’t force everything through Xero

Constraint
Xero was meant to be the data backbone. But businesses set up their accounts in bespoke ways, API data arrived inconsistently, and some setups had no clean mapping.
Options
Keep pushing for a perfect Xero pipeline, or build other ways to create a plan.
What I chose
Multiple entry routes: Xero, results input, CSV import and manual entry.
Trade-off
Manual data is less reliable. I compensated at the point of entry with year-on-year % growth beside every monthly figure, and a warning when monthly targets didn’t add up to the total.
The targets step offering four data sources, Xero or MYOB, results input, CSV import and manual entry, above monthly revenue targets with year-on-year growth beside each month. A summary card warns that the monthly targets add up to $40,000 less than the total target.
Growth beside every month makes an unrealistic number visible as it’s typed, and the summary catches totals that don’t add up.

Decision 4

The pivot, from a customer’s tool to an advisor’s tool

Hypothesis disproven
Customers would fill in their own plans. Phase 1 completion stalled at 25% through the Victorian lockdowns, when owners had no headspace for planning.
What I did
On the beta calls I shadowed, advisors ran Plan live with customers. Once it was usable, it gave them confidence and became something they wanted to show.
What I chose
In phase 2, less dependence on customer input, a choice of plan type (full business plan or marketing projections only), and advisors building plans with or for customers.
Trade-off
Customers had less hands-on ownership of their plan. In return, plans got done, and planning became part of the advisory relationship rather than homework.
Result
This wasn’t one person’s call. It emerged from what the beta showed us, and I designed phase 2 around it.

Influence

  • Alignment through our managing director’s own logic. I kept him on board by showing how his reasoning came through as a tool, freeing him up to do other work. When the targets flow broke a business down the way he would, he was on board.
  • Testing in the field, not in a room. I trained advisors first, then shadowed their live calls with 9 beta customers across revenue plans with Xero or MYOB, ecommerce, customer and leads plans, and different levels of technical comfort. What I saw on those calls shaped phase 2.
  • Stretch goals with engineering. I worked closely with our software engineer and data scientist and kept setting the bar just beyond what felt possible. When Xero data couldn’t support it, we agreed on alternative entry routes rather than stalling.
  • What outlasted the project. I documented the visual rules from Plan as EngineRoom’s Data Display Strategy: colour for data type, pattern for planned versus actual, card hierarchy for summary versus depth. Plan also became the foundation for Annual Strategy reporting.

Solution

The full phase 2 Plan performance page for a sample business. Business performance cards for revenue, transaction yield and new customers sit above a revenue versus leads chart. Digital marketing cards for leads, conversion rate and sessions sit above a leads versus sessions chart. Program cards for Google Ads, SEO and Facebook Ads sit above a leads versus investment chart. Every figure shows its plan and percentage to plan.
The phase 2 Plan performance page. Every card leads with one figure and its plan, then breaks into the parts that build it.
  • Data cards with drill-down. Each card leads with one figure, its plan and % to plan, such as $2,100,040 against $2,026,000. Underneath, it breaks into the parts that build it, so an advisor can see why a customer is off track.
  • Solid for actual, striped for planned. Pattern carries the meaning as well as colour, so planned versus actual reads at a glance and doesn’t rely on colour alone.
  • “Choosy charts”. Two pickers let you compare any two measures over time, such as total revenue against leads, which is how EngineRoom showed its marketing was hitting its own targets.
  • Key dates on the chart. Moments like end of financial year sit on the timeline, so a dip or spike has context.
A chart comparing total revenue against leads for each month of the year. Solid bars show actual revenue, striped bars show planned revenue, a solid line shows leads and a dotted line shows planned leads. End of financial year is marked on the timeline, and a tooltip compares December’s results with plan.
Pattern as well as colour separates plan from actual, so the chart doesn’t rely on colour alone. The pickers let an advisor put any two measures side by side.
  • Plan type first. Choosing between a full business plan and projections only means the form asks just for what that plan needs.
  • Every calculation shown. Each step ends in a summary card with its working, finishing at leads required per month.
The phase 2 targets flow. It starts by choosing a plan type, business plan or projections only, then the target type. Completed steps collapse to their answers, and the flow ends in a leads required card showing the working from revenue to customers to leads needed each month.
Plan type comes first, so the form only asks for what that plan needs. It ends in the number advisors actually work to: leads per month.
  • Projections to sell from. Advisors model the leads, cost and revenue of each program, with their own commentary, on one page they can use in a conversation about investing more.
The EngineRoom projections page. Revenue, new customer and leads projections each sit at 100% of target above an investment summary. Program cards for Google Ads and SEO show projected leads, cost and sessions, with the advisor’s own comments explaining each projection.
Each program carries the advisor’s reasoning in their own words, so an advisor can use it to make the case for investing more.

Outcomes

Planning went from something 3 customers did to something nearly every customer had.

  • 39 of 47Customers with a plan for the financial year (83%) by September 2021, up from 3 of about 50
  • 70%Customers with projections entered, by September 2021
  • 8 of 9Beta customers who completed a plan
  • 100%Advisor satisfaction, against an 80% target
A team chat message from the advisory team on Thursday 9 September announcing that 83% of all 47 customers now have plans for this financial year and 70% have projections, with a row of reactions underneath.
The advisory team’s own announcement, a few months after phase 2 shipped.

Reflection

  • Making existing depth usable is its own skill. Plan wasn’t a blank canvas. My job was to make our managing director’s expertise legible to everyone else, and the most effective move was literally showing the working.
  • Test who the user really is. We designed phase 1 for customers filling in plans themselves. Watching advisors on live calls told us the real user was the advisor-and-customer pair. I’d get into those calls months earlier next time.
  • Know the difference between “this didn’t work” and “this hasn’t had its moment”. Phase 1’s 25% was partly timing. The second run worked because we also changed who did the planning.

Product screens are rebuilt in code with sample data for a fictional business. The team message is recreated, with names changed.

Next case study