There’s no shortage of conversation about technology in accounting - cloud platforms, automation, and now AI. But beneath the buzzwords, one truth holds: technology is only as good as the data that powers it.
As a cloud-native firm, ConnectCPA’s day-to-day operations are built on a fully digital stack. That gives us incredible flexibility - but also places an even greater premium on data integrity, consistency, and architecture. Without that foundation, even the best automation or report risks introducing more confusion than clarity.
Over the past five years, we’ve consistently iterated on our approach to get it right. In this article, I’ll share lessons from our own data and automation journey - how we built a scalable stack, matured our workflows, and are now approaching AI with both excitement and caution.
One Step Back, Two Steps Forward
When co-founders Mike and Lior launched ConnectCPA, the first CRM/job management system that was implemented back in 2016 was Podio. That system remained in place for nearly a decade as the company scaled from 10 to over 130 employees. But as our client base grew in complexity - from multi-entity relationships to specialized edge cases - the limitations of our early data model became clear. Over time, new apps and fields were added within Podio without a unified data strategy, and legacy assets were never properly decommissioned.
The decision to take a step back - conduct a comprehensive evaluation of our current data structure - and remodel it was deemed a necessary step for the company to continue to move forward efficiently. Ultimately, we built a new data model from the ground up, informed by nearly eight years of learnings in Podio. This new structure was deployed in Airtable before migrating all client data.
Good Data In; Good Data Out
With dozens of applications in play, it’s critical to establish a source of truth - a single system where core data is maintained and from which updates flow to the rest of the tech stack. For many firms, that’s a CRM or customer database. At ConnectCPA, that system is now Airtable.
For example, when a client’s name changes, the update is made in Airtable. From there, an automated workflow syncs the change across all connected platforms - including Clockify, ClickUp, Google Drive, Slack, and key Google Sheets - to maintain data integrity.
By maintaining consistent and accurate entity names across all platforms, we reduce the risk of errors, avoid integration failures, and unlock reliable automation at scale. In today’s landscape, building a disciplined approach to data management isn’t optional - it’s foundational.
Tooling Maturity
Over the past five years, ConnectCPA’s technical capabilities have evolved dramatically - a reflection of our broader shift to embrace technology not just for efficiency, but for strategic advantage.
Our data stack began small: a few thousand Zapier actions per month and high-level Klipfolio dashboards connected directly to connected platform APIs.
As we encountered limitations in this initial tooling, we progressively expanded our approach. The constraints of purely no-code workflows in Zapier led us to introduce custom scripting. These Python scripts, deployed as AWS Lambda functions, now run on daily, weekly, and monthly cadences - automating processes we never imagined possible.
One example of this is how we manage deferred revenue for Shopify clients. To ensure accuracy, we run an automated job at midnight (store-local time) on the first day of each month. This pulls all unfulfilled orders as of month-end into a structured Google Sheet. By capturing this point-in-time snapshot automatically, we eliminate the need for bookkeepers to manually extract and manipulate the data - a process that not only consumes time but risks inaccuracies, as orders unfulfilled at month-end could later be fulfilled before the manual data extract, distorting the reporting.
As our data volume grew, API-driven reporting began to break down under the weight of scale and complexity. This resulted in the deployment of our first data warehouse in Google BigQuery. Centralizing our data enabled us to query, blend and analyze information more flexibly, opening up new avenues for insight and reporting.
Of course, not all tools play nicely. Even mature products have API limitations. This resulted in us hitting unavoidable roadblocks in automation implementations. In response, we adopted Robotic Process Automation (RPA) in late 2022. Since then, our RPA lead, Vincent, has developed and deployed more than three dozen bots. In May 2025 alone, these bots collectively ran an average of 19 hours per day - the equivalent of eight full-time team members.
With this evolved toolkit in place, every new automation project undergoes a deliberate evaluation: which tool - no-code, code-based automation, or RPA - is best suited to the task? And when it comes to presenting data, we ask: does the end user need a visual dashboard (Klipfolio) or a flexible spreadsheet they can manipulate (Google Sheets)?
The journey hasn’t been linear, but that’s precisely the point. Building a robust data toolkit is an iterative process. The real differentiator is mindset: continuous improvement, and a willingness to adopt the right tool at the right time.
Workflow Maturity
As firms scale, so does client complexity. With that comes the need for equally sophisticated internal workflows that can support deeper client relationships without compromising efficiency or consistency.
One of my favourite examples of this at ConnectCPA is our month-end bookkeeping workflow which has undergone four major iterations over the past five years. What began as a simple five-step Zapier automation triggered by a ClickUp task - notifying clients that their books were closed - has grown into a robust system of five integrated workflows, totalling over 230 steps and built to handle diverse client scenarios.
These flows now dynamically determine whether a review is required, whether communications should be automated or personalized, whether controllership services are active, and whether a sales tax filing is due. Based on these inputs, the relevant team members are notified - managers or controllers for reviews, tax associates for filings, and clients when month-end is complete.
Behind the scenes, this orchestration moves data seamlessly between ClickUp, Airtable, Google Sheets, Google Drive, Clockify, with Slack alerts to dedicate internal client-specific channels and emails via Gmail for notifications to clients. It only works because of a shared commitment to clean, accurate, well-structured data.
AI
It would be impossible to write about technology without addressing the obvious: AI. For firms like ours, it’s no longer a question of if - it’s when. We’re actively developing an AI strategy grounded in practical use cases and governed by strict data security and privacy protocols. Our goal is to deliver tangible value - internally and for clients - while avoiding the pitfalls of poorly integrated, half-baked solutions.
That means asking the right questions now: Will our current tooling adequately support AI-powered workflows, or will we need to introduce new AI-specific platforms to close the gap? Which processes are ripe for intelligent automation - and which require a human-in-the-loop for context and judgment? We’re starting small, experimenting deliberately, and focusing on AI applications that enhance - not replace - our team’s capabilities.
Moving Forward
At ConnectCPA, our journey with data, automation, and technology has never been about chasing trends - it’s been about building durable systems that scale with intention. From rebuilding our data model from scratch, to expanding our automation toolkit, to introducing AI cautiously and deliberately, every decision has been grounded in a simple principle: great outcomes start with great data.
The road ahead will undoubtedly bring new tools, new capabilities, and new expectations - from clients, from our team, and from the broader industry. AI will accelerate the pace of change, but it will also raise the bar for accuracy, clarity, and trust. To meet that moment, we believe the firms that thrive will be the ones that stay grounded: in their processes, in their data discipline, and in their willingness to continuously evolve.
We’re not done - not even close. But the foundation we’ve laid gives us confidence in where we’re going: a future where automation is smart, workflows are seamless, and data isn’t just a byproduct of work - it’s the engine that drives it forward.
