Hacker News Reader

Comment by shadowgovt | Hacker News Reader

I'm modestly surprised to learn Google was using Salesforce internally at all; the NIH runs deep with that company (they even have their own bugtracker because every other option just wouldn't cut it).

On the other hand, the past decade-ish has seen them grow very rapidly via acquisition, so perhaps this DB was grandfathered in via an acquired company and hadn't yet been replaced by anything internal.

(For Salesforce in particular though, I'd be willing to believe Google doesn't have an in-house alternative... People asked for a Salesforce-like in Google Workspace for years and the company had no interest. I have a hunch that most Googlers find the idea of creating a new CRM to be a profoundly boring intellectual exercise).

View on HN

Fwiw, I was hired by Google in 2015 to help answer questions like "if Google were to add a CRM to the GSuite portfolio, should they build one, buy one or partner with key players". My team's charter was to create business cases with various options and run them up to chain (at the time, Prabhakar was running product for "Google for Work"). On more than one occasion we presented cases with 3 year ROIs in the $xxxM range and were shot down every time with a "too small" comment. A couple years later, Google had partnered with Copper CRM and supported extension builds into Workspace/GSuite, but had also begun a major enterprise rationalization project to consolidate a multitude of Salesforce instances into a single one, at the same time as adopting standard enterprise features & processes of Anaplan.

This led to consolidation of a number of back office IT teams that ultimately ended up with far more enforcement clout than they'd historically had. By the time Ruth changed roles, most of the "normal" business processes had been fairly standardized. Fwiw, the Cloud instance of SFDC, which is by far the most complex & customized, has been in full use for almost five years now and is the canonical source of truth for sales data.

I wonder if the Cloud SFDC is the one that was compromised. It's a little telling Google didn't go into details about which arm of the octopus got attacked (or if they did, I didn't see that reporting yet... Unless Cloud is the implied victim because the description of the attack showed up on the Cloud blog).

I feel you about the ROI. In hindsight, it's a little funny to me that Salesforce is doing revenue numbers a little under half of Google Cloud; you'd think that would be large enough value to get Google interested in biting into that pie.

I'm surprised Google could get away with only a single SFDC instance. AWS has multiple SFDC installations and is forever having to deal with "Oh, yeah, that data is in this other SFDC installation"

Yeah, they have the world class Salesforce engineers there. One of Google's Salesforce's last tech leads wound up becoming the Director of the proprietary Salesforce language Apex.

> they even have their own bugtracker because every other option just wouldn't cut it

Of all the things to NIH, this is one of the most defensible -- lots of bugtracker options just aren't very good.

I found this to be true too, but I don't really get it. Doesn't seem like that complicated of a software. Maybe I'm only thinking like a SWE, and not PM and other laypersons that also need access.

It's definitely not that complicated. It's just one of those bits of software that is paid for and managed by people that aren't actually using it, so you get Jira shit.

I used to work for Dyson and they moved to Teamcenter for CAD management (basically a shitty VCS for CAD). Similarly to Jira it had all the features you'd ever want in its white paper, but it was abysmal to use - even worse than Jira.

Anyway the nicest bug tracking software I've used so far is Phabricator. Quite a lot nicer than anything else, but it is tightly integrated, and I wouldn't really recommend Phabricator these days because a) no integrated CI system, b) it's semi abandoned, and c) PHP. And yes that does matter. (Though TBF it beats Ruby.)

I've generally not had an interest in working for one of the big tech companies, but the opportunity to escape JIRA is tempting.

Yeah, Google's Buganizer was the best bug tracker that I ever used.

...and it still wasn't great.

It's pretty much perfect in my eyes. Not being open source is probably the biggest thing I'd fault it for. The world deserves better than GitHub issues and jira, pity it can't be used by anyone else.

Salespeople are VERY familiar with Salesforce and are not very technical. Probably significantly increases onboarding and training time to have a weird new tool.

Easy to hire experienced salespeople and have them hit the ground fast if they use standard Salesforce conversion flows.

It still amazes me that Salesforce, which is good, mind you, is still basically just Microsoft Access as a Service, and yet here we are.

Google uses lots of non-Google solutions for many things —just imagine all the facilities stuff. But so does any software company, including Microsoft and Amazon.

That said, you can hire people for any purpose (specific roles) and you can build what you want. It’s more a question of whether it’s worth it to build such solutions, after all you have a main line of business to tend to. That’s to say even Google and Apple have so called “boring “ roles and there are lots of people who don’t see it that way and want to work doing those things.

Actually lot of the facilities stuff is inhouse too - floor plans (not just the seat map but actual floor drawings that include physical infrastructure); the ticketing system for maintenance; work hour tracking for contractors; probably lot more that I'm forgetting.

But yes your point stands, sometimes it just makes more sense to use an existing product.

The floor plan tool isn't really in house. It's just an extension of the industry standard real estate management platform they use Tririga (https://www.ibm.com/products/tririga) ... in the same way that go/teams in just an custom visualization of a standard employee directory.

You might be surprised how much of what runs Google (Anaplan, for example, for XWS) is fairly industry standard.

They did acquire (then sell) SketchUp which is what I use for floorplans.

Given the low expected profit margin, a CRM solution at Google would likely come from a 20% project (or rather, the equivalent thing these days since last I checked 20% is basically dead as a formal concept). Nobody expected GMail to blow up the way it did, for example; it happened because some Googlers decided they could probably do a web-client-fronted mail client with a Google search engine attached to it and if they did it'd be really cool.

But even with their, what, 180,000 people these days, I think it's entirely possible nobody is as excited about CRM as Paul Buchheit was about email services.

iirc google cloud’s entire support ticket system is built on top of sf - it went down when saleforce had an outage a few years back

Reply on news.ycombinator.com