This is such a powerful breakdown! I love seeing you share your journey so openly with the community. It’s so applicable to the work you do and the way you think about solving problems. Turning a real constraint into a scalable workflow is makes your perspective so valuable. Proud of you! Master systems guy LET'S GO!!!
Brilliant aproach to solving the LinkedIn blind spot problem! The part about using Apify to convert profile URLs into structured data before feeding to OpenAI is key here since most people don't realize LLMs cant parse webpages natively. The SERP API integration for 250 free searches per month is a clever cost optimization too. What impressed me most was using OpenAI not just for income estimation but also for the role/industry categorization, essentially ofloading the judgemnt calls that would normaly require manual review.
I like how you combine practical automation tools with strategic business thinking. The way you set up the workflow using N8N, profile enrichment, and scoring logic feels like a great example of turning raw data into meaningful, actionable insights.
I’m curious: as you refine the model, how much weight do you think behavior-based signals (activity, engagement, inferred intent) should carry compared to firmographic or profile-based signals (title, company size, industry)? Given that behavior can change rapidly, do you think there’s a risk of overprioritizing dynamic signals over stable firmographic filters?
Hey Phillip, thank you for the thoughtful question. For this specific use case, which is B2C (job seekers looking for their next role), they have already been pre-qualified in our marketing via intent signals (i.e. they have put "Looking for Work" on their LinkedIn profile). There aren't too many other useful intent signals to extract beyond that, so profile-based signals are most helpful in getting me ready for my sales calls.
If Relentless pursued a B2B angle, for example, by partnering with companies to refer candidates to them, intent signals would be much more important. We might start by doing an ICP analysis and finding all relevant firms that meet our criteria for partnership based on firmographic data, but then we'd use a tool like Clay to scan for intent signals that they're primed for a big hiring push (e.g. the # of job postings on LinkedIn has increased substantially, they've just raised a lot of money, leadership at the company is posting about hiring, etc.) which would be a good signal to email the company and see if a partnership makes sense.
This is such a powerful breakdown! I love seeing you share your journey so openly with the community. It’s so applicable to the work you do and the way you think about solving problems. Turning a real constraint into a scalable workflow is makes your perspective so valuable. Proud of you! Master systems guy LET'S GO!!!
Brilliant aproach to solving the LinkedIn blind spot problem! The part about using Apify to convert profile URLs into structured data before feeding to OpenAI is key here since most people don't realize LLMs cant parse webpages natively. The SERP API integration for 250 free searches per month is a clever cost optimization too. What impressed me most was using OpenAI not just for income estimation but also for the role/industry categorization, essentially ofloading the judgemnt calls that would normaly require manual review.
Glad you liked it and thank you for the thoughtful comment!
Love this!
Thanks Will!
would love your thoughts on some of my stuff. follow me back I could DM you?
Sounds good - just followed you. Shoot me a note.
I like how you combine practical automation tools with strategic business thinking. The way you set up the workflow using N8N, profile enrichment, and scoring logic feels like a great example of turning raw data into meaningful, actionable insights.
I’m curious: as you refine the model, how much weight do you think behavior-based signals (activity, engagement, inferred intent) should carry compared to firmographic or profile-based signals (title, company size, industry)? Given that behavior can change rapidly, do you think there’s a risk of overprioritizing dynamic signals over stable firmographic filters?
Hey Phillip, thank you for the thoughtful question. For this specific use case, which is B2C (job seekers looking for their next role), they have already been pre-qualified in our marketing via intent signals (i.e. they have put "Looking for Work" on their LinkedIn profile). There aren't too many other useful intent signals to extract beyond that, so profile-based signals are most helpful in getting me ready for my sales calls.
If Relentless pursued a B2B angle, for example, by partnering with companies to refer candidates to them, intent signals would be much more important. We might start by doing an ICP analysis and finding all relevant firms that meet our criteria for partnership based on firmographic data, but then we'd use a tool like Clay to scan for intent signals that they're primed for a big hiring push (e.g. the # of job postings on LinkedIn has increased substantially, they've just raised a lot of money, leadership at the company is posting about hiring, etc.) which would be a good signal to email the company and see if a partnership makes sense.