| Pitfall | Symptom | Fix | |---------|----------|-----| | Too few sessions | Bayesian confidence intervals stay wide (> 30 %). | Extend test duration or increase traffic by widening targeting. | | Testing too many variations | Results become noisy; decision latency ↑. | Stick to 2‑4 variations for the first round. | | Ignoring secondary metrics | Primary conversion improves, but bounce‑rate spikes. | Add secondary goals (e.g., “Time on Page”) before launching. | | Hard‑coding API keys | Keys leak in repo → security breach. | Use environment variables + CI‑CD secret management. | | Neglecting mobile | Desktop tests win, mobile conversion falls. | Always add a mobile‑specific segment and test on it separately. | | Over‑optimising small gains | Diminishing ROI, resources wasted. | Apply the 80/20 rule: stop after achieving a ≥ 5 % lift unless a larger opportunity is identified. |
| Resource | Link |
|----------|------|
| FullMaze Academy (video tutorials) | https://academy.fullmaze.com |
| API Documentation | https://docs.fullmaze.com/v3 |
| Community Forum | https://community.fullmaze.com |
| Best‑Practice Playbook (PDF) | https://fullmaze.com/resources/cro-playbook.pdf |
| Live Office Hours (Wed 14:00 UTC) | Register in the Dashboard → Support | better fullmazacom 300 new
1️⃣ Collect data → 2️⃣ Analyse patterns → 3️⃣ Form hypothesis
(heat‑maps, funnels) (segment performance) (e.g., “CTA color causes drop”)
↓ ↓ ↓
4️⃣ Design test (Scenario‑Builder) → 5️⃣ Run A/B‑Multivariate → 6️⃣ Validate results
↓ ↓ ↓
7️⃣ Deploy winning variant → 8️⃣ Monitor post‑launch → 9️⃣ Iterate
