How a CRM Manager Reads A/B Test Results Across 4,800 Players in 18 Minutes

AzimuthBet is a Cyprus-based multi-market operator headquartered in Limassol, serving approximately 28,000 monthly active players across European and emerging markets. The platform generates around $9M per week in gross gaming revenue, running a sophisticated CRM stack with fifteen or more automated lifecycle campaigns active at any given moment — welcome sequences, win-back flows, VIP escalation triggers, deposit match offers, and free spins series all firing in parallel around the clock.

Products used: A/B Test Analytics, Campaign ROI, Statistical Analysis

18 minutes | full test analysis time

4,800 players tested | statistically powered split across two bonus types

23% | revenue-per-player advantage for deposit match over free spins in high-value segment

---

Challenge

Four weeks ago, Nadia Petrov split AzimuthBet's mid-tier retention pool in half and ran the operator's most structured experiment in years. One group received a deposit match offer — 50% up to €150 — on the fourth day after their last deposit. The other received an equivalent-cost free spins bundle. The question wasn't which bonus type players preferred. The question was which one actually made AzimuthBet more money while keeping players engaged long enough to matter.

AzimuthBet's CRM stack is powerful, but reading test results across a live platform with fifteen campaigns running simultaneously is genuinely difficult. Nadia had the raw output — redemption rates, revenue logs, activity timestamps — but turning those into a decision required reconciling numbers from three different systems, manually adjusting for the sessions where both groups had other campaigns fire simultaneously, and then calculating statistical significance by hand in a spreadsheet she'd built herself. On a good week, that process took her most of a Friday afternoon. On a bad week, she'd spot a confound on Monday morning and have to redo the work.

"Running the test is the easy part. Reading it correctly — accounting for sample contamination, checking whether the result is real or just noise, understanding how it played out differently across segments — that's where the hours go. I have the data. I just need to trust the analysis."

— Nadia Petrov, Retention / CRM Manager, AzimuthBet

The stakes this time were higher than usual. The winning approach would be rolled out across AzimuthBet's entire mid-tier retention stack, a population of roughly 11,000 players. Getting the answer wrong in either direction — deploying the wrong bonus type, or missing a segment-level nuance that required a hybrid approach — would cost margin at scale. Nadia needed results she could defend, not results she could argue with.

---

Solution

Nadia opened Gaming Mind AI Friday morning with four weeks of test data ready and a board-style presentation to prepare for Monday. She had 18 minutes before a cross-functional call. The AI had access to both test arms, all player activity, the campaign firing logs, and AzimuthBet's historical retention benchmarks. Here's how the conversation went:

---

Nadia: "The 'deposit match vs free spins' 4-week retention test just finished. Give me the headline — did either arm win cleanly, and is the result statistically significant?"

Metric	Deposit Match Arm	Free Spins Arm	Winner
Sample size	2,412 players	2,388 players	Balanced
Revenue per player (28-day window)	€38.40	€31.20	🟢 Deposit Match (+23%)
Offer redemption rate	54%	71%	🟢 Free Spins (+17 pp)
Statistical significance	Both arms sufficient to detect 5% difference at 95% confidence	—	✓ Significant

⚠️ Gaming Mind flags: Two different definitions of "winning" — revenue per player favors deposit match by 23%, but conversion rate favors free spins by 17 pp. The right answer depends on which player segment you're targeting.

The result was clean on statistical significance, but more nuanced on the winner. Gaming Mind confirmed that with 2,412 players in the deposit match arm and 2,388 in the free spins arm, both groups had enough volume to detect a 5% difference in the primary outcome with 95% confidence. The deposit match arm generated €38.40 in revenue per player over the test window; free spins came in at €31.20 — a 23% gap that cleared statistical significance comfortably. But conversion rate — the share of players who redeemed the offer at all — ran the other direction: 71% for free spins versus 54% for deposit match. Two different definitions of "winning," both technically correct.

---

Nadia: "Walk me through the revenue difference. Is it coming from higher deposits, better retention, or just higher-value players ending up in the deposit match arm by chance?"

Metric	Deposit Match	Free Spins	Notes
Pre-test GGR per player difference	<2%	—	🟢 Arms balanced — result genuine
Pre-test deposit frequency	Equivalent	Equivalent	No randomization artifact
Avg deposit amount (test window)	€194	€147	+32% higher real money commitment
Avg session count (28 days)	8.1 sessions	7.9 sessions	Nearly identical — not more engaged, just higher stakes

⚠️ Gaming Mind flags: The randomization quality check — run automatically without being asked — confirms the revenue gap is real, not a baseline imbalance. Deposit match players deposited €194 vs €147 for free spins, with nearly identical session counts — the offer selected for higher real-money commitment, not more engagement.

Gaming Mind ran the randomization quality check first, without being asked — a critical step Nadia usually had to remember to do manually. The two arms were balanced at baseline: pre-test GGR per player differed by less than 2%, deposit frequency was equivalent, and no meaningful age or registration cohort skew existed between groups. The revenue difference was genuine, not a randomization artifact. Within the test window itself, deposit match players deposited an average of €194 versus €147 for free spins players — and crucially, their average session count over the 28 days was nearly identical at 8.1 versus 7.9. The deposit match arm wasn't more engaged; it was simply putting more real money into play each time it showed up.

---

Nadia: "What about retention? I care about D7 and D30 — which arm kept players longer?"

Retention Metric	Deposit Match	Free Spins	Notes
D7 Retention	53%	61%	🟢 Free Spins leads early (zero-risk re-entry)
Crossover point	Day 11	Day 11	Curves cross — deposit match takes the lead
D30 Retention	38%	35%	🟢 Deposit Match leads long-run

⚠️ Gaming Mind flags: Free spins excel at pulling lapsing players back short-term because zero-risk play lowers the friction for players not yet ready to commit real money. But players who do commit via deposit match show stronger long-run attachment. The bonus type is selecting for different player psychology — not just different player value.

Here the picture reversed, and this was the insight Nadia had been expecting but couldn't prove without the analysis. Free spins outperformed on early retention: D7 retention ran 61% versus 53% for deposit match. The gap narrowed over time and the two arms crossed at day eleven, after which deposit match players showed slightly higher activity through day 30 — 38% retained versus 35%. Gaming Mind flagged the crossover explicitly: free spins excelled at pulling players back in the short term, likely because zero-risk play lowered the friction for lapsing players who weren't ready to commit real money. But players who did commit real money via deposit match showed stronger long-run attachment. The bonus type was selecting for different player psychology, not just different player value.

---

Nadia: "Break the results down by player segment — I want to see high-value, mid-value, and casual separately."

Segment (by 90-day pre-test GGR)	Deposit Match Revenue/Player	Free Spins Revenue/Player	Deposit Match D30	Free Spins D30	Winner
High-value (€500+ GGR)	—	—	52%	43%	🟢 Deposit Match (+29% rev, better retention)
Mid-value	—	—	~42%	~44%	🟡 Deposit Match (+11% rev, FS slight D7 edge)
Casual (<€100 GGR)	~equal	~equal	~31%	~35%	🟢 Free Spins (higher conv, better D7, similar rev)

⚠️ Gaming Mind flags: This is the decisive finding. In the high-value segment, deposit match wins on every metric. In the casual segment, free spins win outright. A blanket deposit match rollout would degrade performance for a third of the target population.

This was the decisive finding. In the high-value segment — players with €500 or more in GGR over the 90 days before the test — deposit match won on every metric: 29% higher revenue per player, comparable D7 retention, and meaningfully better D30 retention at 52% versus 43%. In the mid-value segment, the gap narrowed to 11% in favor of deposit match, with free spins holding a small D7 edge. In the casual segment — players under €100 in pre-test GGR — free spins won outright: higher conversion, higher D7 retention, and nearly identical revenue per player despite the lower nominal amount. For casual players, a no-risk free spins bundle was the right offer. For high-value players, a deposit match that rewarded real commitment outperformed by nearly a third.

---

Nadia: "What's the cost efficiency look like? Both offers cost AzimuthBet roughly the same in bonus liability — confirm that and show me the ROI per arm."

Metric	Deposit Match	Free Spins
Total bonus liability	€48,200	€46,800
GGR generated	~€92,700	~€74,500
ROI (GGR per €1 bonus liability)	€3.84	€2.96
High-value segment ROI	€5.20	~€3.20
Non-redemption rate	46%	29%

⚠️ Gaming Mind flags: The 46% non-redemption rate for deposit match is a better outcome, not a worse one — bonus liability is concentrated in more engaged players who were going to deposit anyway. AzimuthBet is not subsidizing players who will churn regardless.

Gaming Mind confirmed the offers were cost-equivalent in aggregate: total bonus liability ran €48,200 for deposit match and €46,800 for free spins — close enough that cost wasn't the variable. The ROI diverged sharply. Deposit match returned €3.84 in GGR for every €1 of bonus cost. Free spins returned €2.96. The gap widened further in the high-value segment, where deposit match reached €5.20 per €1 of liability. Gaming Mind noted one important nuance: the higher non-redemption rate in deposit match (46% of recipients didn't use it at all) meant bonus liability was concentrated in more engaged players — a better outcome, not a worse one. AzimuthBet wasn't subsidizing players who were going to churn regardless.

---

Nadia: "Are there any campaign contamination issues I should know about — players in the test who also received other retention campaigns during the 4 weeks?"

Contamination Check	Result
Players receiving ≥1 additional campaign during test window	19% of both arms
Types of additional campaigns	Win-back, VIP upgrade nudges, weekend reload offers
Sensitivity analysis: results excluding contaminated players	Directional results held in every segment
Revenue gap between arms (clean subgroup)	Slightly widened — deposit match advantage mild understated in full sample
Overall conclusion	🟢 Contamination did not introduce bias — result is robust

⚠️ Gaming Mind flags: The contamination didn't weaken the deposit match finding — it slightly dampened it. The clean subgroup result is actually stronger, giving Nadia a number she can defend under scrutiny without qualification.

Nineteen percent of players in both arms received at least one additional campaign during the test window — win-back messages, VIP upgrade nudges, or weekend reload offers from AzimuthBet's always-on stack. Gaming Mind ran a sensitivity analysis excluding those players and re-computed the primary outcomes. The directional results held in every segment, and the revenue gap between arms actually widened slightly in the clean subgroup. The contamination hadn't introduced bias; if anything, it had slightly dampened the deposit match advantage by mixing in win-back campaigns that performed differently across the two groups. Nadia's test result was robust.

---

Nadia: "Based on everything, give me a rollout recommendation I can take into Monday's cross-functional call."

Segment	Population	Default Offer	Fallback (if no redemption in 48h)
High-value (€500+ GGR)	~3,200 players	Deposit Match	—
Mid-value	~5,100 players	Deposit Match	Free Spins (capture D7 strength as rescue mechanic)
Casual (<€100 GGR)	~2,700 players	Free Spins	—
Incremental GGR vs uniform free spins	11,000 total	+€41,000/month	At identical bonus budget

⚠️ Gaming Mind flags: The incremental €41,000/month projection comes from applying segment-appropriate offers without increasing the bonus budget. The free spins fallback in the mid-value arm deliberately uses free spins' D7 strength as a rescue mechanic within a deposit match-first strategy.

Gaming Mind produced a tiered recommendation Nadia could defend at the table without hedging. High-value players — roughly 3,200 players in the mid-tier retention pool — should receive deposit match offers as the default retention lever. Mid-value players, approximately 5,100, should receive deposit match as the primary offer with free spins as a fallback for those who don't redeem within 48 hours. Casual players, the remaining 2,700, should default to free spins. Applied to AzimuthBet's full mid-tier population, the model projected an incremental €41,000 per month in GGR relative to running free spins uniformly — without increasing the bonus budget. Nadia reconfigured the three campaign branches in the CRM stack before the end of the day.

---

Results

The winning approach identified in 18 minutes, not half a Friday

Nadia's full analysis — from test summary to rollout recommendation — took eighteen minutes of conversation. She didn't open a spreadsheet, run a significance calculation by hand, or spend time checking whether the randomization held. Gaming Mind handled all of it, including the contamination sensitivity check she would have spent an hour on alone.

A hybrid rollout found that a single-winner answer would have missed

The most valuable insight wasn't that deposit match won — it was that free spins won in the casual segment, and a blanket rollout of deposit match would have degraded performance for a third of the target population. The segment-level breakdown, which Gaming Mind surfaced without Nadia asking for it explicitly, turned a binary A/B result into a three-way campaign structure that captured value across the full player spectrum.

€41,000 per month in projected incremental GGR

Applied across AzimuthBet's 11,000-player mid-tier retention pool at the segment-appropriate split, the new campaign structure projected €41,000 in monthly incremental GGR versus the previous uniform free spins approach — at identical bonus cost. The reconfiguration went live the same Friday the test concluded.

Campaign contamination ruled out, results defended with confidence

Nineteen percent of test players had received other campaigns during the window. Before Gaming Mind, Nadia would have flagged this as a concern and spent hours running sensitivity checks manually or asking the analytics team to help — a conversation that typically added two days to the decision timeline. Gaming Mind ran the check in the same session and cleared the result as robust, giving Nadia something she hadn't had before: a number she could defend under scrutiny.

D7 vs D30 retention nuance preserved in the rollout design

Because Gaming Mind surfaced the crossover in retention curves at day eleven, Nadia built the free spins fallback trigger into the mid-value segment's campaign branch specifically to capture the short-term re-engagement benefit. Players who didn't redeem the deposit match within 48 hours received a free spins offer instead — effectively using free spins' D7 strength as a rescue mechanic within a deposit match-first strategy.

"Before, I'd have a test result and a spreadsheet and a nagging feeling I'd missed something. Now I have a recommendation with the statistical work already done, the contamination already checked, and a segment breakdown I didn't even know to ask for. I walked into Monday's call and defended every number. Nobody pushed back, because there was nothing to push back on."

— Nadia Petrov, Retention / CRM Manager, AzimuthBet