Choosing a metric for a news project is less about finding the biggest number and more about deciding what success should mean. A useful metric connects your editorial purpose to observable evidence, while leaving room for context, judgment, and public value.
Start with the project’s real purpose
Before choosing a metric, write down what the project is supposed to accomplish. “Get more traffic” is an outcome, but it is rarely a sufficient purpose for journalism. A news project might aim to help residents understand a local issue, reach people affected by a policy, encourage informed participation, expose misconduct, or provide reliable information during a crisis.
Use a one-sentence purpose statement:
This project exists to help [specific audience] do or understand [specific change] through [type of journalism].
For example:
- This project exists to help renters understand how a new housing rule affects their rights.
- This project exists to help residents evaluate candidates before a local election.
- This project exists to explain changing flood risks to people living in vulnerable areas.
The purpose statement gives you a standard against which to judge candidate metrics. If your goal is comprehension, pageviews alone are weak evidence. If your goal is reaching an overlooked community, total audience size may hide the result you care about.
Separate outputs, outcomes, and impact
News teams often mix different kinds of measurement. Separating them makes metric selection much easier.
- Outputs are what the team publishes or produces: articles, investigations, newsletters, videos, explainers, translations, or events.
- Audience outcomes are what people do with the work: read, watch, return, subscribe, share, save, discuss, or use the information.
- Public impact is the broader change the work contributes to: improved understanding, better access to services, accountability, community action, or a more informed public debate.
Outputs are usually the easiest to count, but they do not prove that the work mattered. Impact is often the most meaningful category, but it may be difficult to measure and may develop slowly. A strong measurement plan connects all three.
For instance, a local investigation might track the number of stories published, the percentage of readers who reach the key evidence section, and documented follow-up by public officials. None of these measurements proves causation by itself, but together they create a more useful picture.
Define the audience precisely
A metric becomes more meaningful when it is tied to a defined audience. “Everyone online” is too broad for most projects. Identify the people who most need the journalism, not only the people most likely to click it.
Consider:
- Where the audience lives or works
- Which language or languages they use
- Their likely level of background knowledge
- Whether they have reliable internet or mobile access
- What decision, problem, or question the project addresses
- Which groups are likely to be missed by ordinary analytics
A project about public transportation, for example, may have a smaller total audience than a general-interest article but greater value for people who depend on a particular route. Segmenting your data by location, device, language, referral source, or membership can reveal whether the intended audience is actually being served.
Be careful with sensitive characteristics. Collect only information you genuinely need, explain how it will be used, and avoid treating personal data as a free source of insight.
Choose a primary metric and a small set of supporting metrics
Most projects need one primary metric that represents the main objective. They may also need two or three supporting metrics that explain quality, reach, or unintended effects.
A practical structure is:
- Primary metric: the clearest available indicator of the project’s main goal.
- Quality metric: whether people engaged meaningfully with the journalism.
- Equity or reach metric: whether the intended or underserved audience was reached.
- Safeguard metric: whether optimization is creating harmful or misleading results.
Suppose your project is a service guide helping residents apply for energy assistance. A reasonable primary metric might be completed guide actions, such as clicking to the official application instructions. Supporting metrics could include time spent with the eligibility explanation, visits from the target region, and user reports of confusing or incorrect information. A safeguard metric might monitor complaint rates or high exits from a critical instructions page.
Avoid selecting ten “primary” metrics. When everything is equally important, the team has no clear basis for making trade-offs.
Match the metric to the reader’s journey
A reader usually moves through several stages, and different metrics are useful at each stage:
| Stage | Useful question | Possible metric |
|---|---|---|
| Discovery | Did the intended audience encounter the project? | Qualified reach or referral visits |
| Understanding | Did people engage with the essential information? | Completion rate, scroll depth, or knowledge prompt |
| Use | Did the information help with a relevant task? | Guide clicks, saves, downloads, or return visits |
| Trust | Did the audience find the work credible and useful? | Direct feedback, corrections, or repeat use |
| Public response | Did the reporting contribute to accountability or action? | Verified follow-up, official response, or documented service use |
These are not interchangeable. A large reach number can coexist with poor understanding. A low article completion rate might be acceptable if the key information appears near the beginning and readers quickly take the intended action. Interpret the metric within the journey it is meant to represent.
Prefer meaningful engagement over raw attention
Attention metrics can be useful, but they need careful interpretation. Pageviews, impressions, and total watch time measure exposure, not necessarily comprehension or value. They can also encourage headlines and formats that maximize curiosity without serving readers.
More useful measures may include:
- The percentage of users who reach the essential section
- The proportion of readers who return to use a service guide
- Newsletter sign-ups from the relevant geographic area
- Clicks to primary documents or official resources
- Saves, downloads, or shares that indicate future use
- Direct responses describing what readers learned or needed
- Completion of a clearly defined action that supports the project’s purpose
Do not assume that a long session is always better. A reader may spend ten minutes because the article is confusing, because a page is left open, or because the information is difficult to find. Combine behavioral data with editorial review and, where practical, direct audience feedback.
Test whether the metric is actually measurable
A metric should be specific enough that two people would calculate it in the same way. Define the following before launching:
- Name: What exactly is being counted?
- Numerator: Which actions or people are included?
- Denominator: What is the metric compared with?
- Time period: When does measurement begin and end?
- Audience: Which users or communities are included?
- Data source: Where will the data come from?
- Owner: Who checks and interprets it?
- Decision rule: What action might the result trigger?
For example, “engagement” is too vague. “The percentage of visitors from the project’s target counties who reach the section titled ‘What to do next’ within 30 days” is much more precise.
Check for technical limitations before committing to a metric. Consent settings, blocked scripts, app-based consumption, shared devices, ad blockers, browser privacy protections, and social-platform reporting differences can all make the numbers incomplete. A metric should be described as an estimate when it does not represent every user.
Set a baseline and a realistic target
A target is useful only when it reflects a baseline and a reason. Start by asking what similar content currently achieves, if comparable data exists. If there is no baseline, run a short observation period or record the first meaningful sample without treating it as a final benchmark.
Targets can be expressed in several ways:
- A minimum number of qualified users reached
- A percentage improvement over a comparable project
- A completion or return rate above a defined threshold
- A response-time goal for corrections or audience questions
- A number of verified community or institutional follow-ups
Avoid arbitrary targets such as doubling traffic in a month unless the project’s purpose genuinely depends on scale. Targets should also account for seasonality, news events, distribution changes, and differences between platforms. A metric that rises because a major unrelated event sends visitors to your site may not indicate project success.
Use qualitative evidence alongside analytics
Some of the most important results of journalism are not visible in standard dashboards. A reader may explain that an investigation changed how they understood a local decision, helped them challenge an incorrect bill, or gave them language to participate in a public meeting. These accounts should not replace quantitative measurement, but they can explain what the numbers miss.
Useful qualitative methods include:
- Short reader surveys with one or two open-ended questions
- Interviews with people in the intended audience
- Structured conversations with community organizations
- Review of recurring questions, corrections, and support requests
- Editorial review of representative user journeys
- Collection of documented examples of reporting being used
Ask neutral questions. “Did this article help you?” is more useful when followed by “What were you trying to do?” and “What, if anything, did you do next?” Be transparent about limitations: a person who responds to a survey may not represent the wider audience.
Watch for misleading incentives
Every metric changes behavior. If a team is judged on clicks, it may prioritize sensational topics. If it is judged only on completion rate, it may shorten articles or move important context elsewhere. If it is rewarded for subscriptions, it may focus on existing loyal readers and neglect people who need public-service reporting.
Before adopting a metric, ask:
- What behavior would this metric reward?
- What valuable work might the team ignore?
- Could the number rise while journalism quality falls?
- Who is missing from the measurement?
- Could optimization encourage privacy-invasive or manipulative practices?
- What would a bad result mean, and what would it not mean?
Include a safeguard against the most likely distortion. If reach is important, pair it with a quality or trust measure. If conversion is important, monitor whether the audience includes the people the project was intended to serve.
Review metrics at planned checkpoints
Measurement is not a one-time choice. Set review points during planning, production, publication, and follow-up. Early data can reveal distribution problems, confusing language, or missing audiences. Later evidence may show effects that were invisible on publication day.
A simple review rhythm is:
- Before publication: confirm the purpose, audience, definitions, privacy approach, and tracking setup.
- After launch: check for technical failures, unexpected audiences, and obvious usability problems.
- After the first meaningful period: compare results with the baseline and review reader feedback.
- At project close: document what the metrics showed, what they could not show, and what should change next time.
Keep a short measurement note with the project. Record definitions, dates, known gaps, and decisions made from the data. This prevents future teams from comparing numbers that were calculated differently.
Troubleshoot common metric problems
The number is high, but the project seems ineffective. Check whether the metric measures exposure rather than understanding or use. Review the audience mix, traffic sources, and reader feedback. Add a metric closer to the project’s intended outcome.
The number is low, but the journalism appears valuable. Investigate distribution, accessibility, language, search visibility, device performance, and whether the target audience is represented in the data. Valuable journalism may need a different channel, partner, format, or time horizon.
Different platforms report different results. Do not force them into a single total without documenting definitions. Compare trends within each platform and use a shared event definition where possible.
Tracking is incomplete. State the limitation plainly, estimate its likely effect if you can, and avoid false precision. Use surveys, interviews, or representative reviews to supplement missing behavioral data.
The team is optimizing the dashboard instead of the journalism. Return to the purpose statement. Require every metric review to include a short explanation of what the result means for editorial decisions and what it does not prove.
A practical final checklist
Before approving a metric, confirm that it:
- Reflects the project’s purpose rather than a convenient proxy
- Names the audience it is intended to represent
- Has a clear definition and calculation method
- Can be collected consistently and ethically
- Includes a realistic baseline or explanation for its absence
- Is paired with at least one quality or equity measure
- Acknowledges privacy, technical, and attribution limitations
- Leads to a specific decision or learning question
- Will be reviewed alongside qualitative evidence
- Does not reward behavior that undermines public-service journalism
The best metric is rarely the most impressive number. It is the measure that helps a news team make better editorial decisions, notice who is being missed, and understand whether its work is useful to the people it intends to serve.