Skip to content

Search the site

GCPCloudMatt RennerGTMsalesInterviews

Google Cloud's Matt Renner: “Half of all AI PoCs are just a bad idea"

"Salesfo… no CRM system’s going to delight you"

Engineers in a sandbox admiring their AI PoCs.

There was a time, some years ago, when Google Cloud (GCP) salespeople would turn up at major banks wearing “board shorts and flip-flops” – as one senior staffer recalls to The Stack – and wonder why they were not taken seriously. It’s been a long road since then: GCP has shrugged off the “killed by Google” fears that it simply might get discarded as a bad bet by Alphabet. It has grown strongly (to $9.6 billion in revenues last quarter, up 28%) and has hired numerous executives with serious enterprise experience, for example CISO Phil Venables, a former CISO at Deutsche and board member at Goldman; this has all helped improve credibility. 

There’s work to do still on its go-to-market though, says Matt Renner, who was appointed President, Global Field Organisation in March 2024. Not least, he says, around how AI capabilities are sold, amid a broader cloud market in which he suggests that “thousands of AI PoCs [proof-of-concepts] were poorly qualified… just bad ideas and they failed.”

Renner, who spent 20 years at Oracle and then nearly four years at Microsoft as President, US Enterprise, joined GCP in April 2023. 

A year into life with Google, Renner sat down with The Stack to talk about his priorities, overhauling GCP’s sales software, avoiding AI fatigue, and more. (The following Q&A was conducted at GCP’s NEXT event in Las Vegas in early April 2024. It has been lightly edited for brevity and clarity.)

Matt, what are your priorities in this new role?

To continue to acquire more customers and grow the customers we have. And then drive value through AI. Those are the three pillars. 

AI is fairly infused now across everything we do.

It’s a catalyst [and] it's also pulling through a lot of revenue…  We're [also] looking to mature, as a sales and services organisation, as we grow.”

I’m detecting a lot of AI hype fatigue out there but thousands of press releases about generative AI… Do you worry GCP is losing focus on the “meat and veg” of the established IaaS/SaaS for enterprise customers?

We have absolutely not taken our eye off the ball for the general business.

What [cloud providers] have been very good at over the last few years is getting customers to sign up for large financial commitments. 

That ends up in a term called backlog, which is basically things that have been committed financially and that aren't necessarily running.

If you look at how sales teams are organised, if you look at how they're compensated, [that] is to actually drive towards converting that backlog. 

AI can contribute to that, as it is going to burn… to spin dials. 

But they [customers] weren't sold to [with an] eye on true business value, and then [without] realistic data structures to go execute against.

There’s ‘fatigue’ here typically because of how AI was sold.

We think the door is open for us to become the adult in the vendor room that shows you the path; shows up with a complete [AI] solution [and helps you deliver value].

How do you want to do this diffently to, say, Microsoft?

The difference between the way I sold at Microsoft, and the way we're about to sell is this: Microsoft tends to have a tonne of generalists with massive portfolios of products. And then they package things where they make things simple for [sales]. So if I’m a rep at Microsoft there’s two ways I’m gonna be successful: One, give me things I can slap on my renewal with Office; it’s how they’ve sold security for example, or two, give me things that they don’t need to get too deep to try; that was ChatGPT. 

I don't need technical depth to come up with that pitch.

I don't need deep experience or industry knowledge to show up and say, ‘ChatGPT’s cool, you should try it; knock yourself out. [That’s resulted in] thousands of AI POCs; none of them properly evaluated… So most failed. 

See also: Google Cloud NEXT: AI for Workspace, BigQuery; compute, storage innovations, an Arm CPU...

[By contrast at GCP] Oliver Parker [VP, AI GTM] has built a team that has sales specialists super deep on just AI; pre-sales specialists super deep on AI. And then solution architects, these are the big brain folks that are even deeper in AI, which we think is required to make someone successful. 

“That's how we're going to market; we're enabling the field to have these really skilled folks to ‘paratrooper’ in and help deliver success. 

Why DO so many AI POCs fail? 

When you see the price of what it takes to take something to production, you have to have a business assessment for it to have value. 

Like ‘I’m going to build an HR bot to help people choose benefits and answer HR questions.’ So my business case is I'm going to eliminate X heads out of 1000 in my HR department, etc.? Well, nobody checked with the HR department which already lost 200 people last year. 

It’s just a bad idea. It’s just not qualified, not a good process. 

And it really matters what's going on with your data, where you're going to point it to; where is it; how can you get access to it; how consistent is it; what shape’s it in. Half the time these folks sure as heck didn't know in HR, for example. Literally, some of these POCs are just a really bad idea. 

The criticisms I’ve heard in the past about GCP have been about inconsistent support and lack of commitment for want of a better word. Think of the “Killed by Google” meme etc. What are your views?

Well maybe some of that was true at some point – and I’m talking about support, not product-wise. I think five years ago Google had a lot of questions to answer on enterprise-readiness. ‘Was Google serious? ‘Within Google, is this going to be a long term thing? Because you're a non-profitable entity at the time? Is Google going to be in enough countries? Are you going to be able to support an enterprise customer’s needs to be global?’ I think all of those have [been answered]. 

Customers aren’t worried about whether we’re serious anymore. 

The biggest issue with support for me has always been if the product works in a consumption environment, you don't need much support… But across support and other pieces we're getting more mature as we grow

What else would you like to see mature more at GCP?

We’re doing the right things. 

We need to mature in how we manage our talent. Number one.

Number two [as you grow so fast], you're also going to have a whole series of systems and other things in the back office you got to work through; countless small ideas that at the time, were a great idea.

We had 1,500 sales plays when we got here! Where do you start? Wouldn’t it be better to have like 36 and they’re all complete? We're trying to get simpler and [make it] easier for our reps and our team. That's the stuff we're trying to get more mature; just making it easier for our reps to wake up in the morning [and have] systems that delight them.

Salesfo… no CRM system’s going to delight you. [We just want to make sure reps have] good assets, good data; that’s ‘mature’.

Do I hear the sound of an CRM replacement looming?

No, it’s just [about] maturing, getting simpler; get to one version of it. 

We’re definitely not ripping and replacing.

By the way, pick a company... my last company; they had a heck of a time with their CRM systems – and they sell one, which they don’t always use.

So using a Tier 1 player wasn’t the problem. They [CRMs] just get implemented over time and with turnover and data problems you can’t always just clean it up. We've got teams that are making it better every day.

Final question. You have years of experience building sales teams and GTM. We see a lot of even quite advanced/late-stage tech startups failing here. What does it take to do this well?

Start with talent; you can’t beat talent. 

Number two, think about alignment and how you’re lining them up.

These days it’s table-stakes to start moving towards industry [specific verticals]. You have to be relevant to customers. The fastest way to do that is to isolate the number of industries they [sales teams] operate in. 

Then what are you providing them on the asset side?

How complete is it, whether that’s a sales play, a resource, a definition. That’s super important, how you enable them when they show up.

So: Bring in the right talent; line them up right; provide them with the right tools – and have a good product that drives value for customers! You'd be surprised how many companies mess up one of those badly.

Join peers following The Stack on LinkedIn