REPORT · INDUSTRY RESEARCH
State of the Operating Stack · 2026
A year-long study of 412 mid-market companies and how their operating stacks evolved between 2024 and 2026.
- 23Median tools in stack
- 14Tabs open at 3pm
- +38% YoYConsolidation rate
- 4.6hReconciliation time
ABSTRACT
What the data says.
Between January 2025 and February 2026 we surveyed 412 mid-market operators (50-500 seats) across six industries. We measured stack cardinality, operator tab-counts, reconciliation time, and the rate of stack consolidation. The trend is clear: consolidation is no longer a thesis, it is an observed pattern.
METHODOLOGY
How we measured.
Quarterly telemetry from 412 companies that opted into the study, cross-checked with operator interviews every six months. Sample was weighted to match mid-market demographics across US, EU and APAC.
BENCHMARKS · KEY MEASUREMENTS
The receipts.
- Median tools in stack02024 baseline was 27
- Tabs open at 3pm0median operator
- Consolidation rate0% YoY
- Reconciliation time0hper operator per week
VISUALIZATION · INTERACTIVE
See the trend.
Toggle either series to isolate. Hover the chart to inspect values at each datapoint.
FINDINGS · WHAT WE LEARNED
The pattern.
Stack cardinality peaked in Q3 2024 and is now in measurable decline.
Median tools in the mid-market stack fell from 27 in 2024 to 23 in early 2026, and 38% of companies we tracked reported formally sunsetting at least one tool in the preceding twelve months.
Finance leads the consolidation; operations trails.
Of companies that consolidated, 61% started with finance tooling and 22% with revenue. Operations consolidation is happening later in the cycle, usually as a follow-on from finance migrations.
Tab-count is the strongest predictor of dissatisfaction.
Operators reporting >15 tabs open during working hours were 3.4x more likely to describe their stack as “broken or breaking” than operators reporting <8.
AI does not consolidate on its own.
Adding AI layers to an unconsolidated stack reduced satisfaction in 41% of cases. AI effectiveness was tightly correlated with the presence of a shared memory layer — not with model quality.
APPENDIX · SOURCES
Where the numbers came from.
- 01Aixys Council Telemetry, 2024-2026
- 02Aixys Operator Survey, quarterly panel (n=412)
- 03G2 SaaS Spend Index, 2024-2026
- 04Okta Businesses at Work Report, 2025
NEXT STEP