#339 Streamlining Customer Success with AI: A Conversation with Shanif Dhanani

[00:00:00]

Mehmet: Hello and welcome back to a new episode of the CTO show with Mehmet. Today I'm very pleased joining me from New York, Shanif, founder of Shanif. Shanif, the way I love to do it is I keep it to my guests to introduce themselves. Uh, so just a little bit about you, your [00:01:00] background, and what you're currently up to.

Mehmet: Cool.

Shanif: Well, uh, Mehmet, thanks so much for having me on the call. My name is Shanif. I am a data geek. I'm a techie. So my background is in software engineering and machine learning. Um, mostly in the world of startups. So, uh, I've done startups quite a bit. Mostly, let's see, first company I built was a company called Tap Commerce, which was one of the world's first mobile ad tech platforms.

Shanif: Uh, at our peak, we were doing like a million ads a second, actually, which was kind of big. Sold that to Twitter. So I worked at Twitter for a while. Then I built a predictive marketing company called Aptio. And today I'm the founder of Locusive, which is essentially an AI co pilot for customer success teams, helping them pull the information that they need to prepare for customer calls or respond to their customers.

Shanif: Building on top of generative AI tools like ChatGPT. So, let's get started. heavily working in that space today. And I'm excited to chat and be here

Mehmet: again. Thank you very much for joining me today, Shanif. I'm really also excited because you know, the topic, [00:02:00] um, I mean, the area you are focusing on is very important.

Mehmet: I'm not talking about AI, of course, AI is very important here, but I mean, you know, it's driving. business to use AI for something significant, which is customer experience, growth, and all this. Now what, you know, this is a traditional question I like to ask everyone. So what were the main challenges that you saw or like, what were the obstacles you saw other companies, they face when it comes to You know, have this smooth customer experience like, you know, because you mentioned what you focus on is to have make sure that their customer success managers are prepared for the calls.

Mehmet: So what were, you know, the main use cases you spot when you decide to start the company?

Shanif: It's such an important question, because when you start a company, you're not working with a lot. And so you really have to go off of a lot of assumptions that you. work to validate. [00:03:00] And that's still something we're doing today.

Shanif: One of the, but you know, to get to answer your question, one of the things that we've continuously found in our, in our discovery calls with CS teams is that they have a lot of information stored across lots of different systems. And that makes it very hard for them to, One, be able to answer customer requests when customers reach out, maybe because they have the data, but it's in a database that they don't have access to, or maybe they have to pull together information from two or three different systems and spend 45 minutes to analyze it.

Shanif: And two, it's very difficult for them to basically put everything together in a way that makes sense to them because they've got tons and tons of data. Calls and notes and follow ups. And so consolidating all of this and making sense of it all is one of the biggest things that we've heard from folks.

Shanif: It takes a lot of time, for example, to prepare for a call, maybe an hour or two, at least for each call. It takes a lot of time to pull together the information you need to get back to a customer. And [00:04:00] so it's, you can think of it as sort of a death by a thousand cuts where you've got information stored across lots of different systems.

Shanif: And your job as a customer success manager is really to drive net revenue retention by providing the customer a great experience, but you're spending all your time pulling information, searching for data, trying to figure out how to best answer a customer. Um, and so that's really the problem that we've come across time and time again.

Mehmet: Cool. Now, Shanif, you mentioned something and it came multiple times with different use cases. Now it's about the data. Now, out of curiosity, uh, did you see also the problem that usually these faults would have that they don't know where their data is? Or for example, it's fragmented across multiple places.

Mehmet: Did you spot something similar when, when you started to, to speak to customers?

Shanif: Yeah. Yeah, absolutely. A lot of times, um, sometimes people won't know where their data is. But a lot of times the data is actually spread across [00:05:00] lots of systems. So for example, you might have contract information and Salesforce and product information in Looker and maybe some email information in your email system with the person with the last time you spoke to your customer account, and then maybe information from Slack based on your team members, knowledge of the account, and so you've got all of these different systems that each contain different pieces of information.

Shanif: And that makes it really hard for you to figure out. You know where to look, where to go, where can I find this information? So information fragmentation is really important and even information discovery, like you said, first, I meant the first problem is sometimes you don't even know what information you've got or where it's stored.

Shanif: And, um, you know, if you don't know where to look, then you can't even begin to answer a customer's question. So data is really important and customer success folks. love working with data. They need the data to do their job well, but sometimes it's just hard to get access to it all.

Mehmet: Absolutely. Of course, like, because I used to work in, in, in that space [00:06:00] as well as sales rep.

Mehmet: So exactly. I know what you're talking about. And when you try to remember, did I send that over slag that I did that, you know, over an email and then, you know, it's like something overwhelming now using, you know, AI. In sales organization, in SaaS companies specifically, how do you think, you know, it can boost actually the conversion rates, um, when it comes also to, to, you know, enhancing their, their revenues?

Mehmet: Where do you see the most, uh, you know, practical example about that?

Shanif: Man, that's, that's such an interesting question. And it's, it's, It covers so many topics, you know, I'll just touch on a couple of them and then I'll talk about the one that we're focused on in customer experience. I actually do a lot of consulting outside of, um, just the core locus of product.

Shanif: We're an early stage company and I'm talking to a lot of CEOs about how they can use AI to implement revenue [00:07:00] conversions, higher revenue upsells, things like that. A lot of the things that we talk about are creating new products based on AI. So, um, there's a few different ways you can do that. One, if you're trying to create.

Shanif: Brand new Greenfield project that uses AI to service your customers better. You can think about, um, automating things that you were never able to automate before, and then using that to create a feature. Uh, perhaps one example that I've worked closely with before are personalized recommendations where perhaps you've got a large library or a set of offerings for a customer, and maybe you don't have a great machine learning engineering team, but you want to.

Shanif: somehow figure out the top 10 or top 20 offerings to offer. One of the companies we work with is a organization that helps companies find grants and funding opportunities. They've got several thousand opportunities, but if a new customer comes in, they need to quickly find the top 10 or 20. So AI can help here by taking, for example, the company's profile and comparing it [00:08:00] against all of the grants that they have and finding the best grants using AI systems.

Shanif: So I can help in terms of creating new products. That way, I can also help in terms of I know you didn't specifically talk to this, but lowering costs, for example, you can start to automate workflows. You can start to create things that automate a lot of tasks that may have been manual before. Um, and then when you come to customer success and customer experience, what we're trying to focus on is allowing the customer's success manager to free up more of their time so that they can focus on strategic things.

Shanif: Sort of research strategic value ads for their customers. What we generally hear is, you know, these folks, their, their main metrics are net revenue retention or net new revenue. And while they're supposed to be driving towards that, they just don't have the time because they're spending hours and hours with followups or responding to customers or trying to catch up with notes.

Shanif: And so what we're trying to do is allow them to not have to worry about that. So they can do things like research where a company is at. and provide new product recommendations [00:09:00] for them, or reach out to a customer that's maybe not using your product as often and saying, Hey, I noticed you didn't get set up very quickly.

Shanif: Here's what I think you should do. And so AI in that case might be an indirect way to, to grow revenue, but still it, what it's doing is it's automating and freeing up a lot of time so that you as a human can focus on more strategic work.

Mehmet: So can we say Shanif, like within this use case, Because I used to, you know, closely work with CS teams, customer success teams, and, you know, one of the things that they always, they were tasked to do, but they were not able because exactly what you mentioned is what we call it, you know, upselling some other offering that We have, right.

Mehmet: So, so that's that, you know, help, you know, in, in, in getting that to them. And, you know, can these AI integrations also as well provide indirectly or directly, I'm not sure, maybe here you can enlighten [00:10:00] us more because part of the job that usually success teams, they do is that they feed back to the product team about, for example, maybe some feedback from the customers, new features that they want.

Mehmet: So. Does that touch that also as well?

Shanif: Yeah, I think you sort of, Mehmet, you hit on the heart of what we're trying to do. There's sort of two sides to what we're trying to do. On side one, we're trying to make the customer success jobs, um, the customer success manager's job a little bit easier by, for example, Providing summaries of conversations or identifying follow ups.

Shanif: Uh, you mentioned the, the ability to surface product rec, like new product upsells and cross sells, we are working on some of that by basically allowing customer success managers to do that. To free up their time so that they can figure out ways to sell to their customers because selling is still a very human activity.

Shanif: But if you don't have time to do it, then you're [00:11:00] not going to be able to do it. But the second part of what you mentioned is a lot closer to what we're building, which is when we have an AI assistant that's tied into the conversations you're having with your customers in terms of The emails that you're sending or the gong messages that you're the gong recordings that you're analyzing.

Shanif: We have the ability to surface what our customers saying at the highest level. What trends are they showing? Are these trends, uh, you know, do we need to worry about them because they look like they're going to churn or does the same, you know, 20 product recommendations keep coming up across all of our thousands of conversations.

Shanif: We can start to create insights around these sorts of conversations because now you have AI that can analyze it all and summarize it all. So certainly to, to your second point, that is essentially the, the, the product, the one of the side effects of the products that we're building.

Mehmet: Fantastic. You know, like this is, will save a lot of time.

Mehmet: I can imagine. Of course, it will avoid the churn as well, because usually the problem that used to happen, things comes up for a feature request, maybe, or, [00:12:00] uh, some enhancement that the customers they ask for get neglected, not because they want to neglect it, but because they are overwhelmed by all the tasks they do.

Mehmet: And this goes, you know, unnoticed I've, I've, I've seen it, you know, in front of my eyes, I can say now in your instructions, and if you mentioned that you use generative AI in Similar to chat GPT. Now, again, out of curiosity, uh, companies usually they are discussing, okay, how we can expose our company data to these AI tools.

Mehmet: So, and I know you talk about, you know, some, some infrastructure, you know, to effectively have this company. internal data feeded somehow. So I'm curious to know about that. Plus, you know, from technical perspective, you know, the challenges behind the internal data access to chat GPT and how this can be addressed.

Shanif: Yeah, you know, that is the core product that we're offering is [00:13:00] this we're building what's called an autonomous agent. So something that can take your request in natural language, figure out how to service it and then give you the best answer. And an agent is a really good framework for connecting your internal data.

Shanif: To an LLM that can make sense of it. So, you know, what's an agent, what am I talking about? Well, you probably use chat GPT and you know, it's smart. It can answer a lot of your questions, but it doesn't know anything about your business, right? It, you know, you haven't connected it to any of your systems. It can't access your user stores.

Shanif: It can't access your contract information. And every company we're talking to wants to be able to do this. So an agent, an autonomous agent. Allows you to plug in your external systems like your salesforce account. So you're even your database, which is something that some of our customers are doing. And then it allows the system to figure out which data structure or data store to pull information from when a user makes a request.

Shanif: So, for example, common requests that some of our customers get is, Hey, can you give me a report of utilization? Can you tell me [00:14:00] how many of my customers are using your software? And this is, they don't say this to us. They say this to our customers, our customers, our B2B companies that sell SAS. So when a customer gets us request, sometimes they have to go into their own database and they have to download some data and then they have to put it into a CSV and they have to answer some questions from a product knowledge base.

Shanif: But if our system is connected to all of their systems, The systems can figure out, Hey, look, I know I need to pull information from a database, but then once I have that, I need to upload it to a CSV and I need to put it into a format that the users can read easily. And so the cool thing about LLMs is if you connect your LLMs to all of these internal systems, then you pass in a user request and they say, Oh, I know that I need to get information from this database first.

Shanif: And then I need to get information from this knowledge base because LLMs are smart enough to do that. So what we've done is we've built the infrastructure to connect your internal systems to the LLMs using things like OAuth, using things like database integrations, using [00:15:00] things like APIs, um, and making those integrations available to the LLM to chat JVT so that it knows where to pull from.

Shanif: When a user request comes in, hopefully it was straightforward.

Mehmet: Yeah, yeah. So if I want to summarize, if I understood it the right way, they don't have to feed the data directly to these LLMs. So basically you act as a, let's call it a proxy or man in the middle kind of architecture where you are sitting next to the data and you just.

Mehmet: You know, connect using the proper authentication, uh, the access to this data and then send it to the LLM. If I, if I understood it correctly, right?

Shanif: I think you got it. You know, we don't send the, we don't send your entire database to the system. We don't send your entire knowledge base. What we do is we work with the LLM and the LLM tells us, Hey, I think I need information from the database.

Shanif: Here's what information I need. So we then execute a SQL query and we give it back to the [00:16:00] LLM or we pull the information from that Salesforce API and give it back to the LLM.

Mehmet: Sure. Now, again, like, uh, I don't like this phrase, but if I want to play the devil advocate here a little bit, and what about sensitive data, because, you know, we saw a lot of organizations.

Mehmet: Of course, but they were using the interface that people know about OpenAI, which is ChatsBT. Of course, like in such enterprise setup, things will be much different. So I'm trying now to simplify this to a SaaS founder who might not very much, you know, into The security aspect of it. So what are the security measures you know that that can be taken so no sensitive data actually is sent to to these LLMs?

Shanif: Well there's a there's a couple of things to think about when you think about data privacy and security. There's the idea of sensitive data going to the LLM and the LLM provider is doing something with it. There's the idea of prompt injection where a user might ask [00:17:00] the LLM to do something dangerous and then there's the idea of the LLM.

Shanif: Providing incorrect responses that might be dangerous. So let's talk about each of those really quickly. When it comes to data privacy and sensitivity, there's a few things that we try to do to make things a little bit more secure. So most of the, all of the time, the data sits entirely on your systems.

Shanif: We are not, like I said earlier, we're not extracting all the data from your systems and sending it to the LLM on every request. But the LLM does need context. It needs to know, for example, if you're asking it, Hey, how many users are active in my in my platform? You need to pull that information from the database, and then you need to send that to the LLM.

Shanif: So what we do is we use the API is that these LLMs provide it. to provide context sensitive data on demand. And what's great about LLMs like chat GPT is if you use their API, they have basically a policy of saying we're going to delete your data after 30 days and we're not going to use it to train systems.

Shanif: So that's good enough for many organizations and if organizations need more, we can work with stricter APIs. [00:18:00] So data privacy is taken care of on that sense. Then you've got a malicious prompts or prompt injection is what it's called today. This is a bigger problem with chatbots that maybe sit on a website that any user can access.

Shanif: You might've heard the story of somebody who put like a car dealership who put a chatbot on their website and users tricked it into selling them cars for only 25. Well, and that's rough, but we don't actually put our chatbots on websites. What we do is we're an internal tool for your customer success teams to get information.

Shanif: And because we're working with enterprises, we tend to have better customers to deal with. And we also implement every request within the context of the customer of the customer success manager who's asking the question. So, for example, if you're making a request of the Salesforce API, that Salesforce request API gets done within the context of the user who's asking for it, so they can't get additional additional data that they're not supposed to.

Shanif: And then finally, um, there was the cons. Oh, man, I mentioned one other one, and I can't remember it quite. Yeah, the [00:19:00] LLM trying to do malicious things. Well, when you prompt an LLM, you have to give it very specific prompts and guardrails. So what we do is we have monitors in our system that check for malicious activity or incorrectly, uh, dangerous activity.

Shanif: So, for example, if we're pulling information from your database, we actually run it as a read only user and we do it on a production, on a non production system. And then we also have monitors in place to check for things like deleting data or dropping tables and we disallow those things. So what I'm trying to say is that every sort of point in the system at every juncture, there's things that you can do to protect against data, data, maliciousness and dangerous actions.

Shanif: And what we're trying to do is put those common sense systems into place as we're building our products. Of course, we have to work with our customers as we do this. For example, they're the ones who essentially provide us with read only access users or, uh, replicas of their production system. So, for the moment, you know, we're early, and so we're working with customers to implement these security processes.

Shanif: But they're the same sort of standard processes that you've seen in [00:20:00] software for a long time now, just applied to LLM applications a little bit differently.

Mehmet: Yeah, absolutely. That's very clear, uh, Shanif, to me. Now, another problem I know, and it was discussed a lot, You know, even here in the on the podcast, we discussed about it, which is the accuracy of the results that these large language models can give you.

Mehmet: Specifically, you know, everyone talks about hallucination. I know that there are now a new generation of startups that they just do this, uh, you know, RAG AI, which is the retrieval augmented generation, uh, AI. Um, but again, in the context of what you're trying to do. So first, let me ask you this because the data is.

Mehmet: Limited. I'm limited. I'm saying limited because it's the customer data, right? So, so they have their databases, their CRM, their email systems. Is there really a chance that still the LLM will do their hallucination thing or providing like wrong answers? And if yes, what, what usually are the measures that we can take to avoid such, [00:21:00] uh, things to happen?

Shanif: Yeah. Yeah. No, LLMs, you know, when you come down to them, they're smart. word prediction machines, and the way they've been trained is to almost always provide an answer so that the humans who are using them can be satisfied in their use. So there is always a chance that an LLM will make something up, but you can significantly minimize that chance, almost down to zero, by doing a few very simple things.

Shanif: The first thing you can do is, Just give it the right prompt by saying, um, when you are asked a question, only provide an answer if the answer is available in the context that I provide you below. And if it's not available, say you don't know. So by very simply telling the LLM not to provide an answer unless it can find the question, uh, find the answer in the prompt that it's given, you significantly reduce the chances of an LLM hallucinating.

Shanif: Uh, so that's sort of the main, main way to avoid hallucinations is by being very straightforward with the prompts and instructions. Then you just have to do a good job of providing the right context [00:22:00] to an LLM. And that's where things like RAG come in, or that's where things like database queries come in, where you can essentially allow the LLM to look up the information it needs and then analyze that information.

Shanif: So hallucinations are always a problem. Um, one way to get around them, like I said, is to tell the LLM. Only provide an answer. The second way is to tell it specifically don't guess or don't use your built in information And so the way you prompt these things goes a very long way The third thing that I recommend which is sort of outside of this process is if you provide an answer to a user to a customer success manager, for example, you should also provide the series of steps that the LLM took To get access to the, to answer the question.

Shanif: So for example, if you have to write a SQL query or the LLM to write a SQL query to get an answer, provide that query to the user so they can go in and double check things, uh, for themselves.

Mehmet: Yeah, makes sense. And the prompt here is, is very important, but I believe in, in, in what you're building currently, Shanif, do they really need to [00:23:00] take care about that?

Mehmet: Or it's something because you're integrating that within the API. So actually you're providing the, the prompt, right?

Shanif: Yeah. We provide. Yep, you're right. We provide the prompt there. What they need to worry about is, uh, making sure that the way that the data comes back from their systems is how they wanna present it to their users.

Shanif: Now, what do I mean by that? You might have a user who asks you, Hey, can you provide me a a, a usage report? Everybody who's active in the system and an LLM will be able to sort of pull the information from your database and without you having to tell it how to do that. But sometimes maybe it doesn't pull every piece of information you need.

Shanif: Maybe it only pulls the names of the users who are active, but you also want the emails and the number of times they logged in, for example. When you have systems like this where the LLM has to guess, then you as the user have to provide additional information about what information, about what sort of columns or context you want the LLM to provide.

Shanif: So it's still on your, it's still your responsibility as a user to provide enough context to the LLM. That it can answer your [00:24:00] questions properly. So it's not quite a hallucination problem, but you as the user still have to make sure it's doing what it's going to be doing. Um, now once it's done the right thing, once or twice, you can essentially give it instructions to always answer a question in the same way that it did.

Shanif: The first or second time where it was correct, or you can provide it instructions to say always follow these five steps when a user asks about this, uh, so you can start to build in the ability for it to be trustworthy over time. I digress. Um, essentially, to answer your question, you're right. Uh, you don't need to worry about that as the user because we provide the system prompts on our website.

Shanif: That's

Mehmet: fantastic. Now, if I want to put some as someone who would, would, would, uh, adopt any technology or any solutions, usually we, we put some metrics from, you know, I know like you are still in the early stages, but if you want to put some numbers when, uh, because you talked about also [00:25:00] helping, uh, your, uh, Your customers with their workflows and enhancing the customer experience and so on.

Mehmet: So if I want to put like some, you know, I'm in point A today and point B. So this point A to point B, is it, do you usually, I mean, do they measure it with. Um, how much more revenue did they get? Is it like the share ratio? Like how really, you know, I can materialize this by using your technology.

Shanif: It's a great, you know, I'm a data scientist, so it's so important to be able to quantify things.

Shanif: In the early days, though, it's very hard to quantify. So, Long term, the way that we are hoping to make an impact is by showing that our system can, um, increase upsells and cross sells and renewals and churn and reduced churn. And the way we, we might show that is exactly how you just said, you know, implement this product for some group of customers and then, uh, have it run for a certain amount of time and then show the lift in terms of, [00:26:00] uh, reduced churn rate or increased renewal rate.

Shanif: But like we're early. And so what we're trying to do at the moment is understand one. What's the accuracy rate of our system? So for example, if a user asks for something, how often is this is this system providing them with the correct answer? And how much time is that saving them? So at the moment, You know, you got, it's an operational workflow optimization problem, but down the line, it's much easier to sell a tool that shows revenue growth versus, uh, you know, productivity gains.

Shanif: And so it's one of those things where we're still working on it. Um, current current metrics are basically accuracy and time saved. Long term metrics are probably going to be net revenue retention lift. Down the line.

Mehmet: I love this, you know, because I'm not a data scientist like you, Shanif, but I believe in numbers as well, because it gives more, we can visualize that, right?

Mehmet: So this is why I like to, to, to listen to these, uh, use cases you just mentioned. [00:27:00] Now, a little bit, I'm shifting gear here from the deep technical thing to more about like building, uh, the company, Shanif, and you did it before also as well. So, um, Are you currently like, you know, have you bootstrapped the company?

Mehmet: Are you funded? Like, what's the status if I might ask, of course? Yeah,

Shanif: absolutely. I, you know, this is my, I guess my third, second, third or fourth startup, depending on how you look at it. Let's go with third. The first two that we did were heavily VC backed and the first one did very well and the second one crashed and burned.

Shanif: And this is my, just, you know, my learnings are, I think it's important to Bootstrap as long as you can. So what we're doing with this current one, the third one is I'm bootstrapping it as long as I can. The goal is to get somewhere between, you know, 30 to 50 paying customers. Um, As, as quickly as not as quickly as possible, because when you're bootstrapping, you have a little bit more time to get things right, but as efficiently as possible.

Shanif: Then once we have that, those customers there, we know that our [00:28:00] churn is low. We know that customers are using us. The idea is that we should be able to be able to repeat that. We should be able to raise money on favorable terms. Now we could have gone the VC route like I did in the first two companies, but what I've learned is if you raise money too quickly, that basically starts the clock on you and finding product market fit is something that doesn't have a set defined amount of time.

Shanif: It can take a little bit of time, so you might run out of money and then you might have to go raise more money. But if you haven't shown that traction, then you have a lower chance of raising money. So we're bootstrapping as long as we can now, you know, obviously down the line, we're going to need to raise something to grow quickly.

Shanif: But I think we are, we have a little bit of ways to go before we get there.

Mehmet: I like this because, you know, you're just waiting the right moment, right time to do this. And just, you know, I asked, I like to start to ask these questions because I get, um, you know, discussions with first time founders, especially the first time founders.

Mehmet: Uh, that they are in hurry. And you mentioned this and like the other day I [00:29:00] had the same discussion with that on another episode. Again, the same thing was mentioned to me that I'm not in hurry, you know, like it was with Gucci. So he told me I'm not in hurry. I like this. He said, I want to build it the right way.

Mehmet: And I don't want to burn myself because for first time founders who are listening to us, your clock ticks same as Shanif mentioned, and you have between 12 to 18 months. So either to make it positive, uh, of course you might raise money later for scaling. And this is what I want to ask you Shanif about like, you know, the scaling phase.

Mehmet: So, so of course now you're bootstrapping. Now, when, when it comes to, to healthy growth, what like also like kind of, um, Lessons or strategies you found yourself most effective when it comes to to scale and especially in a competitive market.

Shanif: Yeah, there's so much here. I would say, I think the main things are understanding the levers that you can pull that are working.

Shanif: So for example, if you [00:30:00] are seeing a lot of growth from one or two channels, then double down on those channels versus trying new things and really exhaust those channels. So for example, we currently are probably going to be a B2B sales company identifying and targeting sort of enterprise and mid market sales, mid market SAS companies going down the line.

Shanif: And what we want to do is make sure that we can. Make that sales B to B sales process as efficient as possible. So we want to focus all of our time and energy on that, which means we might have to forego or delay trying new tactics or topics like I don't know, market like inbound marketing or even things like conferences.

Shanif: Now, I don't know for sure what our go to market strategy is going, what the best go to market channel is going to be for us. But to answer your question, I met the thing that works well when scaling is doubling down, tripling down on the one or two channels that work really well for you and making that as efficient as possible.

Shanif: Um, so if that means in our case, if that means we have to hire [00:31:00] a bunch of enterprise sales reps, then, then there's a reason to go and raise money. But we do that after we've identified our payback period. And we do that after identifying our sales. And we do that after identifying our average contract value.

Shanif: And so then you have the ability to spend 2 to make 5. My point is find what works and double down on that. And then only after you've exhausted that, which takes some time, five years, six years, only after you've exhausted that, should you start thinking about either new strategies or new products to grow from there?

Mehmet: Absolutely. And again, Shanif, I'm asking you this because you're doing this to your point for the third time. So how important also to have kind of, and I like to call it the end in mind. So, um, yeah, yeah. So, so how, how important to have the vision and, you know, keeping this because your first startup, it was sold to, to Twitter X now.

Mehmet: So like, do you, do you advise people to have this from day one to [00:32:00] visualize where they want to reach, uh, before, or, you know, keep it as it goes, keep it, keep it with the flow as they say, which approach you prefer.

Shanif: Yeah, you know, as I've, I've done startups for 10 or 15 years now, and my, my thinking on this has evolved.

Shanif: If you had asked me when I was younger, you know, what do you want to, like, what should, what should people do? I would probably have said, try to grow as fast as possible. And then that gives you a lot of options. And if you can sell or exit, great. To a certain extent, I still believe that any, any startup you can walk away from is a, is a good thing.

Shanif: Good exit, but not everybody builds startups because they want the big exit or the big IPO. A lot of people want to be able to live a good lifestyle and have control of their company. And you can make, you know, 20 or 30 million a year and still own 80 percent of your company and have a great lifestyle and do really well.

Shanif: So my thinking today. Is have a very strong understanding of why you are trying to start a company. Is it so that you can have that ownership and freedom? Is [00:33:00] it so you can make the money? Is it so you can help people and keep that in mind and always work towards whatever that is, because not everybody wants to have an exit.

Shanif: Um, and that's totally fine. Like for me, I've. I'm at a point now where I want to grow a very large business that survives 200 years into the future. And in order for me to do that, I have to work differently than someone who's going for an IPO in 10 years. And that's totally fine for me, totally fine for them.

Shanif: And it's, it's all very personal, I think.

Mehmet: Absolutely. Yeah, it's, it's good. And to your point, And this is, you know, there's the famous, uh, you know, video from Steve job when they, he mentioned that if people comes to him and say, we're starting up just to make money and the lifestyle alone, like they probably they're going to fail.

Mehmet: They need to have, you know, this I like to call it a purpose. Some people, they call it passion, um, to, to build thing. Of course you can make a lot of money down the road, this a hundred percent sure, but yeah, like don't just focus on, you know, uh, Um, let's say the material [00:34:00] part of it, it should be like have something more as you to your point.

Mehmet: It's also like something very, very personal. Um, like you mentioned about the second startup and you mentioned about some failures. So, uh, so what is, you know, the best approach to do when things go in a way that you didn't plan for it? Like, what is your experience?

Shanif: What's funny is there's so many things that can go wrong.

Shanif: There's way more things that can go wrong than go right. And so you have to be able to adapt. In our case, we did a few things wrong. We, um, we raised money before we had product market fit and that put a clock on us. We didn't have the right team in order to go to market. We also weren't sure about the problem that we were solving.

Shanif: Once we finally figured all those out, we only had about six months of runway left. So despite the fact that we were growing 600 percent in our last year, we hit a lot of forces that Cause this to fail. So how do you deal with this? Well, best approach is to not get [00:35:00] into those situations to start with bootstrap for as long as possible, stay lean for as long as possible, understand your product and the problem you're solving for as long as possible.

Shanif: But what happens if you get into this trap? I would, for me, it's been very important to maintain my integrity. So once I realized that we were. In trouble, we might not be able to raise money. I communicated to my investors as quickly as possible. I told them the situation eight months in advance. I told them, here's what we're trying to do.

Shanif: We tried to raise another round. We tried to sell a company. We tried to figure out if we could raise a bridge. We tried to figure out if we could do a few different things. So being very transparent and maintaining your integrity is really important, I think, because that allows you to grow and make sure you don't burn any bridges.

Shanif: Um, Doing everything you can, making sure you're taking care of your team. For me, once the writing was on the wall, I wanted to make sure my team had enough, um, upfront knowledge about what's going on so that they could start looking for other jobs, and so that I could start maybe helping them in their job search.

Shanif: And [00:36:00] to this day, you know, I've written some references and recommendation letters, and I've done a few things in that, in that sense. So, once you realize that things are not looking so great, In addition to doing everything you can to solve that, once you realize things are not going to work out, then trying to make sure everybody is as taken care of as possible.

Shanif: We actually even returned investor money because we realized that the amount we had in the bank wasn't going to do anything for us. Um, and it didn't make sense for us to burn the rest of our cash. Um, maybe the investors could do something with it. So I think it all comes down to maintaining integrity and taking care of the people around you.

Mehmet: Integrity, transparency also as well, uh, Shanif, I think. And, you know, to one thing you just mentioned now, it's also triggered something in, in, in my mind about, you know, the team and how much important also is to surround yourself as a founder with the, with the right team. So, um, any traits, maybe, you know, founders that they should look for when, when they [00:37:00] do the hiring for.

Mehmet: Their, their first hearts, at least, let's say.

Shanif: A lot of folks have thoughts about this. Some people say hire for attitude. Some people say hire for the ability to learn. For me, I think one of the things I've learned is the first, let's say 10 people in your startup. They all need to work really well together.

Shanif: So, you know, I've got like this, no jerks policy, make sure everybody can get along well. Make sure everybody can learn and is, uh, is interested and excited about what they're doing. And. Of course you need to make sure that they're good at what their functional roles are. Like if, and if I'm hiring an engineer, I got to make sure he can code, but that's almost a given, like once you understand that people can do the job that you're hiring them for, there's a lot more you want to look for.

Shanif: For me, like I said, making sure that. People can work together in a team, uh, making sure you're not hiring any jerks, making sure that you can hire folks who communicate well and learn well and can adapt well because startup early stage startup life is not like corporate life. It's [00:38:00] not like anything else.

Shanif: There's going to be a lot of downs, a lot of ups, um, salary. I mean, you can't pay salary for a lot of folks if you're not bootstrapped, if you're not VC funded. And so you have to find folks who are okay with scraping by, but also still enthusiastic. So. It really comes down to, again, this is a sort of a personal thing.

Shanif: Every founding team is different. Every founder looks for something different. And for me, I'm looking for folks who can play well in a team who've got the right attitude and who are excited about growing a business for the longterm. And that reflects in multiple different characteristics, like I just mentioned.

Mehmet: Cool. Now, one thing, you know, before we start to wrap up and because you are again, a data scientist, Sadif, so, um, How important to also act fast when we see, uh, something going wrong, like, you know, maybe, maybe we didn't have the traction that we expected. Maybe we are not getting enough, um, I would say demos booked, uh, [00:39:00] because one of the topics that comes, but I, I want to ask you this from both a, Data scientist perspective plus founder.

Mehmet: Also, how much important is really to, you know, have kind of a North star and to get out of the ego? Because, you know, we hear a lot of stories that, you know, the founders, they get so stubborn. They said, No, we can't be wrong. So how important also to have this agility in accepting that? Yes, it's We might have our hypothesis wrong and we need to pivot.

Shanif: I would argue that's, that's everything. A startup is not a business. It's a, it's a way to try to find a way to make money. Right. And so in our first startup, we pivoted 13 times in my last startup, I pivoted seven times. I'm sure I'm going to pivot six or It's all it is, right? You don't, it's not about the idea.

Shanif: It's not about your original thinking or the hypothesis. It's about finding a way to make repeatable sales from a particular service or product that you're offering. So those [00:40:00] founders who have an ego, who basically gets so tied to their idea and their product aren't building a startup, they're just, they're just.

Shanif: And so startups are all about sort of testing hypotheses, validating them and making sure you can get the right answers from your experiments. Now that's much easier said than done. I've gotten caught in the trap of not pivoting quickly enough, but you have to be open to being wrong. I mean, you're going to be wrong.

Shanif: 99 percent of what I've done has been wrong. Um, and so you have to be able to take everything. all the feedback you're getting and make adjustments. Now from a data science perspective, the problem is in the early days, you don't have data. And so you have to be able to make strong conviction responses with very limited data, which is hard for folks to do, especially hard for data scientists to do.

Shanif: But once you get okay and acclimated to that concept, then you can, uh, make the changes you need to make.

Mehmet: Yeah. So, so you need at the beginning to rely on your Got feeling, as they say, I believe a little bit,

Shanif: a little bit, a little bit. You should [00:41:00] still like what we've done is we've collected about 70 user interviews.

Shanif: And then before we even built anything, after hearing from those users, we put together a prototype that took us a day to build and we put in front of them. So you should be collecting data along the way, but it should be, yeah, it should be structured.

Mehmet: Yeah. I love it. Like this is again, something, and I'm asking these questions to any of To enlighten other founders, actually, uh, download.

Mehmet: Hopefully, hopefully they, they will get benefit out of it. Finally, Shaneef, thank you very much for, you know, the time today. Any final things you want to add? Maybe something I didn't touch base on and also where people can find more about you and your company.

Shanif: The, yeah, the founder journey is, uh, it's very difficult.

Shanif: You are probably going to fail many times before you succeed. That's okay. Um, I have come to believe that. If you just stick with things long enough and you continue to give it your all for five years, 10 years, 15 years, you will reach success, but it is extremely difficult to stick with something that's not [00:42:00] working.

Shanif: And so it comes down to mental willpower. So just stick with it if it's something you really, really care about. So understand why you're doing something. And if you really care about it, stick with it. Um, folks can reach me at, uh, my email is Shanif S H a N I F. Locus, L-O-C-U-S-I-V e.com. Also on LinkedIn, just for search for sif.

Shanif: I'm probably gonna be one of only four or five, so feel free to reach out if I can help with anything else.

Mehmet: Thank you very much, SIF. Of course. I will put the, uh, you know, the links in, in the show notes and really it was an amazing, uh, you know, discussion with you today. Thank you very much for the time and all the use cases.

Mehmet: We talked about AI and how it can support SaaS companies mainly. driving revenue, enhancing their customer experience. And, you know, my close friends who use in customer success departments, also this is for you. Thank you very much for sharing this and your founder journey or so as well. And, you know, this is usually how I end my podcast episodes.

Mehmet: This [00:43:00] is for the. Uh, for the audience. If you just discovered this podcast by luck, thank you very much for passing by. I hope you enjoyed it. If you did so, please subscribe. You are available on all podcasting platforms and you can also share it with your friends and colleagues. And if you are one of the loyal followers who keep coming, sending their messages.

Mehmet: Suggestions. Thank you very much. I really appreciate it. Keep them coming. If you're interested to be on the show, you know how to reach out to me. I'm available mainly on LinkedIn. Write me what you want to discuss. If you have a startup idea, you know, you want to have some visibility or maybe just to have something you want to share related to tech or startups in general, reach out to me.

Mehmet: I would love to hear from you. Thank you very much for tuning in. And we will meet again very soon. Thank you. Bye bye.