Results
Evidence, with the methodology attached
Every figure below comes from live deployments and is reported with the measurement behind it.
Case study
Gamingtec: personalisation across 8 casino brands
Expand the full case study — challenges, solution, and A/B tested results.
Agent · 30-day operational proof
Support automation, measured over 30 days
Operational measurements from a live iGaming support deployment. Containment and confirmed AI-resolved are reported separately because they answer different questions.
- AI conversations handled
- 89K+AI conversations handled
- contained without human escalation
- 57.2%contained without human escalation
- confirmed AI-resolved
- 19.4%confirmed AI-resolved
- FTE-equivalent capacity freed per month
- ~12FTE-equivalent capacity freed per month
Operational measurement vs. external benchmark
25.1s
Average response time — measured in the deployment.
4+ min
Industry benchmark for live chat response, external source: Comm100.
Agent outcomes · 30-day window
89K+ conversations- Average response time
- 25.1s
- Capacity freed / month
- ~12 FTE
Lobby · 40 experiments, 11 deployments
Personalisation held to a control group
Uplift is the measured difference between a personalised cohort and a holdout group, significance tested at p<0.05, run per brand.
- completed A/B experiments
- 40completed A/B experiments
- live deployments
- 11live deployments
- significance threshold
- p<0.05significance threshold
- typical time to live
- 4–8 wkstypical time to live
Personalised experience
Ranking applied per customer
Control / holdout group
Ranking withheld
Measured difference, significance tested at p<0.05
- Click-through rate
- +12% to +66%
- Session duration
- +11% to +25%
- ARPPU
- +6% to +21%
- ARPU
- +4% to +7%
- 7-day retention
- +2% to +5%
Ranges observed across 40 completed experiments in 11 live deployments. Results vary by brand, traffic and content mix — the upper end of a range is not an expected outcome.
How we report
Three rules for every number on this site
Separate the measures
Containment is not resolution. We publish both.
Name the source
External benchmarks are attributed; everything else is our own measurement.
Ranges are ranges
Experiment ranges describe what has been observed, not what is promised.
FAQ
What operators ask first
- How long does implementation take?
- It depends on the product, so we don't quote one universal timeline. Agent is typically live in 2–3 weeks. Lobby is typically live in 4–8 weeks, because the first experiment cycle is part of going live.
- How does Whizdom work with our existing systems?
- Integration is API-first into the front-ends, CRM and platforms you already run. There is no replatforming requirement.
- How are results validated?
- Lobby uplift is measured against a control/holdout group and significance tested at p<0.05 — 40 completed experiments across 11 live deployments to date. For Agent we report operational measurements from live deployments, and we separate containment (57.2%) from confirmed AI-resolved outcomes (19.4%), because they are different measurements.
- How quickly can we see results?
- Agent produces operational measurement as soon as it is handling live conversations; the figures we publish come from a 30-day window. Lobby results come from completed experiment cycles inside the 4–8 week window.
- What happens if the pilot doesn't hit the agreed criteria?
- Success criteria are agreed before the pilot starts. If they are not met, there is no further commitment.
- How is customer data handled?
- Data handling is GDPR-aligned, including removal on request, and the agent works against your live operational data rather than storing a parallel copy of your customer base.
See it on your own customer journeys
A 30-minute walkthrough: live customer scenarios and the dashboard. We'll agree the success criteria before anything goes live.