Let me share something that might sound counterintuitive: the biggest obstacle to AI adoption in ASEAN SMEs is not technology, budget, or talent. It is data. Or more precisely, the lack of clean, structured, accessible data that AI needs to deliver value. I have lost count of the number of business owners who have asked me about implementing AI-powered tools, only for us to discover that their most critical business data lives in a patchwork of Excel files, WhatsApp messages, handwritten logs, and the memories of long-serving staff members.
During a consulting engagement with a mid-sized Malaysian manufacturing company last year, the CEO showed me an impressive AI-powered demand forecasting tool he wanted to implement. When I asked to see their historical sales data, his team produced seventeen separate Excel files across three departments, each with different column headers, inconsistent date formats, and several months of missing entries. The AI tool was world-class; the data to feed it was nowhere near ready. We spent the next four months not implementing AI but building the data foundation that would eventually make AI possible — and useful.
This is not a failure story. It is the reality of the data maturity journey, and understanding where your business sits on that journey is the most important strategic decision you can make before investing in any advanced technology. I have worked with businesses across Southeast Asia — from Singapore to Vietnam to the Philippines — and the ones that get the best returns from technology are invariably the ones that invested in their data infrastructure first. The journey from Excel to AI is not a leap; it is a staircase, and every step delivers value on its own.
Why SMEs Struggle
The first challenge is what I call ‘data fragmentation.’ Most SMEs have data scattered across multiple tools, formats, and people. Sales data sits in one system, customer communications in another, financial records in a third, and operational metrics in someone’s personal spreadsheet. Nobody has a complete picture of the business because the pieces of the puzzle are in different rooms. This fragmentation is not a technology problem — it is an organisational problem that technology alone cannot solve.
The second issue is ‘data debt’ — years of inconsistent recording practices that create gaps, errors, and contradictions in your historical data. If your team has been entering customer names in ten different formats, if product codes have changed three times without updating old records, if your sales figures are recorded in different currencies depending on who entered them — you have significant data debt that needs to be addressed before any advanced tool can deliver reliable results.
The third barrier is the misconception that data maturity requires a data scientist. It does not. It requires discipline, consistency, and someone in your organisation who takes ownership of data quality as part of their role. I have seen businesses transform their data practices with nothing more sophisticated than a shared Google Sheet and a clear set of naming conventions.
Step-by-Step Roadmap
Step 1: Assess Your Current Data Maturity
Be honest about where you are. I use a simple five-level maturity model with the businesses I consult for. Level 1: data is mostly in people’s heads and paper records. Level 2: data is in spreadsheets but scattered and inconsistent. Level 3: data is in cloud-based tools with some consistency and sharing. Level 4: data from multiple systems is integrated, with dashboards and regular reporting. Level 5: data is clean, integrated, and actively used for predictive analytics or AI. Most ASEAN SMEs I work with are at Level 2, which is absolutely fine — it is a starting point, not a shortcoming. For the Malaysian manufacturer, the assessment revealed they were at Level 2 for sales data, Level 1 for production data (still on paper logs), and Level 3 for financial data (thanks to their cloud accounting software). This assessment told us exactly where to focus first.
Step 2: Consolidate Into a Single Source of Truth
Pick your most important data set — usually sales or customer data — and consolidate it into a single, shared location. This might be a well-structured Google Sheet, a CRM like HubSpot (free tier), or a cloud database. The key principle is that there should be one version of the truth, accessible to everyone who needs it, with clear rules about how data is entered. Define your conventions: how are customer names formatted? What date format do you use? What are your product categories? Document these rules in a simple one-page data dictionary and share it with everyone who enters data. For a Vietnamese logistics company I worked with, consolidation meant migrating three years of shipment records from fourteen separate Excel files into a single Google Sheet with consistent column headers. It took two weeks of effort from one team member, but the result was the first time the company had ever been able to analyse shipping patterns across their entire operation.
Step 3: Automate Data Collection
Once your data is consolidated, start eliminating manual data entry wherever possible. Every time a human types information into a system, there is a risk of error, delay, and inconsistency. Look for opportunities to capture data automatically: web forms that feed directly into your CRM, point-of-sale systems that sync with your inventory, time-tracking apps that populate your payroll system. Even simple automations — like using Zapier to log every WhatsApp Business enquiry into a spreadsheet automatically — can dramatically improve data quality and completeness. The Penang physiotherapy clinics from an earlier article automated their appointment data by replacing the paper diary with a cloud booking system that captured patient details, appointment types, and no-show records without any manual entry. The data quality improvement was immediate and dramatic. Within three months, they had a clean dataset that revealed patterns they had never seen before — including which appointment slots had the highest no-show rates and which services were most frequently rebooked.
Step 4: Build Simple Dashboards
With clean, consolidated, automatically collected data, you can now build dashboards that give you real-time visibility into your business. You do not need expensive business intelligence software. Google Sheets with pivot tables and charts, or free tools like Google Data Studio (now Looker Studio), can produce dashboards that would have required a dedicated analyst a decade ago. Start with three to five key metrics that your leadership team should see weekly: revenue trends, customer acquisition costs, order volumes, or whatever matters most in your business. The key is to make data visible and habitual. When your team sees the numbers every week, they start asking better questions and making better decisions. I built a simple dashboard for the Johor logistics company using Google Sheets that showed daily delivery volumes, on-time rates, and fuel costs. The operations manager told me it was the first time he could see his entire operation at a glance without waiting for the monthly Excel report his finance team produced.
Step 5: Introduce Basic Analytics
Once you have clean data and regular reporting, you are ready to move beyond descriptive analytics (what happened) to diagnostic analytics (why it happened). This is where you start identifying correlations, trends, and anomalies that inform strategic decisions. You do not need AI for this — pivot tables, conditional formatting, and simple statistical functions in spreadsheets can reveal powerful insights. For the Malaysian manufacturer, basic analytics on their consolidated sales data revealed that 35% of their revenue came from just 8% of their product catalogue, while 40% of their products generated less than 2% of revenue. This insight, which had been invisible when the data was scattered across seventeen files, led to a product rationalisation exercise that reduced complexity and improved margins. At this stage, you might also explore simple regression analysis or forecasting using spreadsheet tools — nothing fancy, just enough to start making data-informed predictions about demand, staffing, or inventory needs.
Step 6: Evaluate AI Readiness and Start Small
Only now — with clean data, integrated systems, regular reporting, and basic analytics in place — should you seriously consider AI-powered tools. And when you do, start with the most constrained, well-defined use case you can find. AI excels when given a specific task with clean, abundant data: predicting which customers are likely to churn, recommending inventory reorder quantities, automating responses to common customer enquiries, or classifying incoming support tickets. Do not start with a grand AI strategy; start with a single AI experiment. I helped a Singaporean e-commerce company implement a simple AI-powered product recommendation engine after six months of data cleanup and consolidation. Because their data was clean and well-structured, the recommendation engine was effective from day one, increasing average order value by 12%. Had they tried to implement the same tool six months earlier, before the data work, it would have produced irrelevant recommendations and eroded customer trust. The data maturity journey is the foundation that makes AI worthwhile.
Before and After: A Real Example
A chain of traditional Chinese medicine (TCM) clinics in Penang and Kuala Lumpur with five branches had been operating for twenty years using paper-based patient records, handwritten prescriptions, and a basic Excel file for monthly revenue tracking. The founder, a second-generation practitioner, wanted to modernise but was overwhelmed by the options. Every technology vendor he spoke to wanted to sell him a comprehensive clinic management system costing RM 100,000 or more. He came to me asking whether AI could help him predict patient demand patterns. I told him we needed to walk before we could run.
We started at Level 1 and built systematically. Month one: we digitised the patient intake process using a tablet-based form at each branch’s reception, feeding into a central Google Sheet. Month two: we introduced a simple cloud-based appointment system that replaced the paper diary and automatically logged appointment data. Month three: we connected the appointment data to a basic dashboard in Looker Studio that showed patient volumes by branch, day, and treatment type. Month four: we added practitioner utilisation tracking, showing which practitioners were overbooked and which had available capacity. By month five, the founder had something he had never had in twenty years of practice: a complete, real-time view of his business across all five branches.
The insights were immediate. The data revealed that two branches were consistently underperforming on Wednesdays — a pattern nobody had noticed because each branch manager only saw their own paper diary. Cross-referencing with local events, they discovered that a popular farmers’ market operated on Wednesday mornings near both locations, drawing their typical demographic away. They shifted their Wednesday opening hours to the afternoon and introduced a ‘market day’ promotion. Wednesday revenue at those branches increased by 45% within two months. Today, twelve months into the journey, the clinics are at data maturity Level 4 and are genuinely ready for AI-powered demand forecasting — not because they bought expensive software, but because they built the data foundation step by step. Total investment over twelve months: approximately RM 8,000, including all software subscriptions and the tablets for patient intake.
Quick Wins to Start Today
Win 1: Create a One-Page Data Dictionary
Document how your team should record the five most important data fields in your business — customer names, product codes, dates, currencies, and categories. Agree on consistent formats and share the document with everyone who enters data. This fifteen-minute exercise prevents months of cleanup later.
Win 2: Consolidate Your Most Important Spreadsheet
If you have multiple Excel files tracking the same type of data, merge them into one shared Google Sheet this week. Standardise the column headers, fix the date formats, and remove duplicates. This single act of consolidation will reveal insights that fragmented data kept hidden.
Win 3: Automate One Data Entry Task
Identify one piece of data that someone manually types into a spreadsheet and find a way to capture it automatically — a form submission, a tool integration, or a Zapier automation. Eliminating one manual entry point improves data quality and frees your team’s time for higher-value work.
Key Takeaway
The journey from Excel to AI is a staircase, not a leap. Every step — consolidation, automation, dashboards, analytics — delivers value on its own while building the foundation for what comes next. Do not skip steps, and do not let anyone sell you an AI solution before your data is ready to support it.
If you want help assessing your data maturity and building a practical roadmap, explore the Business Data Collection & Integration workshop at Being Specific. We guide ASEAN SMEs through every stage of the data journey, from consolidation to AI readiness. Visit being-specific.com/contact to begin.
Ready to Transform Your Business?
If your team is talking about AI but still living in Excel, the journey from one to the other is more achievable than it looks. Our Business Data Collection & Integration workshop at Being Specific helps ASEAN businesses build a clean data foundation, then layer analytics and AI on top in a sensible, sequenced way. Visit being-specific.com/contact to find out more.
