What metrics should you track in growth-driven design?

The method's improvement cycle opens with a focus metric. The team chooses a number it wants to improve and ranks the ideas most likely to move it. It builds them, studies what happened and shares the result with marketing, sales and service. The instruction is to choose one number per cycle, so the first job is deciding which numbers are worth choosing from.

Begin with the outcome the business wants from the site, such as new sales opportunities. A visitor action leads toward it, such as a demo request, and certain pages lead toward that action. Each level has its own measure, and a sprint can focus on any of them.

Choosing a focus doesn't mean the other numbers stop mattering. A change that lifts the focus metric can lower something further along, and keeping the later measures in view is what lets the team see that trade. We cover how ideas get ranked against the chosen metric in how to organize a growth-driven design roadmap.

Traffic and engagement come first on a visitor's path. In Google Analytics 4, an engaged session lasts longer than ten seconds, includes a key event, or has at least two page or screen views. Engagement rate is the share of sessions that were engaged, and bounce rate is the share that weren't.

Under that definition, a session that lasts eleven seconds on one page counts as engaged. That tells you the visitor stayed past the threshold. It doesn't tell you whether they understood the offer or would make a good customer. So engagement works as a signal about where to look. If a page that feeds the demo form shows a falling engagement rate after a change, that is a reason to study the page with recordings, interviews or a test.

No engagement rate counts as good for every site. A long article and a pricing page attract different visits, so compare a page with its own earlier period and with pages that do the same job.

A conversion rate is a count of actions divided by a count of chances to take that action. The second number needs as much care as the first. Demo requests can be divided by sessions on the demo page, by users who visited the site, or by visitors who started the form. Each version is legitimate, and each answers a different question. A rate that switches denominators between sprints can move even when visitor behavior hasn't changed.

Write the definition beside the metric: which action, divided by what, over which dates. Then keep it fixed for as long as you compare results.

A form submission is an early recorded step toward the outcome the business cares about. Whether it becomes a sale is recorded in the CRM, so the metrics after the form come from there. HubSpot's lifecycle stages are one common structure. A marketing qualified lead is a contact marketing has judged ready for sales. A sales qualified lead is one sales has accepted as a potential customer. An opportunity is associated with a deal. A customer, in HubSpot's wording, is a contact or company with at least one closed deal.

Those stages can be customized, and a contact can be moved by a person or by an automated rule. So a stage count means as much as the rule behind it. If "sales qualified" means a rep has had a conversation and confirmed a budget, write that down. If the rule changes during the year, the rates before and after it measure different things.

Sales feedback belongs beside these numbers. Lean Labs' article on the three stages recommends a short weekly or biweekly meeting where sales shares the questions, objections and misunderstandings it hears from prospects. Those observations go into the improvement backlog and are scored with everything else. The agency's growth-driven design program lists sales feedback among its inputs, alongside page behavior, forms and CRM lifecycle data. The form count shows how many demo requests came in. The CRM stage and the rep's account of the call show whether those requests came from the people the company wanted to reach.

Suppose a software company's demo page gets 4,000 sessions in a quarter and 120 demo requests, and assume each request comes from a different contact. That is a rate of 3% of sessions. Sales accepts 36 of those contacts as sales qualified, which is 30% of the requests.

The team spends a sprint shortening the form. Suppose the next quarter brings the same 4,000 sessions and 200 requests, so the rate rises to 5%. Sales accepts 40 of them, which is 20%.

The focus metric rose by two-thirds. The number of qualified leads rose by four, and sales reviewed 80 more requests to find them. Whether that trade is worth making depends on what a rep's time costs and what the removed fields were doing. If one of them screened out companies too small to buy, the next sprint could restore it and try a different change. With only the first rate on the dashboard, the team would have recorded a clear gain.

The example also depends on counting the right contacts. The 40 qualified leads come from the 200 people who submitted the form in that quarter, followed for long enough that sales could reach them. A group defined by when it entered and then tracked forward is called a cohort. The common shortcut is to divide this month's closed deals by this month's new leads. The two counts can describe different groups of people, because a deal that closes this month can belong to a lead created months earlier. If lead volume is rising and deals take months to close, the shortcut can understate the close rate, since the deals come from earlier and smaller groups.

Revenue is the outcome most teams want to connect to a page, and attribution reports are the usual tool. HubSpot's attribution reporting has separate reports for contact creation, deal creation and revenue. HubSpot lists the deal-create and revenue reports under Marketing Hub Enterprise. Each report uses a model to allocate credit to the interactions HubSpot recorded for a contact, such as page views, form submissions and logged calls.

Credit follows the model you chose. In a revenue attribution report, it describes which recorded interactions the model associates with closed revenue. Suppose the pricing page receives more revenue credit than any other page. If people who are close to buying check pricing before they sign, the credit could reflect who visits the page as much as what the page does. The report can't show what revenue would have been with a different page, and it knows only about the interactions that were recorded.

So use attribution to see which pages buyers pass through and to choose where a sprint should look. When the question is whether a change caused a difference, compare two versions directly. HubSpot's page A/B test, available on Content Hub Professional and Enterprise, shows each version to half of the page's visitors. The test compares the two versions on the action it measures, and lead quality still has to be checked in the CRM. We cover that kind of test in turning user insights into hypotheses and experiments.

Pick the focus metric from a step that loses a large share of the people who reach it. The team should also have an explanation for that loss, one that rests on something a buyer said or did. A weak number alone doesn't tell you what to change.

A target is a number the team chooses from its own baseline. It isn't a benchmark. If the demo page converted 3% of sessions last quarter, a target of 4% states what the team is aiming for with this change. Writing it down before the sprint gives the review something to check against. Compare like periods as well. A quarter with a trade show and a quarter without one differ for reasons unrelated to the page. On a page with few visitors, a comparison between two periods can stay inconclusive.

Not every number has to rise. A sprint that adds a qualifying question to the form may lower the form rate and raise the share sales accepts, and that can be the intended result. Record both numbers in the sprint review so marketing and sales read the same trade.