Vendors publish adoption numbers. Nobody publishes churn. But when a company hires an engineer to "migrate our analytics stack from Power BI to a GCP-based platform", the churn is right there in the posting — funded, dated, public. We read 4.7 million active postings and pulled out every explicit exit.
One thing makes this index different from a survey: we verified the direction of every signal against raw posting text, technology by technology. Extractors love to confuse "migrating to X" with "replacing X" — so seven technologies that looked like they belonged here were thrown out (the full casualty list is in the methodology). What's left are 19 technologies with clean, explicit exit language and 479 companies actively hiring their way off them.
Displacement is accelerating — 1 in 1,000 postings is now an exit req
The replacement signal is the rarest and most valuable thing we extract: a posting that names the technology being killed. Its share of all hiring has climbed all year and hit 100 per 100K postings in June — the highest we've measured. 84% of the signals behind this report are less than 90 days old.
June 2026: exactly 1 in 1,000 new job postings describes ripping a technology out — up ~40% from autumn 2025. Displacement is accelerating even where individual migrations (like SAP's) have plateaued: the churn is spreading across more categories at once.
Who's leaving what: banks quit SAS, SMBs outgrow QuickBooks, everyone modernizes .NET
The index has personalities. SAS is a banking exodus: Citi · PNC · Danske Bank · Banque de France are all hiring to move statistical workloads to Python and Databricks — 12 of its 21 leavers are 1,000+ companies. QuickBooks is the opposite: 24 of 25 leavers are under 1,000 employees — it's not churn, it's graduation, mostly toward NetSuite.
Jenkins is quietly bleeding enterprises (State Farm · ABN AMRO · Suncorp Group), VMware's post-acquisition exodus is real (Ericsson · Dassault Systèmes · NetApp — toward Kubernetes and Nutanix), and even Salesforce shows 32 companies heading for the door, including Booking.com · Veeva Systems · UiPath — with no consensus destination, which is its own story. Oracle's 56 leavers include Siemens · RBC · Salesforce. And Microsoft Access refuses to die quietly: Metropolitan Transportation Authority is still hiring people to finally kill it.
| Leaving | Category | Companies | Enterprise | Common destination |
|---|---|---|---|---|
| SAP ECC | ERP | 81 | 64% | SAP S/4HANA |
| SQL Server | Databases | 60 | 38% | Databricks / Fabric |
| Oracle | Databases | 56 | 50% | AWS / PostgreSQL |
| Salesforce | CRM | 32 | 34% | — |
| VMware | Infrastructure | 25 | 44% | Kubernetes / Nutanix |
| QuickBooks | ERP | 25 | 4% | NetSuite / Brex |
| Angular | Frameworks | 23 | 35% | React |
| SAS | Analytics | 21 | 57% | Python / Databricks |
| SAP R/3 | ERP | 21 | 52% | SAP S/4HANA |
| .NET Framework | Frameworks | 20 | 35% | .NET (modern) |
| .NET | Frameworks | 20 | 35% | — |
| Java | Languages | 20 | 35% | — |
| Jenkins | DevOps | 18 | 56% | GitLab / Terraform |
| Microsoft Access | Databases | 17 | 47% | — |
| Tableau | Analytics | 16 | 38% | — |
| Informatica | Analytics | 15 | 67% | Power BI / Databricks |
| Dynamics NAV | ERP | 15 | 7% | — |
| PHP | Languages | 14 | 29% | — |
| AS/400 | Legacy platforms | 13 | 38% | — |
Where they're going: cloud-native catches the refugees, AI rides along
Displacement money doesn't disappear, it lands somewhere. Among companies actively leaving an indexed technology, the most-adopted destinations are Kubernetes, the hyperscalers, Databricks and Python — the standard modernization playbook. The surprise is what's riding along: RAG (14 companies), LangChain (12) and Cursor (10) all rank among the top technologies leavers adopt. Companies don't just swap old for new — the replatform budget doubles as the AI budget.
| Technology | Raw signals | Why it's out |
|---|---|---|
| SAP S/4HANA | 58 | direction inversion: postings describe migrations TO S/4HANA (1 away-signal vs 42 toward in raw text) |
| Dynamics 365 | 13 | direction inversion: destination of NAV/AX migrations |
| PostgreSQL | 17 | direction inversion: mostly the destination of Oracle exits |
| Snowflake | 18 | contaminated: vendor pitch postings and migrations TO Snowflake |
| AWS / Azure | 64 | on-prem-to-cloud postings mislabel the destination; real cloud-to-cloud churn kept separately below |
| SAP (generic) | 40 | ambiguous: overlaps ECC/R/3 rows and vendor-neutral consulting language |
| SSIS | 12 | implied-deprecation language only; no explicit replacement statements in sampled postings |
Every row of this index is a company list
The free CSV has the 50 largest companies with a verified exit signal — who they are, what they're leaving, what they're adopting, plus hiring context. Each individual row of the index ("companies leaving VMware", "companies leaving SAS"…) is available as a full list with quarterly refresh — that's the product. Custom pull against your account list is free.
How this was measured
Echoloc continuously indexes public job postings; at analysis time: 15,207,852 postings, 4,747,583 active, across 761,828 companies. An LLM pipeline extracts technologies with context (using / adopting / evaluating / replacing). Replacement signals aggregate to company level from active postings only.
- Direction verification: for every candidate technology we ran a lexical screen of raw posting text ("migrating from X", "off X", "replace X" vs "to X") plus manual reading of sampled postings. Seven technologies failed and were excluded — see the table above. This matters: naive extraction ranks S/4HANA as the #3 "dying" technology when it is in fact the #1 destination.
- Excluded from all counts: consulting, IT-services, staffing and outsourcing industries (10 categories), named professional-services firms, and vendor self-rows (a vendor "replacing" its own product).
- Destination pairs are co-occurrence: the same company shows an exit signal for X and an adoption signal for Y. Shown only at ≥3 companies. Absence of a destination means no consensus, not no movement.
- Momentum uses shares of postings per month; months before continuous indexing (Jan 2026) are survivorship-biased toward long-lived postings.
- Counts are companies, not postings, and are conservative: a company migrating without public hiring is invisible to us.
Questions, or a number you'd like re-cut? [email protected].