Brief
Hollybrook Learning is an invented online course provider. It sells self-paced professional courses, bookkeeping, project management and workplace writing among them, one at a time, and a membership that unlocks the whole catalogue for a monthly fee. The founder runs the business with a marketing manager and a part-time developer, and has hired you as a freelance analyst for one month.
The founder's complaint is simple: every month she receives screenshots from three tools that disagree, and she cannot tell whether to spend more on advertising, fix the website or push the membership harder. She wants one page she can read at the start of each month and trust, and a plan that keeps it trustworthy after you leave.
The constraints:
- The developer has a few days a month. The plan must prioritise what he builds first.
- The course pages sit on the main site; checkout and the learning area run on the course platform's subdomain.
- The course platform is the only complete record of purchases, memberships and refunds.
- Reporting runs in Looker Studio on the GA4Google's event-based analytics platform. connector, the ad platform connectors and a monthly export from the course platform in a shared sheet. There is no data warehouse.
- Visitors in Hollybrook's main market see a consent banner, and a share of them decline analytics storage.
- The founder reads monthly; the marketing manager reads weekly.
The business and its data
Hollybrook's last quarter and next quarter's targets
| Measure | Source | Last quarter |
|---|---|---|
| Sessions | Analytics | 84,000 |
| Course purchases | Course platform | 1,260 |
| Course revenue | Course platform | £75,600 |
| Memberships started | Course platform | 310 |
| Course refunds | Course platform | 95 |
purchase events, courses and memberships together | Analytics | 1,690 |
| Sessions referred by the checkout subdomain | Analytics | 2,900 |
| Paid search spend / platform-reported conversions / platform-reported revenue | Ad platform | £9,000 / 640 / £38,000 |
| Paid social spend / platform-reported conversions / platform-reported revenue | Ad platform | £7,500 / 520 / £29,000 |
| Analytics consent rate in the main market | Consent tool | about 68% |
| Distinct event names in the property | Analytics | 23 |
Real purchases and memberships total 1,570 against 1,690 tracked events, a ratio of about 1.08. The event list
includes Enrol_Click, enrol_click, enrolClick and test_event_purchase. Memberships fire the same purchase
tag as courses, with no parameter to tell them apart. Platform-reported return on ad spend is about 4.2 for paid
search and 3.9 for paid social.
Targets the founder has agreed for next quarter: course revenue of £85,000, 420 memberships started, and a course refund rate below 8% (last quarter: 95 ÷ 1,260 = 7.5%).
Task
Produce two deliverables: the measurement plan, written on the measurement plan template, and the dashboard design. Work in this order; the method for each part of the plan is in the measurement plan tutorial.
- Objectives and decisions. Write each objective as an outcome with its period, and beside it the decision it informs and who makes it. Remove any objective that informs no decision. Add a guardrail where an objective could be gamed; refunds are the obvious one.
- Questions and the KPI tree. List the questions each objective raises when it moves, then build one tree per objective with the arithmetic between levels stated, using the KPI tree framework. Every KPIA metric tied directly to an objective. on the dashboard must be a node in a tree.
- Events and parameters. Fill the template's events table: the events each KPI needs, their parameters, their status (keep, rename, new, retire) and which are key events. Decide how memberships are told apart from courses, how the duplicate and test events are retired, and how the checkout subdomain stops appearing as a referral. Order the developer's work by the decisions it unblocks. The columns and change control come from the tracking plan guide.
- Platform map and attribution. For each KPI, name the reporting system and the source of truth, and record why the alternatives differ. Choose the AttributionAssigning conversion credit to touchpoints. model for business reporting and say what platform figures are used for. Write the expected relationship between platform and analytics conversions using the method in the discrepancies guide.
- Owners, cadence and QA. Fill the owners table with named roles, the weekly data quality check and its tolerances, and the QA criteria each new event must pass, drawing on the tracking verification checklist.
- The dashboard page. Design one page that answers the founder's monthly question. Write the question as a sentence at the top, then lay out the zones the dashboard design guide describes: the answer, the trend, the drivers, the breakdown and the detail. For every element give the chart type, the metric and its comparison, the data source, and the decision it supports. Set the default date range, the filters and the definitions shown on the page.
- The analysis behind it. Name the funnel exploration the marketing manager opens when a driver moves, specified as the explorations tutorial and the funnel analysis tutorial describe, and say which dashboard element links to it.
- The monthly report. Show how the dashboard feeds the results and channel tables of the monthly marketing report template, so that the founder's report and the dashboard can never show different numbers.
One row of a strong dashboard layout, for comparison
| Zone | Element | Metric and comparison | Source | Decision it supports |
|---|---|---|---|---|
| Answer | Scorecard | Course revenue, month to date, against the pro-rated target of £85,000 for the quarter | Course platform sheet | Whether the founder approves extra paid spend next month |
The row names one metric, its comparison and its source, and the decision is specific enough that the founder would act differently if the number moved: a month at 90% of the pro-rated target triggers a review of the drivers row before any budget change.
Evaluation criteria
Assess your submission against each row, or ask a peer to. A strong answer meets every row; most first attempts miss two or three.
| Criterion | What a strong answer shows | Common gap |
|---|---|---|
| Objectives tied to decisions | Each objective names the decision, the decider and the cadence; a refund guardrail sits beside revenue | Objectives that are activities, or that inform no decision |
| KPI tree arithmetic | Every node has a formula, and dashboard KPIs trace to the top of a tree | Metrics listed without the relationship between them |
| Event design | Courses and memberships are separated by a parameter or event, duplicates are renamed or retired, and the key events are few | Adding new events while leaving the duplicate names live |
| Data quality | A reconciliation against the course platform with a written tolerance, negative tests for reloads, and a fix for the checkout self-referral | Trusting the tracked purchase count because it looks plausible |
| Source of truth and attribution | The course platform for totals, one analytics model held constant for channel reporting, and platform figures confined to in-platform optimisation | Comparing platform return on ad spend (ROASRevenue generated per unit of ad spend.) across platforms as if it were one measure |
| Consent | The plan states what the analytics figures mean under declined consent and does not present them as complete | Ignoring consent until the numbers fail to match |
| Dashboard hierarchy | One question, the answer top-left with a target, a trend, a few drivers, one actionable breakdown | A grid of every available metric without targets |
| Decision per chart | Every element names the decision it supports, and anything without one is removed | Charts kept because the data was available |
| Definitions and defaults | Scope and source in each label, a default period matching the founder's monthly read, few filters | Definitions held in a separate document |
| Ownership | Named owners and verifiers, a review date, and a report that pulls from the dashboard | A plan that ends when you leave |
Submission checklist
When you have finished, compare your plan with the worked examples in the measurement plan tutorial and return to the intermediate analytics level to mark the level complete.
