This message was deleted.
# general
s
This message was deleted.
a
what stage of the company are you guys in? (or total engineering size is an indicator)
a
depends what your goals are and what type of company
b
There was a blog post about this from the Netflix team (iirc) The numbers I've seen is that a platform team starts generating substantial leverage at 100 engineers That was a few years ago and I think the number has dropped in terms of providing initial leverage However, I've seen teams of 5 service 1000 eng org
d
I’m mostly interested in mid-to-large size organizations at companies where the primary product is software. 100-to-1000 engineers.
a
are you a software infra or saas?
i am guessing you aren’t including core business capabilities in platform
things like identity, auth, core libraries at business logic layer etc
d
I’m coming from the perspective of a saas company.
That’s an interesting question about where the line is drawn about what fits into platform. Is auth and core libraries platform or product?
a
i would target ~20% for saas
if you include things like reliability, tpm etc in that hc
d
In my opinion things like auth and core libraries fit into the platform space, as those are typically the features that every product team needs to bootstrap their services.
a
if you start including data plat, security, core business layers, rpc system etc potentially 30%
when i ran platform/infra at dropbox i had ~40-45% of eng iirc
that included datacenter, network eng, things like that
b
@Andrew Fong with those numbers, it sounds like you're including ops I think it's probably important to tease apart ops from Dev enablement
a
no i am not including that
dropbox builds a storage system from main()
d
My rough guesstimate at average from the companies I’ve worked at is somewhere around 1/3 of funding goes to things that could potentially be bundled into the platform bucket.
a
it’s the same as building a cloud, we had 7y+ R&D projects on hardware
dev enablement will depend on what you’re doing, if you build products that go to hardware then the ratio will increase bc the complexity goes up
i suspect netflixes numbers do not include the work to integrate across vendors for example, it’s purely the saas service
I wrote about the organizational progression for a blog post, I'll be sharing a followup with statistics on distributions of ratios
would love to know what the ratios are for more people here