CII-ISB Study: 65.8 Percent Use AI, Only 21 Percent Run ERP Software
Posted on 20th Oct 2026 12:07:38 in Business, Digital Marketing
Tagged as: MSME, CII, ISB, artificial intelligence, ERP, small business, technology adoption
Indian small businesses have taken to artificial intelligence faster than almost anyone expected. What they have not done, according to a new national study, is put the basic business systems in place that would let that intelligence do any real work. The result is a country of MSMEs where two out of three owners use AI tools, yet only one in five runs the kind of integrated software that turns daily transactions into data a machine can act on.
The finding comes from a joint study by the Confederation of Indian Industry (CII) and the Indian School of Business (ISB), titled MSMEs in India: Challenges, Opportunities and the Road Ahead. The survey covers 2,919 confirmed MSMEs spread across all 36 states and Union Territories and 40 industry sectors, making it one of the broadest snapshots of small-business technology use published this year. The headline numbers, first reported by Business Standard on October 13, describe a gap that every small business owner in India should recognise in their own operation.
What the CII-ISB Study Found
The core comparison is stark. Of the enterprises surveyed, 65.8 percent said they use AI tools in some form. Only 21 percent said they use enterprise resource planning (ERP) software or customised business software to manage operations. In other words, roughly two-thirds of the sample has adopted the most advanced layer of technology available to a small business, while about four-fifths still run the underlying business on disconnected records.
The study also mapped what MSMEs actually do with AI. The most common application is business communication, cited by 34.9 percent of respondents, followed by content creation at 22.3 percent and training at 18.9 percent. Those are useful, low-risk use cases: drafting a quotation, replying to a customer enquiry, producing a product description, or building a short internal training module. None of them, however, touches the operational core of a business, which is where orders, inventory, procurement and payments meet.
On government support, the study found that 74.9 percent of surveyed MSMEs were aware of at least one government scheme, but only 42.6 percent had used one. Firms with recognised certifications, external credit ratings and higher turnover were more likely to have accessed support, which suggests that the smallest businesses — the ones that need help most — face the highest administrative barriers.
Why AI Without a System of Record Stalls
A demand forecast is only as good as the sales history behind it. A cash-flow warning is only as good as the invoices feeding it. This is the practical reason the adoption gap matters. A garment manufacturer that wants AI to plan fabric purchases needs reliable past order data. A food-processing unit that wants to spot recurring wastage needs production records. A wholesaler that wants to identify slow-moving stock needs sales and inventory data that can be compared.
When those records live in a mix of notebooks, spreadsheets, WhatsApp threads and a billing app that does not talk to anything else, the AI tool has nothing to reason over. It will still produce confident answers, which is precisely the danger. The vendor-side view of this problem is blunt. Abhijeet Kate, co-founder and principal consultant at digiCloud Solutions, told Fortune India that the biggest barrier is rarely the AI itself — it is the state of the business around it.
Independent data points reinforce the same picture. Dun & Bradstreet India's 2026 MSME study, covering 210 MSME leaders, found that 76 percent used AI, but only 5 percent had made it central to their operations, and 84 percent relied on general-purpose AI assistants rather than integrated systems. Separately, PHDCCI cited an October 2026 Adobe report showing that just 6.7 percent of Indian enterprises had scaled AI across multiple workflows, while 40 percent remained in the pilot phase for agentic AI.
The enthusiasm is real; the plumbing is missing. PHDCCI also cited a July 2026 Google–India MSME Forum report projecting that AI adoption could unlock more than $490 billion in economic value for Indian MSMEs, with early deployments reporting operational cost reductions of 20 to 30 percent. Those gains are achievable, but they arrive through specific workflows, not through a chatbot in isolation.
What the Gap Looks Like Inside a Small Business
Consider how a typical order travels through an MSME today. An enquiry arrives on WhatsApp. The owner quotes a price from memory or a spreadsheet. The order is written into a register. Raw material is bought on credit. Production is tracked on the shop floor by whoever is supervising. The invoice is raised in a billing tool. Payment arrives weeks later, and reconciliation happens when the owner has time.
Each step is individually manageable. Together they create five or six versions of the same truth. When the owner asks an AI assistant how much margin the order made, the honest answer is that nobody knows, because the cost, the invoice and the payment never sit in the same place. Integrated business software exists to fix exactly this — it does not have to be an expensive ERP. Digital bookkeeping, invoicing and inventory management are enough to create the basic records that more advanced tools require.
There is also a measurement problem. Vishal Devanath, co-founder and CEO of SMERGERS, told Fortune India that the share of Indian business listings mentioning AI on its platform rose from 2.3 percent in 2023 to 4.3 percent in 2026, and cited an embroidery software firm that cut design time from 24 hours to 15 minutes. That is a genuine, measurable gain inside one workflow. It is very different from "we use AI" as a statement of intent.
A Practical Starting Sequence for MSME Owners
- Fix one workflow first. Pick the process that hurts most — usually order-to-invoice or purchase-to-payment — and get it into a single system before adding any AI layer.
- Start with records, not platforms. Digital invoicing, bookkeeping and inventory tracking create the data foundation. An ERP can come later; a clean sales and purchase ledger is the entry ticket.
- Put the AI on top of connected data. Once orders, stock and payments live in one place, AI can forecast demand, flag overdue receivables and plan purchases against expected receipts.
- Measure outcomes, not usage. Track error rates, order turnaround time, stock-outs, receivable days and revenue per employee. If none of these move, the tool is not working.
- Keep an eye on the working-capital wall. Better planning cannot finance a large order. The Economic Survey 2025-26 estimated roughly Rs 8.1 lakh crore locked in delayed payments to MSMEs, so cash-flow discipline has to run in parallel with automation.
The sequencing matters because the reverse order is where most small businesses lose money. Buying software before fixing the underlying process, as the Policy Circle analysis of the study notes, is unlikely to deliver much benefit. The same applies to AI subscriptions bought by individual employees and never connected to a shared record.
What the Report Recommends
The CII-ISB study argues that the next phase of policy support should focus on converting technology adoption into measurable improvements in productivity, quality and business growth, rather than on adoption counts alone. It proposes five national missions: liquidity and payment discipline, an integrated digital journey for MSMEs, technology adoption with implementation support, skills development, and market access, quality and green competitiveness.
One concrete proposal is a unified digital platform that would let an enterprise discover relevant schemes, check eligibility, submit applications, track approvals and resolve grievances through a single interface. That directly addresses the 74.9 percent awareness versus 42.6 percent usage gap. India had reported 9.77 crore registrations across the Udyam and Udyam Assist platforms as of October 6, 2026, so the administrative reach exists; the question is whether the smallest firms can navigate the conditions attached to support.
Implementation support is the part business owners should watch most closely. Training programmes and subsidised licences have a poor record when they stop at the point of purchase. The study's own logic points to a different model: help the firm connect its records, then help it use the resulting information, and judge the programme by shorter payment cycles, lower production costs, fewer defects and quicker approvals.
The Takeaway for Indian SMEs
The CII-ISB numbers should be read as good news with a warning attached. Two-thirds of Indian MSMEs are already experimenting with AI, which is a far higher base of familiarity than most policy debates assume. But experimentation is not the same as integration, and a business that uses AI for letters and social posts while its stock, orders and receivables stay scattered across registers has captured only the smallest part of the available gain.
The report's own caveat is worth stating plainly: participation in the survey was voluntary and the sample was not probability-weighted, so the findings describe the 2,919 firms surveyed rather than every MSME in the country. Even with that limitation, the direction of the gap is consistent across three independent studies — CII-ISB, Dun & Bradstreet and the Adobe data cited by PHDCCI. Broad use, shallow integration.
For an owner deciding what to do next quarter, the sequence is unglamorous and effective: get orders, stock and payments into one system, clean up the records, then point AI at the questions that decide whether the business makes money. The technology is no longer the hard part. The business around it is.
