The chain from goal to metric
Objectives are not invented; they are derived. The business has a goal — revenue, profit, market shareThe business's sales as a fraction of all sales in a defined market, by value or by volume in one unit; a ratio of two estimates, so it is reported as a range., customers, retention. Marketing's objectives are the changes that would move that goal and that marketing can influence. Each objective is proven by one KPI, and each KPI has a target. Written as a chain:
- Business goal
- Marketing objective
- Primary KPI
- Target and date
- Diagnostic metrics
A goal is the business outcome: increase revenue from the online store next year. An objective is a marketing-controllable change that contributes: increase the number of first-time customers from paid and organic search. The KPI is the measure: new customers attributed to search. The target is the number and date: a stated count by year end, at or below a stated cost. The diagnostic metrics are the ones that explain movement in the KPI: search impressions, click-through rate, conversion rate, cost per click. For a free-trial software product the same chain might run from new paying subscribers as the KPI to the trial-to-paid conversion rateShare of free trials that became paying subscriptions. as one of its diagnostics.
Writing the chain out exposes weak objectives immediately. "Increase brand awareness" has no KPI until you say among whom and measured how; "grow social followers" has no goal unless followers demonstrably lead to something the business wants.
Choosing objectives
Three tests separate a useful objective from a wish:
- It would move the goal. If the objective were achieved and nothing else changed, would the business outcome improve? Follower counts often fail this test; qualified leads pass it.
- Marketing can influence it. Revenue depends on pricing, stock and sales; marketing can own the leads, the traffic and the conversion rate, and should be judged on those.
- It can be measured within the period. Lifetime value takes years to observe; a proxy (first-purchase margin, repeat rate at ninety days) is a better KPI for a quarterly objective.
The RACE framework (reach, act, convert, engage) is a useful checklist for coverage: an objective at each stage of the customer lifecycle, so the plan does not optimise conversion while the pool of people to convert dries up. The funnel guide covers the same logic. The marketing objectives library takes each common objective in turn, with its KPI set and the channels that can deliver it.
One primary KPI per objective
Each objective gets exactly one primary KPI: the number that decides whether the objective was met. Two primary KPIs for one objective means neither is decisive, and the team will report whichever looks better.
The primary KPI should be as close to the business outcome as can be measured reliably. For direct-response objectives that is usually conversions and their cost (CPACost of each conversion.) or return (ROASRevenue generated per unit of ad spend.); for acquisition programmes, new customers and CACCost to acquire one new customer.; for awareness objectives, reach among the defined audience (ReachUnique people who saw an ad.) or a brand-tracking measure, not impressions. The core metrics guide explains which metrics are decision metrics at each stage and which are vanity.
Everything else — click-through rate, cost per click, engagement, bounce — goes into the diagnostic layer. It is not less important; it is where the explanation lives when the KPI moves. The distinction keeps reports short and arguments productive.
Setting targets
A target has three ingredients: a baseline, a change and a deadline.
The baseline is where the KPI stands now, measured the same way the target will be. Without it the target is a guess and the result cannot be judged. New programmes with no history use a short test or the economics (what a budget buys at known prices and rates; the budget calculator and CPA calculator do this arithmetic) and replace the estimate with a measured baseline as soon as one exists. For a business with no history at all, the guide to marketing decisions before you have data sets out where each first estimate comes from and when it gives way to a measured figure.
The change should be large enough to matter and plausible given the levers available. A target that assumes conversion rate will double with no change to the site is a hope; one that assumes a modest lift from a specific test programme is a plan.
The deadline matches the planning cycle and the time the KPI takes to respond. Search objectives need quarters; paid-social objectives can be judged monthly; retention objectives need the retention window to elapse.
Targets also carry constraints, which prevent the KPI being hit by means that damage the goal: new customers at or below a cost; leads that sales accepts; conversions excluding discount-driven ones if margin matters. The constraint is as much part of the target as the number.
From goal to KPI tree for a clinic software company
Goal: grow annual recurring revenue. Objective one: more qualified demo requests from clinics of the target size, KPI: sales-accepted demo requests per month, target: a stated count at or below a stated cost per request, by quarter end. Objective two: higher trial-to-paid conversion, KPI: trial-to-paid rate, target: a stated increase on the baseline within two quarters, through onboarding emails and a revised pricing page. Diagnostics under objective one: search impressions and clicks on category terms, LinkedIn reach among practice managers, landing-page conversion rate, cost per click. Diagnostics under objective two: onboarding email open and click rates, pricing-page visits per trial, support tickets per trial. Nobody is judged on a diagnostic; everyone can explain the KPI with them.
The KPI tree
Drawing the chain for every objective produces a tree: the goal at the root, objectives as branches, primary KPIs at the end of each branch, and diagnostic metrics as the leaves beneath each KPI. The tree is the single most useful artefact in a measurement conversation, because it answers "why are we tracking this?" for every number on the dashboard — and reveals numbers that are tracked for no reason. The KPI framework expertise page shows the author's version of the tree and an interactive example; the analytics cluster covers building it in a reporting tool.
Objectives, KPIs and incentives
Whatever is made a KPI will be optimised, including by means nobody intended. Lead volume as a KPI produces cheap, unqualified leads; revenue attributed to a channel as a KPI produces attribution arguments; impressions produce cheap placements. The remedies are the constraints described above, a KPI as close to the real outcome as possible, and a periodic check (an IncrementalityConversions caused by marketing. test, a sales-quality review) that the KPI still tracks the goal. Objectives written with the people who will be judged by them are both more realistic and harder to game. The page on Goodhart's law explains why a measure drifts once it becomes a target and how guardrails and independent ownership of definitions limit the drift.
Reviewing and revising
Objectives are set in the annual planning cycle and reviewed quarterly. A review asks three questions in order: is the KPI moving toward the target; if not, which diagnostics explain it; and does the objective still serve the goal, or has the business changed. Changing a target mid-period because it is being missed defeats the point; changing it because the goal changed is good practice, as long as the change is recorded. Teams that also run OKRs can use the comparison of OKRs and KPIs to keep the two apart: the KPI set stays stable for monitoring, while a quarter's key results commit the team to moving one of them.
Weak and strong objectives, side by side
| Weak | Why it fails | Strong |
|---|---|---|
| Increase brand awareness | No audience, no measure, no target | Raise unprompted awareness among the target segment in the two launch cities to a stated level by year end, measured by the quarterly brand survey |
| Grow social followers | No link to a business outcome | Grow the email list by a stated number of subscribers from the priority segment this quarter, measured in the email platform, as the owned audience for the launch |
| Drive more traffic | Volume without quality or purpose | Increase non-brand organic sessions to comparison pages by a stated share this half, measured in analytics, as the consideration-stage input for the sales objective |
| Get more leads | No quality constraint; invites cheap leads | Deliver a stated number of sales-accepted leads per month at or below a stated cost, measured in the CRM |
| Improve ROAS | Ignores margin and volume | Keep blended return above break-even while growing new customers by a stated share, measured against the order system |
Objective-writing checklist
Common mistakes
- Goals presented as objectives. "Increase revenue" is the business's goal; marketing's objective is the controllable change that contributes to it.
- KPIs without targets. A metric to watch is not an objective.
- Many KPIs. Fifteen numbers on a dashboard with no hierarchy produces reports that describe everything and decide nothing.
- Targets without baselines. The result cannot be judged and the target cannot be defended.
- Vanity KPIs. Impressions, followers and page views as primary KPIs on objectives they do not prove.
- Mid-period target changes. Moving the goalposts hides the learning.
The measurement plan guide takes the KPI tree and decides how each number will actually be captured; the strategy vs tactics comparison explains why objectives belong to strategy and not to individual campaigns.
