Why where your files live affects how fast they are processed, and what we can do about it
The Core Issue
Q: Why does my location affect processing speed?
Our processing infrastructure runs in the US and EU. Before we can extract or enrich anything from a document, we first have to download it. If your files are stored in a cloud bucket located in a different region, such as Asia-Pacific, South America, or the Middle East, that download has to travel a long distance before processing can start.
This cross-region file transfer is consistently the single largest contributor to elevated latency for geographically remote clients. It is not a processing bottleneck, it is more of a distance problem.
Key point: The download step alone can add 500ms to 1.5 seconds or more per request when files are stored far from Veryfi infrastructure. This happens even before any document processing begins.
Q: How do I know if this applies to me?
If your files are stored in a cloud bucket outside the US or EU (for example, AWS ap-southeast-1, ap-northeast-1, sa-east-1, or similar non-US/EU regions), this is very likely affecting your processing times. The pattern is usually consistent high latency across most requests rather than occasional spikes.
If you are unsure where your files are stored, your engineering or infrastructure team will be able to confirm the bucket region. Share that with us and we can assess whether it is a factor.
Q: Is this a bug or something you can fix on your end?
It is not a bug. It is a physical constraint of how data moves across the internet between regions. There is no configuration change we can make on our side that eliminates this penalty without changing how or where files are delivered to us. Addressing it requires changes to your file delivery approach, which we can work through with you.
What We Can Do About It
Q: What are the options for reducing region-related latency?
There are three approaches, listed from most practical to most involved. None of them are one-click fixes, but all of them meaningfully reduce the distance problem. We will always evaluate these with you before recommending one - Get in touch!
# | Approach | How It Helps | What It Involves |
1 | Cross-Region Replication (CRR) or read access grant | Your bucket replicates to a US/EU bucket, or grants us read access so we pull files locally before processing. Removes most of the cross-region download penalty. | Bucket configuration on your side; no integration code change needed; available for SLA Silver+ clients. |
2 | Upload directly to a US/EU bucket | Files arrive in our region from the start, so there is no cross-region transfer at all. Best possible latency for file delivery. | Integration change required on your end; available for SLA Silver+ clients. |
3 | Regional processing infrastructure | Processing runs in your region end-to-end. Eliminates distance as a factor entirely. | High complexity, longer timeline; available for SLA Platinum clients. |
Q: Which option is best for us?
Cross-Region Replication is usually where we start. It requires the least change to your integration and delivers meaningful latency improvements in most cases. Direct upload to a US/EU bucket is the most effective if you are willing to adjust how files are submitted. Regional infrastructure is reserved for strategic, high-volume accounts where the investment makes sense.
We will not recommend any of these without first understanding your setup and getting alignment from our engineering team on the right approach for your situation.
Honest Expectations
Q: How much improvement can we realistically expect?
For clients where cross-region transfer is the dominant factor, CRR or direct uploads typically yield improvements in the range of 500 milliseconds to 1.5 seconds per request. The exact improvement depends on the originating region, file size, and volume patterns.
We cannot guarantee a specific number in advance, but we can share observed ranges from similar setups and help you baseline your current performance to measure against.
Q: Is there a latency guarantee across all regions?
No. We do not offer a universal latency guarantee that applies equally regardless of where files are stored. Geographic distance has a real and measurable impact, and we want to be upfront about that rather than set expectations we cannot consistently meet.
What we do commit to is working with you to understand your constraints, document realistic targets, and identify the best path to meeting them.
Q: What about options that do not involve changing how files are delivered?
There are ways to reduce processing time that do not require changing your file storage setup, such as adjusting which enrichment features run. These can help at the margins, but they do not address the underlying transfer distance. They also involve tradeoffs in output quality that we would walk through with you carefully before considering them.
For clients where region-driven latency is the primary issue, changing the delivery approach is the only way to meaningfully move the needle.
Getting Started
Q: What should we share with you to assess our situation?
The more of the following you can share, the faster we can give you a clear picture:
The cloud provider and region where your files are stored (for example, AWS ap-southeast-1)
How you are currently submitting files to us (URL reference, direct upload, or base64)
Typical document size and page count
How latency-sensitive your use case is and what your target response times look like
Q: When should we bring this up?
During onboarding, before you go live, is ideal. If we know your storage region and latency requirements early, we can flag the risk, set accurate expectations, and begin exploring options before they affect your end users.
If you are already live and seeing elevated latency, reach out to your account team and we will assess the situation together.
Questions about your specific setup? Contact your account team. We are happy to review your region, submission method, and latency goals and recommend the right path forward.