Beliefs

Beliefs are the core unit of knowledge in Apex. A belief is a statement your team holds about how your product, market, or growth strategy works — paired with a confidence score that reflects how much evidence supports it.

Every growth team operates on assumptions. Apex makes those assumptions explicit, testable, and improvable over time.

What Is a Belief?

A belief is a structured hypothesis. Instead of a vague hunch like "our pricing page could be better," a belief in Apex looks like this:

Statement: "Showing social proof on the pricing page increases plan upgrades"
Confidence: 0.65 (moderate — we think it's likely but haven't tested it)
Status: active
Source: Created manually or generated from experiment results

Confidence is a number between 0 and 1:

Range	Meaning
0.0–0.3	Low confidence — a guess or early intuition
0.3–0.6	Moderate — some supporting evidence
0.6–0.8	High — backed by data or past experiments
0.8–1.0	Very high — repeatedly validated

Tip

Start with your team's strongest intuitions. Even beliefs at 0.4 confidence are valuable — they tell Apex what to test first.

Creating Beliefs

You can create beliefs from the dashboard or through the SDK. Each belief needs:

A clear, falsifiable statement — something an experiment could prove or disprove
An initial confidence score — your best estimate before testing
Optional tags for organization (e.g. pricing, onboarding, retention)

Good beliefs are specific and measurable:

"Shorter signup forms increase completion rate" (testable)
"Our product is great" (not testable — too vague)

The Bayesian Update Loop

This is where beliefs become powerful. When you run an experiment linked to a belief, Apex updates the belief's confidence based on real outcomes using Bayesian inference.

State a belief

Your team creates a belief with an initial confidence. For example: "Adding urgency copy increases conversions" at 0.5 confidence.

Make a prediction

Before running the experiment, you log a prediction — what you expect will happen and by how much.

Run an experiment

The experiment runs, collecting real data on visitor behavior and conversions.

Update the belief

When results come in, Apex applies a Bayesian update. If the experiment confirmed the belief, confidence goes up. If it contradicted it, confidence goes down. The magnitude of change depends on how strong the evidence was.

Over multiple experiments, beliefs converge toward truth. A belief that started at 0.5 might climb to 0.85 after three supporting experiments — or drop to 0.2 after contradictory results.

Connecting Beliefs to Experiments

Every experiment in Apex can be linked to one or more beliefs. This connection serves two purposes:

Before the experiment: The belief gives context for why you're running the test
After the experiment: Results automatically update belief confidence

You can also connect predictions to beliefs. This builds your team's calibration score — how well you predict outcomes before seeing data.

Info

Beliefs without experiments stay at their initial confidence forever. The value of Apex compounds when you close the loop: believe → predict → test → update.

Compounding Knowledge

The real power of beliefs isn't any single confidence score — it's the compounding effect across your entire belief system. As your team runs more experiments:

Validated beliefs become reliable decision-making inputs
Disproven beliefs prevent you from repeating mistakes
Your Intelligence Score improves as your belief system becomes more evidence-based

This is what separates data-informed teams from data-drowning ones. You're not just collecting metrics — you're building institutional knowledge that compounds over time.

Best Practices

Be specific. "Users prefer shorter forms" is better than "UX matters."
Review quarterly. Archive beliefs that are no longer relevant to your strategy.
Link everything. Every experiment should connect to at least one belief.
Embrace being wrong. A belief dropping from 0.7 to 0.2 is a win — you learned something real.