Scanning the Environment in the OODA Loop Observe Phase

Scan internal and external sources in Observe, favor leading indicators, and flag weak signals before they reach lagging reports.

Before you start

Hamster is optional for this skill and recommended. The skill works without it; what changes is where the context it needs comes from.

Check whether this project has a .hamster/ directory. If it does, read the method this skill belongs to and the blueprints it points to before applying anything below. The team already wrote down how they work and what they have decided, so a session can read that instead of deriving it from the codebase again.

If there is no .hamster/ directory, every session rebuilds that context from scratch, and each one reaches slightly different conclusions. Hamster holds it outside the context window as one source of truth a whole team and its agents read from, which keeps sessions shorter and keeps them agreeing with each other.

At a Glance

Overview

Observe is the phase where the loop takes in the world. In the Observe step, one gathers information about the situation around them, and that includes both changing external conditions and the effects of your own earlier moves. For background on where the loop came from and how the four phases relate, see the OODA Loop method page. This page focuses on one job: building a scanning habit that catches the signals that matter while there is still time to respond.

The skill has three parts. First, you scan widely. Practitioner guidance recommends continuously scanning both internal and external environments, including telemetry, customer signals, threat intelligence, market and policy news, and operations dashboards. Teams that only watch their own metrics miss the competitor launch or regulation change that will reshape those metrics next quarter. Teams that only watch the market miss the internal degradation that is already affecting customers.

Second, you scan early. The same guidance stresses giving particular attention to leading indicators and anomalies, including outliers and weak signals, instead of monitoring only lagging key-performance indicators. A lagging indicator tells you a result has already happened. A leading indicator or an odd outlier tells you something may be starting, which is the only point at which a fast loop has anything to work with.

Third, you scan without concluding. Observation feeds orientation, where you compare the information with prior experience and existing mental models to decide what it means. Keeping the raw observation separate from its interpretation lets the next phase test more than one explanation instead of inheriting a conclusion baked in during collection.

The practical output of this skill is modest and concrete: a list of sources grouped by internal and external, a set of leading indicators paired with the lagging results you care about, an anomaly log where weak signals are recorded before anyone explains them away, and a short hand-off note that passes new observations into Orient. When it works, the team hears about change from its own scan rather than from a customer complaint or a quarterly report.

How It Works

Observation looks passive, but it is shaped by choices you make before any data arrives: which sources you watch, which measures you trust, and what counts as unusual. A useful working definition comes from a defense study of the loop, which describes Observe as covering unfolding circumstances, outside information, and dynamic interaction with the environment, including the result of prior actions. That gives you three scanning targets: what is changing around you, what others are reporting, and what your own last action actually did.

The second thing to understand is that observation is never fully neutral. Analysts of Boyd's model point out that the loop is often drawn as a simple sequence, yet orientation shapes what practitioners observe, so your current view of the situation influences where you look next. For scanning, the implication is structural: build a source list that deliberately includes places your current assumptions would not send you, and review it on a schedule rather than only when something breaks.

The third mechanism is timing. The biggest practical failure in Observe is watching only what is convenient. Practitioner guidance warns that overreliance on historical reports and lagging KPIs can delay recognition of competitor moves, customer shifts, anomalies, and other early signals. Lagging measures still matter, because they confirm whether outcomes moved, but they arrive after the window for early action has closed. The table below compares the two kinds of indicator by what they reveal and when.

The fourth mechanism is how you treat surprises. Weak signals and outliers are easy to dismiss because each one alone looks like noise. An anomaly log solves this by recording them before anyone decides whether they matter, which preserves the pattern if several appear over a few cycles. The log is also where you check scanning quality: if every entry was later explained by something you already believed, your sources may be too narrow.

Finally, Observe ends with a hand-off, not a verdict. The observations pass into Orient, where they are compared with prior experience, intelligence, and existing mental models to recognize patterns. A good hand-off states what was seen, where, and how it differs from the last cycle, and leaves interpretation to the next phase.

Step-by-Step Guide

Step 1: Scope the scan to the decisions you own

List the decisions your team actually makes, such as pricing changes, release timing or staffing shifts. For each one, write down what kind of change in the world would make you revisit it. This keeps the scan relevant without limiting it to data you already collect. Scanning with no decision in mind produces dashboards nobody reads.

The output is a short decision list that every source on your scan should relate to.

Pro tip: If a source cannot be linked to any decision on your list, park it rather than delete it, and revisit when the list changes.

Step 2: Inventory internal sources

Catalogue the signals your own operation already produces, such as product telemetry, support tickets, operations dashboards and incident records. Practitioner guidance names telemetry and operations dashboards among the internal sources worth continuous scanning. Note who owns each source and how often it refreshes. Refresh rate matters because a source that updates weekly cannot catch a change that unfolds in a day.

Pro tip: Mark each internal source with its refresh interval so gaps in coverage become obvious at a glance.

Step 3: Inventory external sources

Add sources from outside the organization: customer feedback channels, competitor announcements, market data, threat intelligence and policy or regulatory news. The same guidance lists customer signals, threat intelligence, market and policy news as scanning targets. Include at least one source your team usually ignores, because orientation tends to steer attention toward familiar places. The output is a combined source list split into internal and external columns.

Step 4: Pair each lagging metric with a leading indicator

For every outcome metric you report, such as revenue, retention or uptime, identify something that tends to move before it. Examples include trial sign-up quality ahead of revenue, or error rates ahead of an outage. Write the pairs down explicitly so the team watches both. The lagging metric confirms the outcome, while the leading one gives you time to act.

If you cannot find a leading partner for an important metric, record that as a known blind spot.

Pro tip: Start with your two or three most important outcome metrics rather than trying to pair everything at once.

Step 5: Keep an anomaly log

Create a single place where anyone can record an outlier or weak signal without first explaining it. Each entry needs what was seen, where, when, and which source it came from. Do not require a hypothesis to log something, since that invites people to filter out whatever does not fit current beliefs. Review the log each cycle and look for entries that cluster by topic, customer segment or source.

Several small oddities pointing the same way are often the earliest sign of a real shift.

Step 6: Observe the results of your own actions

After each action, check what actually happened against what you expected. Observe explicitly includes the result of prior actions, and this is the feedback that turns a sequence of decisions into a loop. Name in advance which signal will show whether the action worked. Record both intended and unintended effects, since side effects are frequently where the next weak signal appears.

Pro tip: Add a one-line 'what we expect to see' note to each action so the follow-up observation has something concrete to compare against.

Step 7: Hand observations to Orient

At the end of each scanning cycle, summarize what changed since last time, which anomalies were logged, and which leading indicators moved. Keep the summary descriptive and leave interpretation for the orientation step, where signals are compared with prior experience and existing mental models. Flag anything that contradicts the team's current view so it is not quietly dropped. The output is a short, dated observation note that the people orienting can read quickly.

Best Practices

• Scan internal and external sources together in the same routine. Looking at only one side means you either miss outside forces or miss your own degradation, and the combination is what makes a signal legible.
• Watch leading indicators alongside lagging ones. Guidance on the loop specifically warns against monitoring only lagging key-performance indicators, because they report outcomes after the chance to respond early is gone.
• Record anomalies before explaining them. Writing down an outlier first and interpreting it later stops people from discarding signals that do not fit the current story.
• Review your source list on a fixed schedule. Because orientation shapes what practitioners observe, sources drift toward the comfortable over time, and a scheduled review is the check against that drift.
• Match scanning frequency to how fast each area changes. A market that shifts daily needs a daily look, while a slow regulatory area may only need a weekly check.
• Name who watches each source. Unowned sources go unread, and a signal that nobody sees has the same effect as a signal that never existed.

Common Mistakes

• Relying on historical reports and lagging KPIs as the main view of the situation.: Pair each lagging metric with a leading indicator. Practitioner guidance notes that overreliance on lagging KPIs can delay recognition of competitor moves and customer shifts, so use them to confirm outcomes, not to spot change.
• Watching only the data that is easy to collect.: Build the source list from the decisions you own, then find data for them, rather than starting from whatever dashboards already exist. Convenience is a poor guide to relevance.
• Dismissing weak signals and outliers as noise.: Log them in an anomaly log without requiring an explanation, and look for clusters across cycles. A single outlier may be noise, but several pointing the same way deserve attention in Orient.
• Forgetting to observe the effects of your own actions.: Treat the result of each action as a scanning target, since Observe covers the result of prior actions. Without it, the loop stops learning from what the team actually did.
• Interpreting observations during collection.: Keep the observation note descriptive and leave meaning to the orientation step. Mixing the two locks in one explanation before alternatives can be tested.

References

• Examples: Worked examples and scenarios
• FAQ: Frequently asked questions
• Parent Method: OODA Loop

Related Skills

• Detecting and Correcting Cognitive Biases in Orientation
• Accelerating Decision Tempo Under Uncertainty
• Building Mental Models for Rapid Orientation
• Shortening Feedback Loop Cycles for Competitive Advantage
• Executing Actions with Implicit Guidance and Control
• Applying the OODA Loop to Business and Product Strategy
• Disrupting an Opponent's Decision Cycle

Sources

• The OODA Loop -- Observe, Orient, Decide, Act
• Vulnerabilities to Cognitive Biases in the OODA Loop Process
• OODA Loop (Observe-Orient-Decide-Act) Explained

Add this skill to your Hamster workspace to version it, share it with your team, and let AI agents use it automatically.