Death of a Salesforce: Why AI Will Transform Sales

0:00:03 – Plugging into anywhere that customer feedback comes in,
0:00:05 and that could be your email, that could be Slack,
0:00:06 that could be Zendesk.
0:00:09 System of records can now have much more context,
0:00:12 and context is what human beings use to make decisions.
0:00:14 There’s insatiable demand right now,
0:00:17 a demand in frankly, in which I’ve never really seen.
0:00:21 – In 1999, some 25 years ago,
0:00:23 Mark Benioff started Salesforce,
0:00:25 with a vision to build, quote,
0:00:28 a world-class internet company for Salesforce automation.
0:00:30 Now today in 2024,
0:00:33 there are thousands of Salesforce partners,
0:00:35 and hundreds of thousands of certified professionals,
0:00:37 all around the world.
0:00:39 Not to mention other multi-billing dollar companies,
0:00:41 like HubSpot or Zendesk,
0:00:43 that are fundamentally built
0:00:46 on the concept of a system of record.
0:00:48 And selling is a strategy as old as time,
0:00:51 but the latest iteration gave sales teams the ability
0:00:55 to track, to assign, and to reward sales more effectively,
0:00:57 all through better technology.
0:01:01 And the good news is that technology is not done progressing.
0:01:03 So that’s what today’s episode is all about,
0:01:06 the intersection of sales and artificial intelligence.
0:01:10 So how will AI change selling as we know it?
0:01:12 What workflows can be automated or redefined?
0:01:15 And better yet, what can the integration of AI
0:01:18 do to fundamentally change the way that selling is done,
0:01:21 just as we saw in the last era of the CRM?
0:01:22 We discuss all this and more together
0:01:25 with investing partners on the A16Z apps team,
0:01:27 Joe Schmidt, and Mark Andrusko.
0:01:29 This is also part of a wider series
0:01:32 on how AI is disrupting the enterprise.
0:01:34 So if you’re interested to hear more
0:01:35 about how it’s reshaping everything
0:01:37 from marketing to accounting,
0:01:41 head on over to our A16Z YouTube channel.
0:01:42 Oh, and one more thing.
0:01:44 This episode was actually based
0:01:46 on a long-form article co-written by Mark,
0:01:48 so we’ll link that in the show notes.
0:01:51 It’s called “Death of a Sales Force.”
0:01:52 Okay, let’s get started.
0:01:56 (upbeat music)
0:01:57 As a reminder, the content here
0:01:59 is for informational purposes only.
0:02:01 Should not be taken as legal, business, tax,
0:02:02 or investment advice,
0:02:05 or be used to evaluate any investment or security
0:02:06 and is not directed at any investors
0:02:09 or potential investors in any A16Z fund.
0:02:11 Please note that A16Z and its affiliates
0:02:13 may also maintain investments
0:02:15 in the companies discussed in this podcast.
0:02:18 For more details, including a link to our investments,
0:02:21 please see a16z.com/disclosures.
0:02:23 (upbeat music)
0:02:28 We are constantly being sold to,
0:02:30 feels like all the time, selling is a constant,
0:02:32 but maybe what has changed is the way
0:02:34 that we’re being sold to,
0:02:36 and specifically technology’s role in that.
0:02:37 So tell me a little bit about that.
0:02:40 How has sales changed over the last few decades?
0:02:44 – Sales is maybe the oldest profession that there really is,
0:02:46 and sales has been going on since prehistoric times.
0:02:49 Like as long as people have been doing something
0:02:50 with the act of trade,
0:02:53 there has been the need to record who is my customer?
0:02:54 Where are they?
0:02:55 What do they need?
0:02:55 What do they bought from me?
0:02:58 So like that act is not necessarily new.
0:03:01 I’d say the first tracking mechanism that you might’ve had
0:03:03 would be like someone writing that down on a piece of paper.
0:03:06 Maybe that evolved into like a Rolodex in the 1950s,
0:03:07 you think of Mad Men.
0:03:10 But then like the modern kind of CRM really came around
0:03:12 in like the 80s with ACT,
0:03:14 or maybe even the 90s with Siebel systems.
0:03:17 And these were the first digital CRM.
0:03:18 But at the end of the day,
0:03:20 it was always the same exact act as that prehistoric time.
0:03:22 It’s like there’s a salesperson,
0:03:23 they’re doing some act of selling,
0:03:25 they’re recording that information
0:03:27 and they’re putting into a certain place.
0:03:29 Salesforce then comes in at the turn of the century in ’99
0:03:31 and says, “Hey, let’s move this into the cloud.
0:03:33 Let’s actually create an easier way for that sales rep
0:03:34 to enter that information
0:03:36 and have access to it on the go.”
0:03:37 So now you can go and log in your system
0:03:39 and have all that information wherever you might be.
0:03:40 At the end of the day though,
0:03:43 it’s always been a sales rep or whoever that might be
0:03:44 with a relationship with a customer
0:03:46 that wants to record that information.
0:03:48 What’s interesting about this moment in time
0:03:51 is that the way that that information is being captured,
0:03:52 there’s an opportunity to do it
0:03:54 in an entirely different way.
0:03:55 Like there’s been this tectonic shift
0:03:57 in the way that technology works.
0:03:59 And so now instead of a human being sitting down
0:04:01 at their computer and entering something in,
0:04:04 we now can have basically AI always on
0:04:06 and recording that conversation in real time
0:04:09 or frankly having that conversation.
0:04:11 And so you start to think about all of the legacy systems,
0:04:15 whether it’s pen and paper, Siebel systems or Salesforce
0:04:16 that have been dependent on a human.
0:04:19 And now you’re starting to see an AI native system
0:04:20 that can have these conversations
0:04:22 and really start to capture that information at the source
0:04:24 rather than the waiting for the human.
0:04:26 – For the companies who are using these tools,
0:04:29 compared to the alternative of not having tools
0:04:31 or pen and paper, like you said,
0:04:35 how reliant are these companies on these tools
0:04:37 and how embedded are these systems?
0:04:37 – They’re extremely embedded.
0:04:39 I mean, I think there’s a reason
0:04:41 that all venture capitalists and technologists
0:04:43 are obsessed with this idea of the system of record.
0:04:45 It’s because historically systems of record
0:04:47 have benefited from tremendous moats
0:04:48 and that they are really hard to rip out.
0:04:50 And the reason they’re really hard to rip out
0:04:52 is they have the source of truth data
0:04:55 about some of your most important parts of your business.
0:04:56 In this case, we’re talking about like
0:04:58 who your customers are and who your prospects are.
0:05:00 – And so you wrote this article
0:05:02 that’s literally called the death of a Salesforce,
0:05:04 which is a pretty strong statement,
0:05:06 but you are talking about some pretty fundamental shifts
0:05:08 of structured to unstructured data.
0:05:09 So tell me more about that.
0:05:12 How does that shift potentially reshape the sales cycle
0:05:13 or in this case, like you’re saying,
0:05:15 the death of a Salesforce?
0:05:17 – One of the concepts I’ve loved for a very long time
0:05:21 is like don’t focus on activities, focus on achievements.
0:05:23 And the problem with historical sales software
0:05:26 is so much of the activities that you track
0:05:27 as like the leading indicators of success
0:05:28 of whether or not you’re doing a good job
0:05:30 or whether or not you are really come down
0:05:33 to like these like very fallible activities by the rep
0:05:36 or by the team manager or whoever it might be.
0:05:38 And so in so far as you can get rid of that
0:05:41 and basically have what is the core piece of truth,
0:05:43 what is happening with the customer,
0:05:45 what’s really going on in the field
0:05:47 and basically using that to power your Salesforce.
0:05:50 Like that’s when this gets to get really interesting
0:05:51 and you can start to focus on,
0:05:53 okay, what are actually the core pieces of activity
0:05:55 that do lead to achievement rather than the things
0:05:58 that you intuitively think are right
0:06:00 or may I don’t really think are right.
0:06:01 It’s like, what is the cold heart data?
0:06:02 I mean, how do we make better decisions?
0:06:03 – What’s an example of something
0:06:05 that you think could disappear completely
0:06:07 with this new approach?
0:06:09 – Before I get to the things that will disappear,
0:06:10 I don’t think we want to get up here
0:06:12 and say sales is going to disappear.
0:06:14 I used to run a sales team in the job before this,
0:06:17 Mark obviously sold, AI is not going to replace
0:06:20 like a very high quality sales process.
0:06:22 You need to figure out who your customer is,
0:06:24 you need to figure out what they want
0:06:25 and you need to build the best product
0:06:28 and the right process around that to be effective.
0:06:30 That human element is not going to go away.
0:06:33 Now, I think parts of that process to your point
0:06:37 will disappear and they will frankly change a lot.
0:06:39 Something that we’re seeing a ton of right now
0:06:41 is like, all right, we call it intelligent pipeline
0:06:44 in our post, but basically how are you identifying prospects?
0:06:46 How are you contacting prospects?
0:06:47 How are you scheduling meetings
0:06:49 in basically doing the qualification process
0:06:52 that’s normally associated with like an SDR?
0:06:53 And so you start to think about, okay,
0:06:55 what is the role of that human being?
0:06:56 It’s almost a right of passage and sales.
0:06:57 Like you kind of have to do it.
0:06:59 It’s a miserable existence.
0:07:00 You sit on the phone all day,
0:07:02 you make however many cold calls,
0:07:03 you send however many emails,
0:07:05 we’ve both done this, it’s horrible
0:07:06 and it can be automated.
0:07:07 You know, there are businesses out there
0:07:10 like 11X or others that are in this segment
0:07:11 that have basically figured out, okay,
0:07:13 how do we package up this human being
0:07:14 that is really the beginning
0:07:17 of any high quality sales process,
0:07:18 at least in the enterprise?
0:07:20 And how do we figure out how to, okay,
0:07:22 we’re gonna identify these most interesting prospects.
0:07:24 We’re gonna figure out like the right way to contact them
0:07:28 and generate like a very persuasive outbound message.
0:07:30 And then we’re gonna have an interaction with the client
0:07:31 and try to schedule it for a more holistic
0:07:34 and wholesome call with an account executive.
0:07:36 Like we’re starting to see that process
0:07:39 basically disappear with certain customers,
0:07:40 which is really, really exciting.
0:07:42 And frankly, like freeze up humans
0:07:43 to do the things that they’re best at,
0:07:45 which is like sell, right?
0:07:47 And so I think we’re gonna see things like that
0:07:49 start to disappear in this new world.
0:07:51 – Yeah, so then that leads to the second question,
0:07:54 which is if you get rid of all of this early stage process
0:07:55 that no one really wants to do.
0:07:57 And like you said, it’s kind of a rite of passage.
0:07:59 And all of those people now have the ability
0:08:02 to spend that time on new approaches, right?
0:08:05 So what’s net new that you think comes from this?
0:08:06 – Yeah, totally.
0:08:08 We as a firm talk a lot about like voice agents,
0:08:11 one concrete example of this is imagine
0:08:13 you’re a rep on a call with a prospect.
0:08:15 Maybe it’s a Zoom call or just a regular phone call.
0:08:17 You can now be getting live coaching
0:08:21 from an AI voice agent that is trained on all of the data
0:08:23 that you have seen from prospects
0:08:24 and from existing customers.
0:08:27 And as your prospect is objecting to something,
0:08:29 it kind of inserts the right answer into your ear
0:08:31 and you can like say it on the fly.
0:08:33 It kind of reminds me of old school
0:08:34 when they’re trying to take a test at the end.
0:08:36 And there’s like guys in the parking lot
0:08:38 telling them the answers to the test is the same thing.
0:08:41 – Right, but it’s also like autonomous vehicles, right?
0:08:42 You will have only so many miles
0:08:45 and instead the car is trained on all of the different
0:08:48 accidents that have happened to all of the cars
0:08:49 that have been reported, right?
0:08:51 So it’s a totally different level of information.
0:08:52 – 100%.
0:08:55 And then I think like the second order implication there
0:08:56 is that everything can just become
0:08:58 so much more personalized.
0:09:00 Like if you think about a sales rep
0:09:02 who has a really high quality, large lead
0:09:04 that they’re working,
0:09:06 a lot of what they’re doing to prepare to close that deal
0:09:09 is like personalizing some sort of collateral or deck
0:09:10 or something that’s gonna convince their prospect
0:09:12 to buy their product.
0:09:15 And now think about like how easily you can do that
0:09:18 at scale and really tailored to like an individual human
0:09:19 who’s like, oh, you know,
0:09:22 some system picked up that this human was on our website.
0:09:25 Let’s customize a deck that speaks to this person
0:09:28 individually today and do that with a click of a button
0:09:30 instead of a rep spending five hours
0:09:32 to put something together that’s super custom.
0:09:34 And now do that across like every prospect
0:09:36 that is in your pipeline.
0:09:37 We’re just talking about like a level of scale
0:09:40 and personalization that hasn’t been possible before.
0:09:42 – Yeah, and how are you seeing that market develop
0:09:44 when you think of the fragmentation
0:09:46 or the co-lesscence of it all?
0:09:49 What parts of whether it’s like prospecting or qualifying
0:09:52 or later on the process you mentioned like personalization,
0:09:54 how are you seeing startups start to show up
0:09:56 and where are they showing up as well?
0:09:59 – So we think about the new types of companies we’re seeing
0:10:01 in the space in kind of four broad categories.
0:10:02 We were deliberately broad
0:10:05 because I think in sales software,
0:10:08 market maps of the past things get hyper granular
0:10:11 to a point where everything is sort of overlapping.
0:10:13 And I think everyone kind of shares this same vision
0:10:14 of spanning end to end.
0:10:16 So thinking about it in the four broader buckets
0:10:18 helps us do that more clearly.
0:10:20 And basically we think about it as intelligent pipeline,
0:10:23 digital workers, sales enablement plus insights
0:10:25 and CRM plus automations.
0:10:26 And we’ve kind of already touched on each of these
0:10:27 throughout this conversation.
0:10:30 But I think now the question will be, okay,
0:10:33 if you start with any of those four as the wedge,
0:10:35 which wedge best positions you to then earn
0:10:37 the rest of that pie over time?
0:10:39 And I don’t think we have an answer to that yet.
0:10:40 I think that’s something that we’re really excited
0:10:41 to see play out.
0:10:42 And I think as you talk to different founders,
0:10:43 they have tremendous conviction
0:10:45 and one being the right wedge over the other.
0:10:48 And so we’re still early in that journey.
0:10:50 – Where are you seeing the most adoption
0:10:53 in terms of today, not future looking,
0:10:55 but today you see different sales forces willing
0:10:58 to engage with one of those four categories.
0:11:01 I think that we’re seeing a ton of adoption
0:11:03 in the intelligent pipeline category.
0:11:05 Back to what I said earlier on like,
0:11:09 what is the job of like the thankless job that is being an SDR?
0:11:13 It is very automatable to think about like, okay,
0:11:14 for the companies that know what they’re doing
0:11:15 and know who they’re selling to,
0:11:17 how do I ingest all that information
0:11:19 about who my ideal customer is?
0:11:20 What’s the right process?
0:11:23 And then how do I basically run that contact mechanism?
0:11:25 There is almost no difference oftentimes
0:11:28 between a very well-trained AI agent
0:11:31 and a very poorly trained 22 year old
0:11:32 who’s hungover from the night before
0:11:34 sitting in a crowded sales floor.
0:11:37 And oftentimes you’re gonna get playbook adherence,
0:11:38 you’re gonna get thoughtful follow-up,
0:11:39 you’re gonna get a personalized outreach,
0:11:41 whereas the other person is sending
0:11:42 like a thoughtless LinkedIn request
0:11:44 or maybe it’s a half-hearted email.
0:11:45 – Yeah, I would like to connect.
0:11:47 – Right, yeah, to the wrong decision maker.
0:11:48 And so now you start to think about
0:11:50 what you can actually do with that.
0:11:53 And that’s why I think there’s insatiable demand right now,
0:11:55 a demand been frankly in which I’ve never really seen
0:11:58 for products in that intelligent pipeline layer.
0:11:59 – Yeah, and as some of these categories
0:12:02 eat parts of the sales process, as you’ve talked about,
0:12:04 how does that change a sales team?
0:12:06 You’ve kind of alluded to this a little bit,
0:12:09 but if you no longer have the young college kid
0:12:11 who’s forced to take all of the early stage calls
0:12:14 and train that way, what does that person do?
0:12:16 How do you actually train a sales team effectively
0:12:18 in this new environment?
0:12:21 – Today there is kind of discreet quotas, systems,
0:12:24 organizational structures for marketing sales,
0:12:25 customer success and account management
0:12:27 and customer support.
0:12:29 But if you like take a step back,
0:12:30 that whole go to market organization
0:12:32 is just trying to do what’s right by the customer
0:12:35 and be the North Star of the voice of the customer.
0:12:37 And so it’s kind of silly that a sales rep closes a deal,
0:12:39 has a quota for a new logo,
0:12:41 and then a customer success or account manager
0:12:43 takes over that logo and has like a different quota
0:12:45 and the relationship is different.
0:12:47 I think these teams are just gonna be able
0:12:48 to work much more closely together
0:12:51 and out of the same shared system.
0:12:53 And I think they can even be comped together.
0:12:55 Like everyone can just kind of reorient around
0:12:57 the thing that matters, which is doing right
0:12:58 by the customer at all times.
0:13:00 And so I think the rosy take is that
0:13:02 this will be a great outcome for customers
0:13:04 because I think teams are just gonna be much more coordinated
0:13:07 and much more aligned and the lines will be blurred
0:13:08 between like who’s generating the pipeline,
0:13:09 who’s closing the pipeline
0:13:10 and who’s managing the relationship after.
0:13:12 Everyone can be doing a little bit more of everything.
0:13:14 – So to date, most of these systems of record
0:13:16 mostly have structured data.
0:13:18 What I imagine as a non-sales person
0:13:21 is status of company and like contact, email,
0:13:22 things like that.
0:13:24 In this new world, what would that really look like?
0:13:26 You’ve mentioned multimodal,
0:13:27 but are we talking images, videos,
0:13:29 and like how would that actually influence the workflow?
0:13:31 – Yeah, 100% and it’s like all of the above.
0:13:33 It’s basically plugging into anywhere
0:13:35 that customer feedback comes in
0:13:37 and that could be your email, that could be Slack,
0:13:38 that could be Zendesk.
0:13:39 Those are obvious ones,
0:13:40 but think about even the non-obvious ones
0:13:43 like customers might tweet about your product.
0:13:45 You might send an NPS survey on Qualtrics.
0:13:46 You might have offline meetings
0:13:48 that you have to now record in this universe.
0:13:51 And I think the key point to take away here
0:13:55 is that system of records can now have much more context
0:13:58 and context is what human beings use to make decisions.
0:13:59 That’s one of the most compelling parts
0:14:02 of the opportunities is you can replicate human judgment
0:14:04 a little bit more as you get more and more context
0:14:05 and as you lack context,
0:14:07 it’s hard to trust AI agents to act autotomically
0:14:08 on your behalf.
0:14:09 – Yeah, and I think that like basically gets back
0:14:11 to the point of what we were saying of
0:14:13 what is the new system of record?
0:14:15 In the past, it’s just been like a UI to interact
0:14:17 with a bunch of human generated data.
0:14:19 It is your interface to understand
0:14:20 what’s going on in the business.
0:14:24 But if that information is no longer just being surfaced
0:14:26 in this like one engagement layer
0:14:30 that like a human being is creating and maintaining
0:14:31 and whatever it might be.
0:14:33 Now you have all these other pieces of data.
0:14:35 It would be almost impossible to actually have that
0:14:37 all on one place and use it effectively, right?
0:14:39 So how do you think about like that data
0:14:42 actually informing every part of your sales process,
0:14:43 every part of your marketing process,
0:14:44 every part of your customer success process.
0:14:45 – And then importantly,
0:14:46 now you start to bridge
0:14:49 into every part of your product and engineering process.
0:14:49 – Correct.
0:14:51 So historically, product and engineering
0:14:52 aren’t really spending that much time
0:14:54 clicking around Salesforce.
0:14:56 But if you actually have this multimodal system of record
0:14:58 that captures all of your customer feedback in one place,
0:15:00 all of your prospect feedback in one place,
0:15:02 what could possibly be a better tool
0:15:05 to feed the input of like roadmap planning
0:15:06 and all the things that actually matters
0:15:07 to product and engineering.
0:15:09 And so now we’re starting to talk about systems of record
0:15:11 that could bridge different parts of the organization.
0:15:12 That’s super interesting.
0:15:13 – Yeah, yeah.
0:15:15 You’re finally gonna fix the whole product versus sales.
0:15:16 – Yeah.
0:15:17 – Debacle.
0:15:18 – I don’t, I don’t.
0:15:19 – Remember, we fix it in this room.
0:15:21 – I think it was actually this podcast
0:15:22 that solved all the problems.
0:15:24 But yeah, no, very exciting times.
0:15:25 – No, I think it’s great.
0:15:26 I mean, you use the term maintenance
0:15:28 and it does feel like a lot of systems today
0:15:29 are after the fact, right?
0:15:30 You take something that’s happened
0:15:33 and you’re maintaining your record of it.
0:15:34 But as both of you are saying,
0:15:36 it’s really this idea of a true second brain
0:15:38 where this information is working for you.
0:15:38 – Yeah, yeah.
0:15:40 Or honestly, it could become the first brain
0:15:42 before it gets to my second brain.
0:15:44 And you know, much more effective version of it.
0:15:45 So, yeah.
0:15:46 – Awesome.
0:15:47 Well, this was great.
0:15:48 Thank you so much for talking through this.
0:15:50 I guess this podcast is effectively sale.
0:15:51 So it’s everywhere.
0:15:52 – Yeah.
0:15:53 – Can’t wait to use some of these tools myself.
0:15:54 – Yes.
0:15:55 Thank you for having us.
0:15:56 – Yeah, thanks for having us.
0:15:58 (upbeat music)
0:16:00 – All right, that is all for today.
0:16:01 If you did make it this far,
0:16:03 first of all, thank you.
0:16:05 We put a lot of thought into each of these episodes,
0:16:07 whether it’s guests, the calendar tetras,
0:16:09 the cycles with our amazing editor, Tommy,
0:16:11 until the music is just right.
0:16:13 So if you’d like what we put together,
0:16:17 consider dropping us a line at raidthespodcast.com/a16z.
0:16:20 And let us know what your favorite episode is.
0:16:23 It’ll make my day, and I’m sure Tommy’s too.
0:16:24 We’ll catch you on the flip side.
0:16:27 (upbeat music)
0:16:29 (upbeat music)
0:16:32 (upbeat music)
0:16:35 (upbeat music)