Intelligence Score

The Intelligence Score is Apex's meta-metric. While most analytics tools measure what happened, the Intelligence Score measures how well your organization is learning. It quantifies whether your team is getting smarter over time — turning experiments into compounding knowledge.

What It Measures

The Intelligence Score isn't a vanity metric. It's a composite of four signals that together reveal your team's growth maturity:

The score ranges from 0 to 100. Most teams start between 10–20 and improve as they adopt the full believe → predict → test → update loop.

Belief Validation Rate

This measures what percentage of your active beliefs have been tested by at least one completed experiment.

• A belief that's been tested (regardless of the outcome) counts as validated
• A belief sitting at its initial confidence with no linked experiments counts as unvalidated

Why it matters: Untested beliefs are just opinions. The more of your belief system that's grounded in evidence, the better your decisions will be.

How to improve: Link experiments to beliefs. When you run a test, connect it to the belief it's testing. Even disproving a belief counts as validation — you learned something.

Prediction Accuracy

This is your team's calibration score — how closely your predicted outcomes match actual experiment results.

A perfectly calibrated team:

• Predicts 60% confidence → right about 60% of the time
• Predicts 90% confidence → right about 90% of the time

Why it matters: Teams that can accurately predict outcomes make better resource allocation decisions. If you know a test is likely to win, you can invest more in the rollout plan before results are in.

How to improve: Log predictions before every experiment. Review your calibration curve monthly. Most teams start overconfident — awareness alone helps correct this.

Experiment Velocity

This measures how many experiments your team runs per time period, normalized for your team size and product complexity.

It's not just about quantity — experiments must be completed (not abandoned) and have sufficient sample size to count toward velocity.

Why it matters: You can't learn if you're not testing. Teams that run one experiment per quarter will always be out-learned by teams running one per week.

How to improve: Lower the barrier to experimentation. Use text-change variants for quick tests. Set up reusable goals so new experiments are fast to configure. Build a backlog of experiment ideas so there's always something ready to run.

Learning Rate

Learning rate measures how quickly your beliefs update in response to new evidence. It captures the feedback loop speed — the time between an experiment completing and its results being incorporated into your belief system.

A high learning rate means:

• Experiments have linked beliefs that update automatically
• Results are reviewed promptly (not sitting in a "completed" state for weeks)
• The team acts on what they learn — updating strategy based on new confidence levels

Why it matters: Speed of learning is a competitive advantage. Two teams with identical experiment velocity will diverge rapidly if one incorporates learnings in days and the other takes months.

How to improve: Link every experiment to a belief. Review completed experiments within a week. Use belief confidence changes as inputs to planning — if a belief just jumped from 0.4 to 0.8, that should influence your next sprint.

The Compounding Knowledge Effect

The Intelligence Score reveals something that traditional analytics miss: knowledge compounds.

Each experiment doesn't just produce a result — it updates beliefs, calibrates predictions, and informs future experiments. A team at Intelligence Score 60 isn't just "better" than a team at 20 — they're operating with a fundamentally different quality of information.

Here's how compounding works in practice:

1. Early stage (Score 10–25): You're running experiments, but beliefs are mostly untested and predictions are unreliable. Decisions are still largely based on intuition.

2. Building (Score 25–50): A meaningful portion of beliefs are evidence-based. Your prediction accuracy is improving. You're starting to know why things work, not just what works.

3. Compounding (Score 50–75): Your belief system is largely validated. Predictions are well-calibrated. New experiments are informed by a rich base of prior knowledge, so they're better designed and more likely to produce actionable results.

4. Mastery (Score 75–100): Your team has genuine institutional knowledge. You can predict outcomes with high accuracy, design experiments that test the right things, and make strategic decisions backed by a validated model of your market.

Viewing Your Score

The Intelligence Score appears on the main dashboard and updates daily. Click into it to see:

• Your score breakdown by component
• Trend over the last 30/60/90 days
• Specific recommendations for improvement
• How each recent experiment impacted your score

Best Practices

• Focus on the weakest component. Improving your lowest-scoring component has the biggest impact on overall score.
• Don't game it. Running trivial experiments to boost velocity will hurt prediction accuracy and learning rate.
• Use it as a team metric. The Intelligence Score works best when the whole team sees it and understands what drives it.
• Track the trend, not the number. A score of 35 that's been climbing steadily is better than a score of 50 that's been flat for months.