Part 4: Automated Reconciliation — The Dream, The Reality, and The Lessons Learned

By now, we’d built a solid fortress: manual tangles were tamed, data was chunked and sanitised, servers stopped melting, and order statuses marched in line. The next natural step? Automation. Surely, the future is here, and machines should do all the heavy lifting, right?

Spoiler alert: it wasn’t that simple.

Why We Didn’t Go Fully Automatic (Yet)

Automation sounds dreamy, but reality bites:

Brands did not have a streamlined data pipeline at their end.
A lot of orders were handled manually, which resulted in incorrect data on those orders. Hence, manual interventions were still necessary.
Constant changes on the brand side demanded ongoing dev work.
One famous brand managed reconciliation entirely in Excel sheets. Yes, Excel. Welcome to the wild west.
Tools like Shopify (yes, Shopify still doesn’t provide an option to place returns!) and Shiprocket lacked native support for returns processing, adding to the complexity.
On top of this, getting brands (who were not as tech enabled as us) to expose APIs, or file sharing mechanisms was a big task. We did integrate with 5 brands, but since those also required a similar amount of manual bandwidth as the non-automated brands, we decided to pause this project for the time being.

The Dream: Real-Time Reconciliation Through APIs

Many brands we worked with used platforms like Shopify, which provide APIs to fetch order details. Our hope was to tap into these APIs, get near-real-time data, and automate reconciliation completely.

Imagine a world where:

We pull order data directly from brands’ systems.
No more waiting for month-end uploads.
Creators see their earnings updated in real time.
Finance gets clean, timely data without manual intervention.

Sounds dreamy? Absolutely.

Reality Check: The Brand Integration Struggle

We started talking to 15+ brands, mostly Shopify-based, eager to integrate and automate. What we found was eye-opening:

No standardization: Each brand customized Shopify or their order management systems heavily. APIs returned different data, formats, and sometimes incomplete info.
No single system: Not all brands kept returns data in Shopify. Almost always, they used a third party to manage returns because Shopify doesn’t provide a way to do so.
Manual interventions everywhere: Despite third parties and APIs, many brands manually tweaked orders, returns, and commissions outside the system.
Tech team availability: Some brands had no dedicated tech team, making integration slow or impossible.
Communication gaps: Misunderstandings about data fields, timing, and expectations led to repeated back-and-forths.
Timing mismatches: We’d be reconciling June’s data in July, but May’s data might still be trickling in from the brand’s side. Matching data sets across months was a nightmare.
No API is perfect: Even Shopify and similar tools lacked features like returns processing or standardized commission reporting.

The Google Sheets Saga

Here’s a funny (and sad) story: one “major” brand did their entire order management and reconciliation in various different Google Sheets.

Yes. Sheets!

No APIs, no systems, just manual uploads, edits, and emailing files back and forth.

This was a wake-up call. The e-commerce world is wild and fragmented.

Trying Other Integrations: Unicommerce and Shiprocket

We didn’t give up.

We built integrations with Unicommerce, a popular order and inventory management platform, for brands that used it.
Orders received from Unicommerce would trigger automated reconciliation attempts.
If the auto-reco failed (due to mismatches or tech errors), it would fall back to manual review.

Similarly, we tried integrating with Shiprocket. For a few months, automated reconciliation was 95–99% accurate — a huge win. But over time, discrepancies crept in again, forcing us to request manual reconciliation sheets from brands to cross-check.

Hybrid Approach: Automation Meets Manual

Our final automated reconciliation approach became hybrid:

Use APIs and integrations to do as much automatic matching as possible.
Fall back on manual uploads and QC when automation failed or data was incomplete.
Keep creators’ experience safe by only showing auto-reco data when confidence was high.
Maintain manual controls for exceptions, fraud detection, and edge cases.

Lessons Learned

No one-size-fits-all: Each brand had unique quirks, systems, and processes. Automation had to be flexible.
Communication is key: Regular syncs with brands and clear documentation saved countless headaches.
Expect the unexpected: Manual interventions, data mismatches, and late changes are the norm, not the exception.
Automation is a journey, not a switch: We automated incrementally, always keeping manual fallback options.

Wrapping Up Part 4

Automated reconciliation was the dream, but reality kept us grounded. Still, the hybrid approach gave us the best of both worlds — efficiency where possible, human oversight where needed.

Next, we’ll explore how reconciliation impacts creators and finance teams, and how we built processes and alerts to keep everyone happy and informed.

Stay tuned. The saga isn’t over yet.