QCI Metrics says its next-day slot intelligence dataset now covers more than 100,000 gaming machines. For slot directors, the important point is not the headline number by itself. The value comes from using a larger pool to build more relevant comparison groups, detect performance changes sooner and test whether a local result is unusual or simply consistent with the wider market.
The service provides QCI customers with next-day benchmarking over rolling 30- and 90-day windows. According to QCI, the daily detail can expose weekday and weekend differences, recent changes inside a longer trend and divergences between play activity and revenue performance. The company also says its theme-name normalization reconciles inconsistent game names across source systems.
A 100,000-machine dataset can strengthen benchmarking, but it does not make every comparison valid. A machine in a destination resort should not automatically be compared with the same cabinet in a locals property, a restricted-hours venue or a different denomination mix. The larger dataset creates more opportunities to find a relevant peer group. Management still has to choose that group correctly.
What The 100,000-Machine Milestone Actually Adds
A small benchmarking pool forces broad comparisons. Products with different denominations, configurations, markets and operating conditions may be combined because there are not enough observations to separate them. That can produce a stable-looking average that is operationally weak.
With more machines represented, an analytics provider can potentially form narrower cohorts while retaining enough observations to produce a useful reference point. Instead of comparing one cabinet with every installation of the same manufacturer, the analysis may be able to retain more context around theme, denomination, configuration and reporting mix.
Scale can improve the precision of a benchmark, especially when extreme results from a few machines would otherwise distort the average. It also gives the system more examples of normal weekday, weekend and seasonal movement. However, statistical precision is not the same as market representativeness. If the dataset is concentrated in particular jurisdictions, property types or customer segments, the benchmark may still be a poor fit for a different operation.
Why Next-Day Slot Intelligence Changes The Management Cycle
Traditional external slot reports may arrive weeks after the activity occurred. By then, the operating team may have changed the floor again, the customer mix may have shifted or the original test conditions may no longer exist. QCI states that its report is current when the operator opens it, covering the preceding 30 or 90 days.
That shorter delay can improve the test-and-review cycle:
- Define the expected result before changing a location, theme, denomination or configuration.
- Record the implementation date and all machines affected.
- Review next-day exceptions without reacting to one day of gaming luck.
- Compare the rolling trend with a relevant external cohort.
- Check whether the change affected the surrounding zone, not only the selected machine.
- Continue, adjust or reverse the test after sufficient volume has accumulated.
The benefit is faster diagnosis, not faster uncontrolled floor changes. Daily information should identify what deserves investigation. Product removals, capital purchases and permanent configuration decisions still require persistence, adequate wagering volume and a clean comparison period.
How A Larger Benchmark Supports WPUPD Analysis
Win per unit per day, or WPUPD, is one of the quickest ways to compare slot productivity after normalizing for the number of machines and reporting days:
WPUPD = Slot win ÷ Active machine-days
Assume a bank produces WPUPD of 240 while the relevant external peer-group median is 300:
WPUPD benchmark index = Property WPUPD ÷ Peer-group median WPUPD × 100
The bank’s index is 80, meaning its reported WPUPD is 20% below the peer-group median. That is a useful signal, but not yet a removal decision. The analyst must check whether the local machines had equal opportunity to earn.
If the bank operated at only 80% availability because of communication faults or extended maintenance, its productivity per available day may be consistent with the benchmark. The correct management response could be a technical intervention rather than a game conversion.
This is why external data should sit inside a broader slot analytics and floor-optimization framework. Location, availability, game age, cost, customer segment and surrounding-floor transfer remain property-specific.
The Metrics That Should Be Benchmarked Separately
QCI says its comparisons distinguish play activity from revenue performance. That separation is important because the same movement in win can have several causes. Coin-in can fall while hold rises. Games played can decline while average bet increases. WPUPD can improve because a small number of machines were removed from the denominator.
| Metric | Benchmarking Use | Main Limitation |
|---|---|---|
| Coin-in per unit per day | Compares wagering demand after normalizing for units and days | Still distorted by downtime and unequal operating hours |
| Games played per unit | Shows whether volume comes from more decisions rather than larger bets | Requires consistent game-count data across systems |
| Average bet | Separates stake-size movement from changes in game count | Can be concentrated among a small number of customers |
| WPUPD | Provides a fast productivity comparison | Actual win can be volatile over short periods |
| Actual hold | Identifies unusual revenue conversion from recorded coin-in | Needs theoretical hold, volume and game variance for interpretation |
| Theoretical win per unit | Provides a more stable expected-value comparison | Depends on accurate paytable and game-mix configuration |
| Net contribution per unit | Supports product, participation and capital decisions | Requires consistent jackpot, lease and direct-cost allocation |
A practical slot-machine performance scorecard should show these measures together. One rank should not hide the components or the reason a machine moved.
What Benchmarking Can And Cannot Say About Hold
A broad dataset can show whether a property’s actual hold, theoretical hold or actual-to-theoretical index differs from comparable products. It cannot prove that a machine is configured incorrectly or that a performance problem exists.
Short-term actual hold is affected by normal gaming variance. A high-volatility game may remain far from its mathematical expectation after the same coin-in that gives a lower-volatility product a relatively stable result. Jackpot outcomes and the distribution of bet sizes also matter.
If local actual hold falls below the benchmark, the operating team should first check:
- Coin-in, games played and the number of active machine-days
- Configured and volume-weighted theoretical hold
- Denomination, paytable and game-mix changes
- Large jackpot and hand-pay outcomes
- Meter mapping, resets and casino-day cut-offs
- Whether the external cohort is genuinely comparable
The distinction between theoretical hold and actual hold remains unchanged by the size of the benchmark. More observations improve context; they do not remove mathematical variance from the local result.
Theme-Name Normalization Is More Important Than It Sounds
Casino systems often describe the same product differently. One database may use a full commercial theme name, another a shortened code and another a legacy name retained after a conversion. Minor differences in punctuation, cabinet identifiers or version numbers can split one product across several records.
QCI says its theme-name normalization reconciles inconsistent names across source systems. That is necessary before cross-property benchmarking can be credible. If the same theme is represented by three labels, its sample is understated. If different versions are incorrectly combined, the benchmark can hide a genuine configuration difference.
Normalization should therefore be auditable. A slot director needs to know which local asset, theme, cabinet, denomination and configuration were mapped into the external cohort. An automated recommendation is difficult to challenge when the underlying product match is invisible.
The Dataset Still Needs A Comparison Hierarchy
A useful benchmark should begin with the closest credible peer group and widen only when the sample becomes too small. An illustrative hierarchy could be:
- Same theme, cabinet, denomination and comparable configuration
- Same theme and denomination across similar cabinet types
- Same product family within a comparable market and property class
- Broader denomination or volatility cohort
- Overall market benchmark used only as context
The report should disclose when it moved to a broader level. Otherwise, management may treat a weakly matched benchmark as if it were a direct product comparison.
Median results are often more useful than simple averages because a small number of exceptionally productive machines can pull the mean upward. Percentile ranges can add more context by showing whether the property sits near the middle of the distribution or at an extreme.
How Slot Directors Should Use The 30-Day And 90-Day Views
The 30-day window is more responsive to recent changes. It can show whether a relocation, conversion or competitor opening coincided with a shift in demand. It is also more exposed to holidays, promotional events, jackpot outcomes and temporary customer concentration.
The 90-day window is slower but more stable. When both windows move in the same direction, the case for intervention becomes stronger. When the 30-day result declines while the 90-day measure remains healthy, management should investigate the recent period before changing the product.
| 30-Day Result | 90-Day Result | Likely Interpretation | First Management Action |
|---|---|---|---|
| Strong | Strong | Persistent product or location strength | Protect availability and verify net contribution |
| Weak | Strong | Recent event, variance or emerging decline | Check daily detail, downtime and local events |
| Strong | Weak | Recent recovery or short-term favourable result | Continue monitoring before declaring a turnaround |
| Weak | Weak | Persistent underperformance against the selected cohort | Test location, configuration, cost and replacement options |
Privacy And AI Do Not Remove Data-Governance Work
QCI states that Metrics uses aggregated, normalized and de-identified information and does not share transaction-level, player-level or campaign-level data. That design allows product benchmarking without exposing individual customer activity.
The company also presents the larger dataset as a stronger foundation for agentic AI and recommendation tools. More timely observations can help models identify patterns, but the output is still limited by source-system accuracy, product mapping and peer-group design. An AI recommendation built on an incorrect denomination, missing downtime or a mismatched theme remains an incorrect recommendation.
Every recommended change should therefore show the data period, selected cohort, comparison metric, sample coverage and operational reason. The slot director should be able to reproduce the logic without accepting a black-box score.
What The 100,000-Machine Dataset Means In Practice
QCI’s milestone gives participating operators a broader and more current external reference than a property can build alone. Its strongest use is not producing a universal ranking of slot products. It is narrowing the list of questions the operating team should investigate while the information is still relevant.
For the next floor review, management should place local WPUPD, coin-in, games played, theoretical win, actual hold, availability and net contribution beside the 30- and 90-day external benchmarks. Any material deviation should then be traced to volume, configuration, location, cost or variance before action is approved.
The dataset can shorten the time between a performance change and a management investigation. It should not shorten the evidence required for removing a machine, changing a paytable or committing capital.
Source note: Product features and dataset figures are based on QCI’s 5 November 2026 announcement and its earlier next-day performance-data release. Operational interpretations and recommended controls are independent analysis.