Yeah, the volume definitely becomes a real bottleneck - I'm dealing with similar numbers and the manual upload process eats up way too much admin time that could be spent on actual candidate evaluation. The duplicate detection by phone is smart, though I've found even that gets tricky when people use different formats or have multiple numbers listed.
That's such a smart approach with the plain text backup! I've definitely been in those situations where a beautifully designed resume turns into a data entry nightmare. The phone number duplicate detection is interesting - I hadn't thought about how people switch emails but keep the same number. For bulk uploads, I've started doing a quick visual scan of the batch beforehand to separate out the obviously problematic formats, which saves time on the backend even if it adds a step upfront.
The bulk upload challenge you're describing really resonates with me - we face similar volume pressures in management consulting where we're constantly pulling talent from multiple sources for different client engagements.
Your point about the creative resume parsing issues is spot-on. We've definitely encountered that nightmare scenario where someone submits a beautifully designed resume that looks great but completely breaks the parsing logic. What I've found particularly frustrating is that these tend to come from exactly the candidates you want to engage with - creative professionals, senior executives with branded personal materials, or consultants who've invested in professional resume design.
The dual submission approach is clever, though as you noted, it's not scalable for unsolicited applications. We've experimented with something similar by requesting a LinkedIn profile link alongside resume submissions, which gives us a structured data fallback when parsing fails. The challenge is that not everyone keeps their LinkedIn current, especially passive candidates who aren't actively job hunting.
From our experience with Talantly.ai, I'd say the bulk functionality works reasonably well for standard formats, but we still hit customization limits when trying to map complex candidate data to our specific workflow requirements. The real-time error reporting you mentioned would be incredibly valuable - right now, we often don't discover parsing issues until we're deep into the review process, which creates bottlenecks downstream.
One workaround we've developed is creating intake templates for different referral partners. Rather than asking them to submit individual resumes, we provide a structured spreadsheet format that captures the key data points we need, along with resume attachments. It's not perfect, but it reduces the parsing variability and makes bulk processing more predictable.
The duplicate detection based on phone numbers is brilliant - I hadn't thought of that angle. We've been relying primarily on email matching, but you're absolutely right that phone numbers tend to be more consistent identifiers, especially for candidates who maintain separate professional emails for different opportunities.
Have you found any effective ways to handle international candidates where phone number formats vary significantly? That's been one of our ongoing challenges as we source globally for client projects.