Torture it long enough & it will tell you exactly what you want?
I have seen similar metrics across a number of SaaS companies. I think what really matters is what actions will be taken with this data to further prove out hypotheses taken from the data. Too often I see people debating a metric number and then the discussion stops there because they are exhausted trying to defend their point. These metrics are a reference point. Your sub head tells it all.
Very interesting! A few points come to mind:
— Low podcast attribution: The fact that the podcast didn't jump out in HDY data & then explain all the "dark" direct traffic seems quite notable, and the folks pushing podcasts + social for e.g. IMO confuse the medium for the message. CW etc use podcasts+social to push a very distinct narrative and then convert people who buy on narrative. If you 'do a podcast' but don't have that kind of message *or* that kind of extremely online market to meme (i.e. other marketers) then you're outta luck in that playbook, and I'm not sure they'd have much more to offer. In this case, it might be interesting to try and plug a podcast-only offer to send folks to a specific page and see if anyone bites as a test of prospect engagement.
— High trade show attribution: The fact trade shows conversely *does* stand out is pretty interesting -- I was surprised by that. Given the effort + expense that goes into trade shows, I wonder if that makes HDY info worthwhile, at least for a period.
— Organic search: This still remains a black box IMO, and on the organic search side I find "organic" as a grouping pretty meaningless when it's brand + SEO, and it'd be interesting to see how much organic search traffic went to non-brand-SERP pages & what GSC looks like for brand search.
— Direct: IMO it's pretty interesting that HDY data *does* shed some light on direct traffic, and if I understand your correlation chart, that what it reveals is as you say -- it's just a mix of the other sources, there isn't a secret hidden pseudo-channel there waiting to be exploited.
Finally, it's interesting to think about what we can infer about awareness ('brand') vs activation ('conversion') from this kind of data too. I wonder what a kind of meta metric that bucketed all 'brand' traffic together (organic brand, paid brand, direct brand) and treated that as function of all other activity (paid + SEO + podcast + trade shows), with a kind of weighted contribution from each. Not sure exactly what it would look like, but I'd like to think one day our marketing tools will better reflect buying reality and not just data collected... we can dream, at least :)