HEART, the North Star and a dashboard

Why:You’ll choose a few metrics that speak of the project’s health and build from them a dashboard that is pleasant and useful to look at. Afterwards you have a HEART table filled in for your project, a North Star tied to the main action from the brief, and a dashboard of five measures, each with its data source and the decision it supports.

25 minutes

Read by speech synthesis (Qwen3-TTS); the voice is not a person’s.

HEART is Google’s framework of five groups of metrics: happiness, engagement, adoption, retention, task success. With a North Star they make up a dashboard.

From many numbers to a few

After seven lessons you have many possible numbers: visitors, page views, bounces, time, scroll depth, conversions by step, retention, test results. If you put them all out, you get a wall of numbers that nobody looks at. So the last task of this part is: choose the few things that speak of the project’s health and build from them a dashboard that can be glanced at in five minutes to understand what to do.

There are two supports for choosing. The first you already know: the North Star, one metric about the person’s benefit, from the lesson on growth. The second is HEART: five sides by which to check whether all is well.

HEART

Where from. In 2010 Kerry Rodden, Hilary Hutchinson and Xin Fu of Google described how to measure the quality of user experience at large scale. They proposed five categories of metrics, after the first letters of the English words: HEART. The paper is in the sources.

  • Happiness. Is the person satisfied: ratings, surveys (“would you recommend us”), reviews.
  • Engagement. How deeply does the person use it: how much they read, how far they got, whether they stayed.
  • Adoption. Has a new person started using it: did they take a first step.
  • Retention. Do they come back, do they stay in touch (you counted this by cohorts).
  • Task success. Can the person do what they came for: errors, time, the share who finished.

The value of this framework is not in the five words but in the process the authors propose: for each side, first write the goal (what we want for the person), then the signal (what in their behavior shows that the goal was reached) and only then the metric (how we count it). The order matters. If you start from the metric, you will count what is easy to count and return to vanity numbers.

You needn’t fill in all five sides for every project: for a business-card site retention often means “stayed in touch”. But going through all five and consciously skipping something is useful: you’ll see what you hadn’t thought about.

Open the diagram. Choose a type of project: the goals, signals and metrics are shown for it. Switch on “show the vanity substitutes” and you’ll see which numbers people usually take instead of working ones. Below, play with a panel of ten candidates.

HEART: five sides of a project’s health
SideGoalSignalMetric
HappinessThe person leaves feeling they got an answera rating in the email after the consultationthe share who rate 4 or 5 out of 5
EngagementThe person reads instead of skimmingread an article to the endthe share of visits that scroll to 75%
AdoptionA new person takes the first stepopened the prices or downloaded a checklistthe share of new visitors who reached the prices
RetentionThe person stays in touchopens the emails, reads the channelthe share of subscribers who clicked a link in 30 days
Task successBooking goes through without difficultysent the formthe share who started the form and sent it
A panel of five measures

Choose no more than five. Which of them will lead to decisions?

Chosen 0 of 5. Start with what supports decisions.

The rows of the table are the author’s examples, not data. The HEART framework was described by Rodden, Hutchinson and Fu of Google in 2010.

The North Star on the dashboard

Here the North Star is the dashboard’s main measure. The demands on it are those of the lesson on growth: it names the action, the period and the audience; the metric is about benefit, not traffic; it rises before revenue; it can be measured by a counter or a form. An example: “the number of people from the group ‘…’ who booked a consultation, a month”. Beside it stands a quality metric, so the number isn’t fudged (Goodhart’s law).

A dashboard of five measures

A dashboard is the place where the project’s main numbers are gathered. A good one has several properties:

  1. No more than five measures. Attention is limited, and the extra hides the main thing.
  2. Each has a data source. Analytics, a spreadsheet of leads, an email service, a survey. A person should know where the number comes from and how to check it.
  3. Each has a decision. For each measure it is written: “if the number fell, we…”. A measure with no decision is vanity (the last lesson).
  4. A combination. A typical set: the North Star, the conversion of the main page or form, the share of right leads (quality), return or contact (retention), one measure of speed or errors.
  5. A rhythm. Set a day when you look at the dashboard and decide: once a week or once a month. A dashboard nobody looks at isn’t needed.
  6. Honesty. Show beside a number the period and what you compare it with. Don’t hide the bad numbers.

What to build it with. For a small project a spreadsheet is enough, with five numbers entered by hand once a week. Separate dashboard services (Looker Studio, say, or Yandex DataLens in Russia) are needed when there are many numbers or they must update themselves. Don’t start with a tool: first decide which five numbers and why.

Sources of data and privacy. Some numbers come from the counter, some from your forms and spreadsheets (for example, how many leads reached a conversation). Remember what you learned in the earlier lessons: the dashboard holds numbers and shares, not personal data. Names and phones don’t belong on a dashboard.

Check yourself: why does a dashboard have no more than five measures, each with a decision?

Because attention is limited: more than five numbers turn a dashboard into decoration. And a measure that supports no decision is a vanity one: it suggests nothing.

What next

This part finishes the chain “attract, measure, understand”: you can find people (the part on growth), count what they do, test changes and look at the site through their eyes with respect for their privacy. Next in the course come quality and the law: the speed and accessibility of the site, security, personal data, advertising and a checklist before launch.

Practice

Build a dashboard for your project.

  1. HEART: a goal, a signal and a metric for each of the five sides; in the “Task success” row the main action from the brief.
  2. The North Star: one phrase with the action, the period and the audience from the brief.
  3. The dashboard: three to five measures, each with a name, a source and a decision. No vanity ones.

The check looks at form: the fields are filled in, the North Star names the action, the audience and the period, the dashboard has no more than five rows with sources and decisions. Whether your dashboard is good is for you to decide: imagine it is Monday tomorrow and you look at it for the first time this week. Is it clear in five minutes what to do?

Practice

Preparing your page…

Test

Question 1 of 6

Match each side of HEART with the question it answers

Find the match for each item.

Cards

These cards go into your review once you finish the lesson.

  • What is HEART?

    Answer

    A framework of user experience metrics (Rodden, Hutchinson and Fu, Google, 2010): Happiness, Engagement, Adoption, Retention, Task success. You go from the goal to the signal to the metric.

  • What are goals, signals and metrics?

    Answer

    A goal: what we want for the person. A signal: what in their behavior shows the goal was reached. A metric: how we count it. Go in that order.

  • What should a project’s North Star be like?

    Answer

    One metric that names an action, a period and an audience, about the person’s benefit and not traffic. Beside it a quality metric, so that it isn’t fudged.

  • What rules are used to build a dashboard?

    Answer

    No more than five measures, each with a data source and the decision it supports. No vanity metrics. Once a week or a month it is looked at and decisions are made.

  • What do you build a dashboard with?

    Answer

    For a small project a spreadsheet into which five numbers are entered by hand once a week is enough. Separate services (for example Looker Studio, or Yandex DataLens in Russia) are needed when there are many numbers or they must update themselves.

Sources

Next

You now know: the HEART framework, A project dashboard.