Marketing KPIs: How to Build a KPI Tree and Metric Dictionary
Published:
Last Updated:
Category: Marketing Budget & KPI
Authors: Shusaku Yosa
Marketing KPIs should explain progress toward a business objective and indicate where a team can act. Design them in this order: goal, intermediate outcome, formula, data source, owner and decision. A dashboard containing every available metric rarely provides that clarity.
In Japanese management terminology, the final outcome is often called a KGI, or key goal indicator. KPIs help monitor its drivers. Improvement in a driver does not guarantee business improvement, so include quality and cost guardrails.
Build one consistent KPI tree
Consider a fictional B2B company targeting JPY100 million in annual new-customer bookings, with JPY2 million per won deal, a 20% win rate and a 25% MQL-to-opportunity rate.
Level | Required value | Relationship |
|---|---|---|
New-customer bookings | JPY100m | 50 wins × JPY2m |
Wins | 50 | 250 opportunities × 20% |
Opportunities | 250 | 1,000 MQLs × 25% |
MQLs | 1,000 | 2,000 leads × 50% |
Leads | 2,000 | Plan channel volume and fit |
The 25% MQL-to-opportunity rate can be decomposed into 50% MQL-to-SQL and 50% SQL-to-opportunity. See the B2B KPI guide for the handoff detail.
At an average opportunity amount of JPY2 million, 250 opportunities represent JPY500 million of unweighted pipeline. This is not the JPY100 million bookings target. Pipeline includes unwon deals with different probabilities and timing, so it should not be presented as confirmed revenue.
Create a metric dictionary
Metric | Formula and scope | Source / owner | Decision supported |
|---|---|---|---|
Qualified MQLs | New contacts meeting fit and interest rules, deduplicated | CRM / marketing | Is the relevant prospect pool sufficient? |
Opportunity rate | Opportunities from an MQL cohort / original MQLs | CRM / sales operations | Where does handoff or qualification stall? |
New-customer bookings | Value meeting the agreed won condition | Deal ledger / sales owner | Progress toward the business goal |
Cost per qualified MQL | Defined cost / qualified MQLs | Cost ledger and CRM / operations | Acquisition efficiency and allocation |
Customer gross profit | Revenue minus corresponding delivery cost | Finance records / finance | Whether revenue creates contribution |
Add period, currency, tax treatment, refresh frequency, exclusions, reporting delay, missing-data rules and definition history. Labels such as “users,” “conversions” or “sales” are insufficient without these details.
Choose campaign metrics by purpose
Purpose | Intermediate measure | Outcome check | Avoid |
|---|---|---|---|
Capture demand with ads | Relevant clicks and landing-page arrival | Qualified conversions, CPA and wins | Increasing spend on CTR alone |
Support evaluation with content | Relevant queries and content use | Consultations and sales usefulness | Judging only by pageviews |
Advance consideration with email | Delivery, clicks and replies | A useful next action | Treating opens as confirmed human readership |
Build understanding through PR | Coverage quality and relevant reach | Changes in understanding or behavior | Treating advertising-value equivalents as profit |
Support existing customers | Adoption and problem resolution | Retention, repeat purchases and contribution | Ignoring the cost of support |
AMEC's Barcelona Principles 4.0 rejects advertising-value equivalents and emphasizes outcomes and impact. Keep coverage outputs distinct from changes in awareness, behavior and business results.
Set targets and review intervals
Use comparable historical performance, the business requirement and operating capacity. An external average is useful only when its year, market, channel, population and denominator are relevant. An unexplained benchmark should not become a performance target.
Daily checks may detect broken tracking or delivery problems; weekly reviews can cover actions and follow-up; monthly reviews can examine cost and mature outcomes. Do not assess a months-long sales process through daily fluctuations.
Record the decision when a metric misses target
Capture the observation and denominator, data-quality checks, decomposition by channel or cohort, alternative explanations, proposed action, owner, deadline and reevaluation rule. Check for missing data and immature cohorts before calling a change a performance decline.
Remove metrics that do not support a decision. If a definition changes, document the break in comparability and update the baseline rather than making the previous series appear continuous.
Related practical guides
Connect execution, costs and reviews
Xtrategy supports project management, monthly budgets and actuals, and customer and deal records. Use a 30-day free trial to evaluate the workflow against your team’s needs.
References
Official sources checked September 29, 2026. Product pricing and availability may vary by market, plan and contract.
