Marketing KPIs: A Three-Tier Set That Tells You What to Do Next

Marketing KPIs are the few numbers you review regularly to decide whether to continue, change or stop something. A metric becomes a KPI when a specific person can act on it. Plenty of data is worth recording that is not worth reviewing, and the discipline lies in telling the two apart.

We recommend sorting them into three tiers. Leading KPIs show early signals you can influence quickly. Lagging KPIs show outcomes, such as revenue, after the delay of a sales cycle. Efficiency KPIs divide cost by result and tell you whether the outcome was worth buying. Most weak dashboards have plenty of the first tier, some of the second and nothing in the third.

Be careful with ratios. Cost per lead falls when you accept worse leads, so it should never travel without a quality measure beside it. The same holds for conversion rate, which rises when traffic shrinks to the most committed visitors. Pair every efficiency number with the volume or quality number it can distort, and read them together.

Leading KPIs include qualified enquiries per week, share of enquiries answered within a set time, and demo or quote requests from target pages. Lagging KPIs include pipeline created, closed-won revenue and customer retention. Efficiency KPIs include cost per qualified lead, cost per opportunity and customer acquisition cost.

Definitions decide whether the numbers are comparable from month to month. Write down what counts as a lead, who qualified it and where the data lives. HubSpot, for example, ships default lifecycle stages (Subscriber, Lead, Marketing Qualified Lead, Sales Qualified Lead, Opportunity, Customer and others), and its help documentation notes you can customize them. Whatever stages you use, a KPI that counts "leads" without saying which stage is one that two people will report differently.

At the awareness stage, track reach among the audience you actually want, measured as search impressions for non-brand queries or visits from target regions, not total traffic. At consideration, track engaged visits to comparison and pricing pages, content downloads and newsletter signups tied to named offers. At decision, track qualified enquiries, quote requests and booked calls, and the speed of the first reply.

After the sale, track repeat purchase or renewal, referrals and review volume. Many small teams stop at the lead, which leaves the cheapest growth unmeasured. Keep two or three KPIs per stage at most; more than that and the review meeting turns into a tour of charts.

Services businesses can map these to sources they already have. Search Console shows non-brand impressions and clicks for free, GA4 shows engaged sessions and key events, and the CRM shows opportunities and revenue. The gap is usually the join between them: unless the enquiry records its source, you can show traffic and revenue but not the link between them. Fix that field first, because every efficiency KPI depends on it.

A vanity metric is one that can climb while the business does not. Followers, impressions, total sessions and email list size all fit when reported without a link to enquiries or revenue. They are not useless: they are context. They become misleading when they stand in for results, usually because they are easy to grow and easy to display.

The test is simple. Ask what you would do differently if the number moved 20 percent in either direction. If the honest answer is nothing, demote it from the KPI list to a diagnostic. Keep it on a secondary tab for the day a leading number drops and you need to find out why.

A worked example, with illustrative numbers: a page grows from 2,000 to 3,000 monthly visits after a viral post, while qualified enquiries stay at 10. Sessions rose 50 percent and nothing the business cares about moved. Now suppose visits fall to 1,500 after you remove thin pages, and enquiries rise to 14. The vanity view says things got worse; the useful view says they improved. Judge traffic by the enquiries it produces and by whether it comes from people you want to serve.

Social platforms and ad networks report engagement counts by default, which is why these numbers dominate so many reports. Fine to glance at, but a reaction count is not a lead. Where a channel truly cannot be tied to enquiries, such as brand awareness work, say so and judge it on a stated, modest goal for a defined period instead of passing it off as a revenue driver.

Match the cadence to how fast the number can change and how fast you can respond. Leading KPIs belong in a short weekly look, because a drop in qualified enquiries or a slow reply time can be fixed within days. Lagging and efficiency KPIs suit a monthly review, since a sales cycle of several weeks makes weekly revenue figures mostly noise.

Once a quarter, review the KPI set itself. Remove anything nobody acted on, fix definitions that drifted, and check that every KPI still has an owner and a source system. Put the targets in writing before the period starts, and compare with the previous period rather than with an industry figure you cannot verify. This approach is overkill for a team of one running a single channel; there, three numbers on a weekly note are enough.

Format matters as much as frequency. A one-page report with the KPI, this period, the last period, the target and one sentence of commentary is read; a twelve-page deck is skimmed. The commentary should say what you will do about the number, because a figure without a decision attached is just a record of what happened.

Work backwards from the outcome. Write the business goal for the year in one line, such as new customers or booked jobs. Then list the steps a buyer takes to get there, and choose one leading, one lagging and one efficiency measure that covers the steps you can influence. Ten KPIs is a ceiling for a small team; six is easier to keep.

For each KPI, record five facts: the definition, the system the number comes from, who owns it, the target and the review date. Targets should come from your own history. If you have none, run the measure for a quarter first and treat it as a baseline, because a target borrowed from an unrelated company is a guess with a number attached.

Test each KPI against a bad month. If the number would stay green while sales fell, it is measuring the wrong thing. If it turns red for reasons no one can control, such as a holiday week, add a note rather than a new metric. Over time, the set should shrink more often than it grows.

Share the draft with the people who will be held to it. A KPI imposed from above gets gamed or ignored, while one the owner helped define gets used. Ask the owner what they would change about the definition, what could distort the number and what they would need to see to act on it. Those answers often improve the metric more than any template.