Excellent article, Diddo — but my first thought was around the first of the nine steps, "Map your tax data pipeline. Find every source feeding a compliance output. Find where the inconsistencies live before the algorithm does."
This may be possible at Uber, but it would be completely beyond most large supply chain multinationals with a significant ERP history. Even with AI's help, understanding how that data is stored and maintained in the first place requires more skills than they typically have available.
Even then, the cost-benefit would only make sense if it goes far beyond tax.
Thanks, Geoff. Good point, it's a cumbersome undertaking. At Uber, we have an internally built data pipeline and master data management solution that makes things easier for us. Interesting to understand - is the bottleneck with large supply chain multinationals connected to legacy ERPs? Complexity of implementation? Skills? Curious to understand the details.
All of the above and then some. Third-party ERPs like SAP and Oracle, by their very nature, have to be flexible enough to handle any type of business from supply chain to banks to airlines — but that flexibility is the very rope that implementers hang themselves with. Then they acquire other companies, lift 'n' shift to cloud, and — as is the current vogue — remove customizations. All this while thinking about processes, not data.
That said, I think it is the sprawling, historical supply chain conglomerates that have the biggest problem.
This aligns with my observations as well. Over the past 2 years, I've seen many companies implement or plan to implement customization layers on top of Oracle/SAP to handle data manipulation. As you rightly pointed out, you can standardize many processes, but the data remains highly specific. I've seen this at Uber. For example, reporting obligations may differ from city to city in the USA, making custom data generation and quality checks necessary. This cannot be provided out of the box by a traditional ERP.
Excellent article, Diddo — but my first thought was around the first of the nine steps, "Map your tax data pipeline. Find every source feeding a compliance output. Find where the inconsistencies live before the algorithm does."
This may be possible at Uber, but it would be completely beyond most large supply chain multinationals with a significant ERP history. Even with AI's help, understanding how that data is stored and maintained in the first place requires more skills than they typically have available.
Even then, the cost-benefit would only make sense if it goes far beyond tax.
Thanks, Geoff. Good point, it's a cumbersome undertaking. At Uber, we have an internally built data pipeline and master data management solution that makes things easier for us. Interesting to understand - is the bottleneck with large supply chain multinationals connected to legacy ERPs? Complexity of implementation? Skills? Curious to understand the details.
All of the above and then some. Third-party ERPs like SAP and Oracle, by their very nature, have to be flexible enough to handle any type of business from supply chain to banks to airlines — but that flexibility is the very rope that implementers hang themselves with. Then they acquire other companies, lift 'n' shift to cloud, and — as is the current vogue — remove customizations. All this while thinking about processes, not data.
That said, I think it is the sprawling, historical supply chain conglomerates that have the biggest problem.
This aligns with my observations as well. Over the past 2 years, I've seen many companies implement or plan to implement customization layers on top of Oracle/SAP to handle data manipulation. As you rightly pointed out, you can standardize many processes, but the data remains highly specific. I've seen this at Uber. For example, reporting obligations may differ from city to city in the USA, making custom data generation and quality checks necessary. This cannot be provided out of the box by a traditional ERP.