India has built single-window systems at every level: NSWS, PARIVESH, state portals, sector-specific dashboards. Behind each dashboard, approvals depend on each other in sequences the portal never surfaces. A BIS application needs an Indian entity; the entity needs GST; GST needs registered premises; premises in a bonded configuration need customs approval. Each authority processes its own file without flagging the cross-dependency. Does a single window mean a single approval, or is the government operating a tracking layer over the same fragmented machinery?
India has, over the past decade, built an extensive architecture of single-window systems to streamline regulatory access. At the central level, platforms such as NSWS, PARIVESH (Pro-Active and Responsive Facilitation by Interactive and Virtuous Environmental Single Window Hub) for environmental and forest clearances, the Directorate General of Foreign Trade (DGFT) portal, ICEGATE (Indian Customs Electronic Gateway) for customs, and MCA21 for corporate filings are designed to provide a unified interface for approvals and compliance. These platforms have meaningfully reduced the friction of access; identifying applicable approvals, standardising submissions, and improving visibility across departments.
What they have not changed is the institutional architecture beneath them. A window becomes single through statutory delegation of approval authority to a single body, not through aggregation of departmental interfaces onto a single platform. The distinction determines whether a single-window portal accelerates approvals or adds a tracking layer over the same fragmented machinery.
Most states have adopted what is formally termed "reverse integration" with NSWS. The application is submitted through the national portal, but it is immediately routed to the state industry department's own system, the state pollution control board's own portal, the labour department's own queue. NSWS becomes a tracking layer, not a processing layer. The application enters and is recorded; but the moment it crosses into the state machinery, it moves at the state's pace, through the state's processes, subject to the state's officer bandwidth. The single window opens onto a set of doors that each operate independently.
The circular dependency a market entrant first encounters illustrates how this plays out at the entry stage. BIS requires entity and GST linkage. GST requires registered premises. Premises in a bonded configuration require customs approval. Customs expects operational readiness that depends on the registrations still in progress. Each authority processes its own file correctly; no authority flags the cross-dependency. The company discovers the loop only when an application is returned with a query that references a credential it has not yet obtained, because that credential depends on a different authority that is waiting for something else. The result, in cases observed repeatedly across sectors, is a four-to-six month delay, multiple re-filings, and a forced workaround; typically a third-party importer arrangement or non-bonded warehousing at higher cost, neither of which was part of the original market entry design.
Six institutional realities operate beneath the single-window surface, and each explains why aggregation at the platform layer does not translate to integration at the decision layer.
First, the correct sequencing of regulatory approvals is not documented anywhere. It exists only as institutional memory, held by those who have navigated the system repeatedly. A BIS application cannot progress without an Indian entity; the entity cannot operationalise without GST; GST cannot complete without premises. Each step has an implicit predecessor the system does not make explicit.
Second, what appears parallel on a project plan is sequential in institutional reality. NSWS shows fourteen approvals all marked "applied." The actual system behaves as fourteen approvals of which five are active, four are blocked by prerequisites never surfaced, three carry dependencies invisible on the platform, and two are waiting for an officer whose bandwidth is consumed by the other files.
Third, errors in sequencing are discovered late. The dependency surfaces not at the planning stage but at the point of rejection or query, when the company has already invested months in one track and discovers that a different track needed to have been initiated first.
Fourth, the same officer may handle multiple files sitting in different digital queues within different systems. File B does not move because File A, in another system, is pending with the same officer. The platform tracks applications, not officer bandwidth.
Fifth, offline validation routinely overrides online submission. Site inspections, document clarifications, and procedural queries are conducted through phone calls, physical visits, and local office follow-ups. The digital status remains "under process" while real progress is being negotiated offline.
Sixth, the dashboard lags ground reality. NSWS updates only after a department records an action, and departments update after internal file movement, not during it. A status that shows "in process" may mean the file is being actively scrutinised, or it may mean the file has not been opened.
A manufacturing client applied through NSWS for approximately twelve approvals across state and central layers. All applications were filed within ten days. The dashboard showed "in process" across the board. Pollution consent did not move because the site layout needed a minor revision, communicated informally by the State Pollution Control Board (PCB) officer but not recorded on the platform. The factory licence did not move because the officer processing it could see that pollution consent status was unclear. The power connection file was not taken up because the load requirement depended on the final plant configuration, which depended on the factory licence. None of these dependencies were visible on NSWS. Progress resumed only after direct engagement with the state PCB officer and the industry department desk, clarifying sequencing and enabling movement that the platform, by design, could not facilitate.
As senior officials involved in these platforms themselves acknowledge, single windows sit on top of the underlying administrative processes; they do not replace them.
A second institutional observation emerges from the design of recent industrial schemes. The BHAVYA industrial parks scheme approved by the Cabinet in March 2026 treats devolution of planning, single-window clearance, and development authority to the SPV as a mandatory pre-condition for release of central capital grant. This is the architecture under which NICDC's earlier industrial corridor townships (Dholera Special Investment Region, Shendra-Bidkin Industrial Area, Vikram Udyogpuri, Greater Noida) operate. The institutional logic the scheme adopts is that single-window systems as typically operated function as one more window over the existing clearance windows rather than replacing them, and that the institutional solution is for states to devolve administrative, regulatory, and legal authority to a single body. The BHAVYA scheme makes this devolution a funding pre-condition, not a reform aspiration. The limit of this model is its domain: industrial township SPVs cover only the geography of the notified park, and the layered-window problem persists for every regulatory matter that sits outside a notified township jurisdiction.
PM Gati Shakti, the National Master Plan for Multi-modal Connectivity launched in October 2021, is the inter-ministerial layer the central government built specifically to address the coordination problem that single-window portals could not solve at the data layer. The platform integrates geospatial data from over forty central ministries and departments and from state governments onto a single GIS-based interface, permitting any ministry planning an infrastructure project to view the land, utilities, environmental constraints, existing infrastructure, and other ministries' planned projects in the same coordinate system. The institutional architecture beneath the platform is the Network Planning Group, chaired by a Special Secretary in the Department for Promotion of Industry and Internal Trade, which convenes regularly to examine infrastructure proposals above defined thresholds and to surface the inter-ministerial dependencies the platform has made visible. A proposal that would have previously been designed by one ministry, approved by its own appraisal architecture, and then discovered at the execution stage to conflict with another ministry's planned alignment is now examined in the NPG room before approval, and the conflict is resolved before capital is committed. For manufacturing investors, the platform offers a specific planning utility: land identification, logistics-cost mapping, and port-connectivity assessment across the national footprint through a single data layer. The institutional limit is the same limit that constrains most coordination reforms: the platform integrates the data; it does not integrate the approval authority. A state-level clearance, a municipal permission, or a sectoral regulator's licence still moves through the same institutional machinery that existed before Gati Shakti, and the platform's visibility into the dependency does not by itself compel the dependency to resolve. Gati Shakti has demonstrably reduced the inter-ministerial data asymmetry that used to drive the worst coordination failures; whether it becomes, over time, another layer that investors must navigate rather than the coordination instrument that resolves other layers is the question the next five years will settle.
Single-window systems have digitised entry. They have not integrated decision-making. They show what approvals are required, but not how the system will actually process them. What remains invisible on every dashboard, and critical to every outcome, is sequencing, officer bandwidth, and inter-departmental dependency. The platform aggregates approvals. The system still executes them independently. And the distance between the two is where regulatory timelines are actually determined. The regulatory system is not horizontally integrated; it is a set of vertically coherent silos. That accountability for the end-to-end outcome of a market entry must be constructed from outside the system, by whoever is managing the regulatory journey. The company that treats the single-window portal as the entire process will be surprised. The company that treats it as the front door and prepares for what lies behind it will not.