Marketing KPIs: which ones to track and which to ignore

A KPI is not a nice-looking metric. It is the metric whose movement forces you to do something different on Monday. That is the only test that matters. If a number goes up, goes down, and your week is identical either way, what you have is a reporting metric, not a key performance indicator.

The three-question test

Before a metric earns a place on the dashboard:

1. If this gets 20% worse, what do I do? No concrete answer means it is not a KPI.
2. Who is responsible for moving it? An indicator without an owner does not move.
3. How often can I read it with confidence? A KPI that needs a month of data cannot support weekly decisions.

Most agency dashboards fail the first question with half their rows.

Metric, KPI and target are three different things

A KPI without a target is a thermometer with no reference temperature: informative, but silent on whether to act. Write the target number before you see the result, not after.

The minimum set per business type

Lead generation. Cost per lead, lead volume, lead-to-customer rate, cost per sale. The first two live in the platform; the last two live in the CRM. A dashboard with only the first two measures marketing without measuring the business.

E-commerce. ROAS, revenue, average order value, site conversion rate. ROAS alone, without order value and conversion rate, hides whether growth came from selling more or from selling pricier.

Long-cycle services. Qualified opportunities, cost per opportunity, close rate, time to close. Cost per lead misleads here constantly: cheap leads that never qualify quietly destroy the quarter.

Subscription. CAC, cohort retention, recurring revenue, and months to recover CAC. Without that last one, a "healthy" CAC can still break cash flow.

The KPIs that almost always don't belong

• Impressions and reach as headline numbers. They grow by spending more. Useful as context, useless as a goal.
• Follower count. Rarely connects to revenue, and when it does it lags by weeks.
• Total engagement. Rises with controversial content, not with sales.
• SEO vanity metrics inside a paid advertising dashboard: different channels with different cycles, mixed into one confusing view.

None of these is false. The problem is opportunity cost: every row occupied by a metric that demands no decision is a row not showing one that does.

Translating a business KPI into an operating one

The owner wants revenue. Whoever runs campaigns cannot move revenue directly — they move bids, creatives and budget. Between the two there is a chain, and the dashboard works when each level sees its own link.

When someone asks for "one dashboard for everyone", they are really asking for four. Showing the director CTR by creative, or the operator only monthly margin, wastes both people's time.

Leading and lagging, and why you need both

Revenue is a lagging indicator: by the time it moves, the decision that caused it is weeks old. Cost per result and CTR are leading indicators: they move first and give you time to act. A dashboard made only of lagging indicators reports history accurately and changes nothing.

The practical rule is one lagging KPI per level, surrounded by two or three leading ones that plausibly drive it. If you cannot explain how a leading indicator connects to the lagging one above it, you are tracking two unrelated things and calling it a funnel.

The KPI almost nobody tracks and should

Time between a lead arriving and the first contact attempt. It is not a marketing metric and it shapes the return on advertising more than most campaign tweaks. A lead contacted in five minutes and one contacted the next day close at different rates, with identical cost per lead. If your operation generates leads and someone follows up, that number belongs on the dashboard even though no ad platform produces it.

The same goes for the share of leads rejected as unqualified. If it rises, your cost per lead can be falling while your cost per customer climbs — exactly the scenario where the marketing report looks great and the business does not.

How to present them so they get used

The most common failure is not picking the wrong KPIs: it is presenting them without context. A number on its own says nothing. A number with these three things says plenty:

• Comparison. Against the equivalent previous period, and against target.
• Trend. The last several weeks, not today's snapshot.
• Owner and action. Who watches it, and what was done this week.

Five well-presented KPIs get read in two minutes and produce decisions. Twenty-five metrics with no comparison get read in zero minutes, because nobody opens the file.

The trap of internet benchmarks

Searching for "a good cost per lead" returns figures from studies run in other countries, other industries and other years. Using them as a target is an elegant way of inventing a number. The benchmark that works is your own history: same indicator, same account, comparable previous period. By month three you have a real baseline; before that, your target is a hypothesis, and the report should say so.

How often to review them

Daily: spend and volume, only to catch something broken. Weekly: operating KPIs, cost per result and trend. Monthly: business KPIs, CAC, retention, margin. Quarterly: whether the chosen KPIs are still the right ones, because the business moves and dashboards tend to stay in the previous phase.

The practical next step is turning this into a report someone actually reads: see client reporting for small agencies and, for the wider framework, how to measure digital marketing results. If you want the channel numbers in one place first, the pricing page shows what each plan covers.