Future-proofing the humans behind the tech. Follow Phil Gamache on his mission to help marketers level up and have successful careers in the constantly expanding universe of martech.
Philippe Gamache 0:07
What's up guys, welcome to the Humans of martech podcast. His name is Jon Taylor. My name is Phil Gamache. Our mission is to future proof the humans behind the tech so you can have a successful and happy career in marketing. What's up everyone today we have the pleasure of sitting down with Michael remanso, co founder and CEO at narrative bi. Michael started his career as an electronics engineer and then a back end software engineer he dived into web dev database management and API integrations. He also took on the challenge of being CTO at an IT startup called flat logic based in Belarus. He then moved over to San Francisco and founded a web and mobile dev consultancy which he ran alongside co founding a natural language search platform called friendly data. With the mission of democratizing access to data. He went through 500 startups, a VC seed fund acceleration program, and friendly data was acquired by ServiceNow in less than three years, and Michael went on to join the company in a central product role to help develop their natural query language AI tool. He's also an investor@founders.ai, a Startup Platform for disruptive SAS products. And his latest entrepreneurial endeavor is narrative bi, a generative analytics platform that helps growth teams turn raw data into actionable narratives. Michael are super excited to chat with you today. Thanks so much for your time. Yeah,
Speaker 1 1:32
thanks for having me. I've listened to many of your podcasts and excited to be here. Nice.
Philippe Gamache 1:37
Appreciate that. This episode is brought to you by our friends at knack. launching an email or landing page and your marketing automation platform shouldn't feel like assembling an airplane mid flight with no instructions. But too often, that's exactly how it feels. NAC is like an instruction set for campaign creation for establishing brand guardrails and streamlining your approval process to knacks no code, drag and drop editor to help you build emails and landing pages. No more having to stop midway through your campaign to fix something simple. nak lets you work with your entire team in real time and stops you having to fix things mid flight, check them out@naqt.com That's kn a K and tell them we sent you. This episode is brought to you by our friends at revenue hero, I can't think of anything worse than finding out a lead waited a week for a response from sales. That's why we recommend revenue hero, it's the easiest way to qualify leads based on Form Values or enrich data and route them to the right sales rep. Their product is packed with a bunch of behind the scenes superpowers that ensures qualified leads are assigned to the right reps following your custom round robin rules and sending key data back to your CRM. That means more qualified meetings for your reps. We all know they want more of those. But more importantly, no more waiting time for your potential customers. They back all of this up with the best product support out there offering 24 Five support on Slack Connect for all customers, no matter your pricing plan. So if you want to three extra conversions with the same traffic, go to revenue hero.io And tell them we sent you your sales team will thank you for it.
Jon Taylor 3:19
Yeah, we're excited to have you on so let's just dive right in. We had another founder on the show whose advice for aspiring founders was to quit their job as soon as possible and just jump right in. Curious if you agree with this and how do you know like, when to take that that leap? I know you had a tour of duty working as a consultant, but like at what point both in your professional life and in your personal life do you know it's like it's time to go live
Speaker 1 3:43
there's no right or wrong approach. Every station is individual Of course it's easier to start a company if you're 20 Something dude you can live in some collision with other founders but if you have a family it's a different situation. But what what is common is that at some point you need to focus 100% Because without the skin in the game, it's just impossible to build something big but if you have no skin in the game how are you going to hire other people and how you're going to convince them to work for you or if you have no skin in the game like Angel Investors will not invest in your in your company their own money and skin in the game can be expressed in a different ways it can be like living your full time job in my case when when I started narrative die I invested my own money into this venture but also for me it was a huge opportunity because because I with my support my former employer and I left behind seven figure compensation on the table which was Harris you plus based Basecamp calm so skin as the game is required and if you want to build something, but you need to make this transition at some point. Yeah, it's
Philippe Gamache 4:57
easier said than done. It takes Uh, a lot of courage for sure. So, yeah, I know there was definitely an opportunity cost there, especially for you knowing that like the company acquired your last company. So there's there's probably some, some golden handcuffs there like you kind of shared. But you also shared on LinkedIn this, like fantastic advice for other founders. But I thought that it was also applicable to people that work in startups, like one thing that you said was kind of a breath of fresh air for me, coming from a founder and CEO is that you shouldn't expect your team's dedication to match yours, they hold a much smaller equity stake, obviously, oftentimes, the expectation to work in a startup is like 12 plus hours to grind, you get a bit of like stock options. And that's why the expectation is a bit higher, smaller teams, you need to like pick up the slack. But along those lines, like I was curious, your take on employees at your company having their own side hustles, Shopify CEOs, notorious, for example, for discouraging side hustles. And I'm curious to ask you, like, do you require unshared attention from from your team and what your thoughts are there?
Speaker 1 6:07
Oh, yeah, honestly, I don't know how many hours they, my employees work. And it's, it's none of my business. So it's fine as a contribute. And our first one as they perform their, their duties. I'm happy with that. In fact, I encourage side hustles. And we're a small team with just eight people, but five of eight people either started companies previously, or not necessarily startups, but some kind of thought business, or has already have side hustles, or some some income stream right now. And I think it's, it's actually a good thing, because they can bring back some outside expertise. And it can be beneficial in some, in some, in some cases, or into at some point, if for someone decide to quit, well, I will be proud to if someone starts something new, a new company, I will encourage you so. So I think as an entrepreneur, I can understand that. And I can all encourage
Jon Taylor 7:05
a lot of that answer. You know, one of the things that we want to dive into and kind of one of the main reasons we wanted to have you on the podcast was to get into this business intelligence space, like the business intelligence space is a fascinating one, and for many years is provided, you know, folks with tons of opportunity, but I still think there's a lot unsolved in this space. One of the big things from film and myself, like we had a background working at a bi startup called Klipfolio, local Canadian company. And we were really focused, particularly when Phil and I were there. Were on the dashboards component of it. Narrative bi, however, like Phil and I both signed up for a trial. And we will dive into some of the things that we've we found through the trial. But one of the kind of cool things that I've found so far is this idea that narrative BI is surfacing these insights is almost out of sites to you through these alerts, it doesn't feel like I need to go in and view a dashboard like in a cockpit where I look up and I'm like, Oh, look at all the things that are that are going on. Instead, narrative bi seems to be bringing insights to me. What are your thoughts on the future of there's a big question, what are your thoughts on the future of the space? And what about dashboards? Are they a dying breed?
Speaker 1 8:15
Yeah, great question. Well, there is not going anywhere. It's so short to build a market and it's still rollin. However, the problem I have with BI is that it's designed. It's mostly designed for data people. And if we consider like most advanced tech companies that uses the most advanced data solution, or the adopting of for BI will, they might be 20 to 25% in this organization. So what about the remaining 8%? What about less less for technical teams, I believe that every person, every knowledge worker should be able to access data insights and data. Insight, knowledge shouldn't be limited by data knowledge or technical model your side beliefs, or the biggest opportunity on the market is to enable previously untapped by that people enables them to get insights. It could be automated insights or some AI generated insights. But the point is to give access to data insights to people who never never used bi, but at the same time, they make a lot of business decisions based on the data. And speaking of dashboards, I personally used dashboards for some of some use cases, but the problem was dark, but most of dashboards are still static. So you have a set of predefined metrics a set of predefined queries, but but if you consider modern organization modern business, the data is changing every day and market dynamics is changing every day. However, the dartboard remains static and Some, sometimes you lose a lot of interesting hidden insights in in the static approach. Firstly, the biggest opportunities to make insights and analytics, ultimate sick, AI powered, and easy to consume. As soon as this is the biggest opportunity. And if we, if someone can enable previously untapped audience, this opportunity is bigger than the AI itself.
Jon Taylor 10:23
I think that's a fantastic insight into your to your head and where the space is going. You mentioned something around knowledge workers and the adoption of business intelligence and analytics tool. This is something that totally jives with the experience that I had working in an analytics company for many years is that the data people love what they build, but getting, you know, the knowledge workers to adopt it. That's really difficult. Now I like what I've seen so far with narrative bi as approach using AI to surface these insights to folks like, there's just always so much data, I think this is probably one of the biggest challenges for knowledge workers is that the volume of data is massive. What do you see as AI is role in helping to democratize data for these knowledge workers? And how do you think that'll influence adoption of BI tools in the future?
Speaker 1 11:16
So first of all, I'm I'm a data nerd myself. So I personally enjoy using di but the problem was even more than natural language query tools that in order to effectively use them, but you need to have some level of data knowledge, like you need to understand how the database is structured. What is the table names, column names? How are they related to each other? And data people are knows that but even if you use like most advanced natural language query input, you need to know how to ask questions, you need to have some some level for understanding. And I don't believe for a lot of non technical people, like markets are so few executives possess this knowledge. So I don't believe that the chart approach itself will solve the adoption, adoption problems. However, I think the opportunities to try to answer questions before the court, people ask them and generate insights more proactively based on previous history based on some personalization. And this is where, where AI can help by protests and a lot of sort of personalized data, modern AI enabled solutions can surface insights that personalized for specific team for specific role, or even for specific person. So this, I think this is this opportunity is even bigger than our natural language query, which which became relatively easy to implement with the development of for large language model.
Philippe Gamache 12:54
Yeah, it's super interesting. You mentioned a couple of things there, I want to pull a thread on. I agree, I think like most marketers don't, don't have like the sequel jobs to go into Looker, Annelle and build the type of stuff that they would want to without like an analysts or a data engineer coming in and helping out. But you've like, your whole mission is that like analytics tools should be simple enough so that everyone can get insights, but not necessarily go in there and build like all potential dashboards that that they could. But while this is somewhat easy to say, for maybe like single sources of data, like GA, or maybe like product analytics, oftentimes, like you said, the power of BI and these data experts is in the pipeline building and understanding how the data is structured, joining different tables together and combining some of that data. Do you think that will ever solve that piece of the puzzle like a world where not only like getting access to insights, but building pipelines end to end and joining data is simple enough that anyone can do it?
Speaker 1 14:05
On I think I can ask for or with some given some background, so my personal experience. So when I was up at ServiceNow, I worked closely with the BI team and ServiceNow was a huge organization. And I talked to the data about their challenges about their daily work. And they told me that there are three types of requests they deal deal with them. First type of requests or requests is common from HR managers, executives, different marketing growth teams, and it's pretty basic like what the what the average revenue per user is some product line, like very basic question for like you can write an SQL like in a minute to get the answer. The second type of questions is more advanced like you mentioned, which involves some data aggregation, some joins or some More complex stuff, which can pull might be an hour or a couple of hours to clean the data to pull it to aggregate. But still, it's it's not it's relatively relatively easy until every day to perfect pumpkin can do it. And the third type is more more advanced, like you need to do some research, you need to prepare data, you need to gather all this data, eventually, you need to prepare some slides or outline some insights to do your deep deep research work with actual insights. And it can take maybe even weeks to, to prepare. And I think the first time most simple and second can be can be automated. First one, if you can ask a question to your data team, you should be able to ask it in AI interface. Second, my demo more complex, but it's technically technically doable, even charge Bootsy can write SQL with, with joints. And we actually solve this problem at friendly data. The third, the third one is more complex. And this is where I believe the BI teams can bring most value. And I think if we can automate more basic tasks, it will be a win win for everyone. Because right now, most of for BI experts spends their time on addressing the clock and answering a simple question they take a few minutes to, to answer but when you have dozens or hundreds of such questions every day, it becomes a nightmare. And then people have asked this question then unsatisfied, because racing, it's a simple question. Why take like few days to answer because of the backlog. So if we can automate the simple or relatively simple stuff, it will delve into them. Because business people will get quick answers. And the AI people can focus on more, more advanced, more impactful thing.
Philippe Gamache 17:01
That's really interesting. Like having used narrative BI for a couple of weeks now, as some of the insights that we've received through like the product lead email alerts aren't necessarily like questions that I had off hands. I'm like, while you were talking there, I was wondering if tools like narrative bi coming on the scene paired with bi are actually going to like save time for these like data analysts, because growth teams and knowledge workers are going to be going down these rabbit holes based off of the Insight emails that they got. And they'll be less focused on asking those like pie in the sky type questions. And maybe the backlogs will get smaller in that sense. But yeah, curious, your your take there on on this idea of instead of like growth teams coming up with questions and like, we wish that we'd be able to monitor this or have these answers. They're focused on actually doing campaigns and growth stuff. And the insights come in from the product lead stuff. And so like maybe it saves a bit of time from bi teams, curious your take there?
Speaker 1 18:09
Well, most of the run in been curls and trying to understand how your company work. And at the end of the day, marketers run those campaigns. So it's important to understand what performs best and where you can probably cut your spending or not effort. But in order to in order to solve this conflict, we we plan to what we're working on. So we recently launched LLM recommendations, which is one button you don't actually have to ask questions, you just click one button and get all of them recommendation. And we try to provide additional context trying to explain why, why Samsung happened trying to suggest some next steps. So you don't actually need to dig deeper, we try to provide this actionable recommendation. The next step for us is to is to to make it more conversational. So you can ask some follow up, follow up, follow up question. So again, I believe that that seems secure just thinking about they just need to focus on a more more advanced, more impactful
Philippe Gamache 19:18
This episode is brought to you by our friends at customer IO over sold them out legacy marketing automation platform that is still struggling to update its user interface. I've done a tour of duty with all the major marketing automation platforms and many are definitely similar customer I O is the most intuitive and beautiful platform. I'm talking about the industry's top visual workflow builder to design and implement your unique messaging strategy. Powerful A B testing features inside your workflows not just on subject line sense. Hold out testing functionality to see the incremental impact to your messages to draft mode so you can QA messages and conditions in production with real users before anything is sent. Copy workflow items so you don't have to repeat the building process again and monitor camp Things tests and keyless membership growth from your personalized dashboard. The icing on the cake marketers using customer to have seen a 20% increase in conversion rates from strategic messaging. So stop using clunky old tools and adopt a multi channel approach that creates joyful interactions with your customers. Start a free trial without a credit card customer data yo and tell them we sent you. This episode is also brought to you by our friends at census the number one data activation and reverse ETL platform left by Activision Canva Sonos notion and more. As you might know, I'm pretty opinionated that the future of martech is composable. And that the single source of truth for your marketing data should be your data warehouse. Since this helps marketers solve an age old marketing problem getting real time complete access to your customer data without needing to write a line of code. Also, if you want your own face as a humans of martec style image, we're doing a fun monthly raffle with census for a personalized t shirt, enter to win at get census.com/humans
Jon Taylor 21:05
One of the kind of questions that popped into my head as we're, as you're talking, there was just this idea. I know one of the trends and business intelligence now is around the semantic layers of you know, having additional information around the data itself. So dressing up the data with additional context. As we talk about late knowledge worker adoption in the given play between the data teams and the knowledge workers. There's a lot of context, the knowledge workers have like getting the alert that Phil and I got was around pageviews, we saw a pageviews spike on one of our podcast episodes, immediately. I'm the GA for experts. So Phil got me to go in and take a look, I was triangulating data that was coming back as like, there's a feedback process here that the next time that alert comes in, if we had a ability to input information or additional context, we start to get more of this kind of feedback cycle that gets better and more refined. How do you see trends like the semantic layer, involving knowledge workers and developing these contextual datasets that kind of help organizations master what they're working with and getting more value out of their data?
Speaker 1 22:13
Yeah, I think credit multiplayer is the necessary part if you're going to build some kind of natural language query tool, because without it, the accuracy will be not not high enough to trust the system. Because every business is unique. Every business has a different data structure. Every business has different business rules and every business count, even conversion differently. So what can be conversion for one business is not counted as a conversion for quaza. And right now with semantic layer is unnecessary, unnecessary evil, evil, because it requires a lot of time to do or to set out to assign like names for different column. And we are trying to, we're not trying to solve this problem, we tried to avoid this problem, how we work with data sources that are quite popular, at the same time, they are well structured. For example, let's say J four, J, four is being used by 10s of millions of websites. And at the same time, the structure is standard. So am KeePass. Also pretty close among our customers. So why buy by setting some meaningful defaults we can avoid, avoid denial with customization for every customer. And same goes for other data sources we support like Google as is a pretty popular is literally among the polling and same goes for Facebook ads and other data, data tools. But but also we allow or some level of customization, you can introduce your custom metrics. And it's somehow overlapping with this concept of segment, semantic layer. Where are we where I believe an opportunity is to personalize with personalized experience based on engagement with with a style. And we have a large, large buttons like you your life that you can like this particular insight, or you can press you don't like it. And based on this, we will we will turn on our machine learning algorithm also on the system to suggest more relevant stuff to your next time.
Jon Taylor 24:25
Yeah, that's fascinating, really cool. Take on that, you know, kind of building off of that I wanted to ask a question about the competitive edge of proprietary data. I know you've mentioned like a key advantage for organizations in the AI space as their unique data. We're talking about like the context around the data, the semantic layer, like massive research or investments in your organization's data are hard, but specialized AI with proprietary data can excel in these niche areas. The one use case that came to mind is leveraging proprietary data across your user base in order to offer additional insights and benchmarks for like customers within similar industry. So like propensity to buy or something like Applecross across his niche. Just love to have you unpack this a little bit with some practical examples.
Speaker 1 25:08
Yes, sure. So we're in a paradoxical situation right now, because only a few organizations in the world can afford to build a foundational LM model GT four level model, because of compute because of resources required. And it's no surprise that in order to build Tamsen, new companies raise hundreds of millions in in Sidhant. On the one hand, you would you need a lot of resources to build something new, like from day four model. On the other, it has never been easier to build an AI. You can use open source models, you can use open AI API's or something similar. And it's super easy to build a basic AI app, you can build a custom VPC in a few minutes. So, but the question is how to differentiate on the server environment, or from the sensor one hand, you have to differentiate against the generic solutions, like, charge up to from Zaza. But you need to differentiate against hundreds of phone startups that pop up every day, using using the underlying technologies like API. And the answer it's data as AI becomes a commodity and as technologies become a commodity became a commodity. That's the only way how how to differentiate this by using proprietary data. And you can use the data in my in many ways, for example, to find some the model itself to inject some proprietary data into prom you can use it for post processing, to give you a particular example how we use it, we use proprietary data for benchmarking so we can suggest like insights based on based on customers in this segment how you perform again also we use it for for prompt injection. So we we built from tailored for specific customer or for specific market segment. And this has been proprietary data is the probably the only differentiator on the market. But it's not all not only about generics if if you want to outperform any AI system like voice recognition, or or image recognition or natural language processing, you need to have domain specific proprietary data. And if you build something verticalized some senses dominance, you can now differentiate that easily by having access to some interesting data that is unaccessible to other player.
Jon Taylor 27:42
It's fascinating. I mean, one thing I want to dive into really quickly before Phil takes his next question is just around this, this idea of like using chat GPT for data analysis, right. And you mentioned about differentiators. Like for Philippi on the podcast. Sometimes we'll plug like screenshots into GPT and see what it spits out for insights. And yeah, it's pretty mid to be honest with you. How do you guys a narrative bi develop a differentiator with your with with your integrations with AI?
Speaker 1 28:12
Sponsoring this question I wanted to consider it from life high level point of view. Well, in order to understand the possibility of an AI system, you need to understand our limitations first. And I personally use tragic Bitsy every day, but for for for tasks that are capable to do better than I do or Magnolia or like some other tools. And I use it for summarization, because it's really good for summarization, you can load a bunch of for CSV files or some unstructured text and summarize it you can write so text when it comes for for data analytics use case, huge problem alum systems have with data managed services that they hallucinate a lot. And this explained by design by design column systems, they are your sense why just to give you some answers, even if they don't know the answer, even if you don't have underlying data, they will come up with sanctions. And I personally cannot trust such tool because I don't know if the answer is correct. And even the tragic itself this your every time gives you a disclaimer that you have to verify that your data you cannot rely on it. You should verify it if you make a decision by based on it. So so this trust issue I think is the barrier for using AI so data analytics allow them for data analytics purposes. And in order to solve it. I think all open problems need to invest some resources into post processing and to do it so purposefully. I don't know if this is purism is a step roar for them. So we solve this problem by not using AI. In the Insight generation part, we use our own proprietary technology, which is machine learning based. What we use AI for is for summarization call on set for some text or natural language processing stuff. And for some formats in our output small, more or more, more precise, and even when we use it, we do a lot of post protests and two microphones to make sure the data is accurate.
Philippe Gamache 30:34
Very cool. Yeah, definitely tracks with my experience using chat GPT for trying to like get insights out of a chart or just like uploading a CSV in a custom GPT. It's been really good at summarizing data and trying to interpret some of it like kind of hit or miss there, but usually pretty solid. But like getting insights out of it. Like especially when there's like a ton of context and nuance around the data. It's just like, you have to spend so much time training it on, like what this data is before it can get anywhere useful anyways. But I think that like having used narrative BI for a couple weeks now, like we said, one of the things that was interesting from the insights alerting that we got was actually like, kind of insights, but also in this category of like anomaly detection, like being able to see like, oh, whoa, like this, this metric that we're tracking in GA is way, way off track right now, or it's like way higher, probably bought traffic, like, which was the case in a couple of times when we did investigate it. But we actually had the founder of metaplasia. Earlier on this year, their focus is a bit more on data teams and pipeline anomalies. But I know that like for your product narrative bi, you've recently rolled out this idea of anomaly detection talk to us like why marketers need to care about this, like we had met a plane on to, like, try to just like get outside of just like data team should care about this, like marketers should care about it to be said something interesting about this idea that anomaly detection can also help marketers uncover untapped opportunities that maybe they haven't, like, kept a close eye on.
Speaker 1 32:19
Oh, yeah, that's actually a simple question. I have so many examples, like literally up to date. And when I say more, our customer or reached out to us and I will say, like, you know, you saved my day, it's not narrative, I would not have a conversion drop. Apparently, there was some landing page where some form didn't work. And we sent sent them an alert right away the conversion drops, they take a look at them, they investigated and apparently some some forms didn't work. But there are many examples like this, for example, it should track your, it should connect your Google Search Console data source. And you can track how your page pages are indexed on Google. Because Google is unpredictable, its algorithm is unpredictable. And so you can model the some page dropped in the index significantly. So we can notify about such such stuff or poor advertising, when when you have significantly increasing CPC. It's probably Samsung to look to look into. But there are also positive examples. For example, sometimes I use narratives that form mostly 4g For and Google Search Console is cases, we don't advertise a lot. So those primary data sources we work with. And when I when I see some interesting pattern in like tragic common pain and similar source, or I always look into it, because sometimes it could be a publication in some blog post that goes or viral. And it could be an opportunity for you to engage with those people to to highlight it somewhere in your social circle. So there are also positive examples of anomaly.
Philippe Gamache 34:03
Very cool. I wanted to ask you, like I was curious about this. Like, we talk a lot about like some of the practical examples were like CMOs and marketers, and I know like, so far, a lot of the integrations are growth slash marketing, focused data sources. I love the strategy that you took with narrative bi and like focusing on one of these verticals and going out to market with them and really like nailing it with with graphs marketers, like I was curious to ask you why you picked growth as like one of the first verticals like John and I wouldn't we were at folio. One of like, our struggles, I guess, was like figuring out what was that vertical that we should go after? So many verticals? Like people across the company, all departments can use data, right? Like some of your examples were around HR, like the product team cares about data too. So like, why did you pick growth and maybe talk about to curate your next vertical, are you always going to focus on growth? Just curious?
Speaker 1 35:05
As one great question, I wouldn't say it was easier to focus on something because it always tempts him to build generic content generic. And this is probably on fake, I might was friendly data, probably at some point. Well, we eventually realized that we need specialization, but probably it was too late. was so it was hard learned lesson from my previous experience and why we picked markets. And because so die product is for focusing on share in your data. So every time something interesting happened, something unusual happens, or we generate a narrative about it, we notify about it. And the marketing use case is particular. Interesting, because when markets and data is changing, especially the one your key metrics, like conversion, revenue, or new users, changing probability means something for business that went south for top line of your business. And it's just easier to explain our potential ROI when you deal with this kind of data because it's always about about money. So this is why we started with markets and girls use cases as at an entry point. But our our reason is to make analysis the I single source of truth forever, model work worker. Recently, we added integrations with CRM systems, we look into product analytics, and we plan to expand it to operational accounting use cases sounds the future. So I think it's a big opportunity.
Jon Taylor 36:45
I think with you building this product for marketers, you've probably end as a CEO as well, trying to grow your own business, the idea of attribution probably comes up quite often, I just wanted to get your take on this because like, right now, Phil, and I interview a lot of people, we talk about attribution all the time, we have some folks on the podcast, who are very, like extremely data driven, are able to, you know, want to find all those data points and then build their marketing campaigns around that. Then we also have the other school of thought, which comes in, which is, you know, you have to have data you have to attribution. But there's some of marketing, that's arts and crafts, that's magic that's taking risks and taking chances. So curious, what do you think about this? And how does BI tools like narrow bi play a role in this in this whole attribution discussion?
Speaker 1 37:33
Yeah, well, as some market sound practitioner, I understand this problem very well. And it's a pain point for me as well. Internally, we solve this problem with a lot of manual actions, like pin points on every new signup and trying to understand what channel it is coming from. And we also match it with our internal customer systems like intercom and our own homebrew Customer Data database. And if someone came under this problem, it could be a huge, huge business, but it's not not me. Good luck to that person. And I'm not going to address that just because mainly because it's her it's risky, because there are a lot of trigger lights from going around attribution around all this compliance and privacy policies. And as a startup we already deal with with enough risks so I'm not going to take one more at surface level we address this problem was narrative deaths by by provides an impact event so we can show what specific campaign or what action led to some increase in your your signups, for example, but not, not on individual signup level. It's, we do not try to solve all markets and problems, we try to solve problems we can help we, but attribution is not known as a problem we can help. It's unfortunately,
Philippe Gamache 39:03
yeah. Now, fair enough. It's not a problem that I would focus on if I was building software, either. It's it's almost like a political debate when it comes to to attribution like there's, there's a bunch of people in the middle, but like people that are very hard on this idea of, you know, I don't need to track everything. I don't need to like spend all this effort and all this money building this multi touch attribution model to assign a revenue number to like these hundreds of touch points that people might have gone through before they start a free trial. I don't I No need to go down that path. But there are companies on the other side of the coin where some C level folks are asking for that like you want a bigger budget marketing team. Sure, no problem but like show me last year all that money on paid ads, all that money on content on my podcast advertising, like what was the ROI on that? And so they're kind of forced to build these models. Whether they trust them or not, to just like, have a way to go back to the exec team and say like, Hey, like, this is imperfect data, it's never gonna be perfect, because like, we can't track every single touch point. But it's better than not doing anything, because we have a rough idea of what's driving revenue or not. But yeah, I'm curious, like, as the company grows, as you guys like, add, add more funding, there's more campaigns, you're doing more things in different channels? Like, is there going to be a time where you're gonna need to prove ROI of stuff? Or are you always going to be in the camp of just like, you know, what, if I see revenue going up on one chart, we're just gonna keep doing what we're doing, because like, things are working well, and I don't need to spend all this time and money on multi touch attribution
Speaker 1 40:44
is a problem, what was such abuse, is that every company has a unique customer journey, you know, clean channels, and even converse, converse for miscounted. So differently now for not just for every organization, but within organization, it can can be counted in different ways, some Google Analytics, or Facebook ads, for example. So it's really hard problem. And we we deal with it by by trouble in which specific campaigns or which specific events led to increase or decrease in some traffic or conversions. But as I said, not an individual level, we were not sure, yeah. Gotcha.
Philippe Gamache 41:27
Yeah, no, makes sense. I agree with your take, we got two more questions for you, I want to give you a chance to walk us through like some of the best features and insights out of narrative bi, with a specific use case. So I'll bring us back again, to our time at Aqua folio, but also like some of my more recent experience, as well. And I think that maybe some of our listeners are going to feel this experience in their soul, and I'm walking them through it. But you're you're coming in, you're starting your day, and you've already got a Slack message from your founder, the CMO, and they're just like, Phil, I was looking at this dashboard, and our free trials are down, or our traffic is down, or MQ ELLs are down. Why are we down? Why is this metric down? And so I've got a lot of ways to tell me that this number is down, I need to block off the first half of my day to dig into that data and come up with some type of answer. What's broken? Why is something down? Is it just seasonality? How is narrative bi, helping me come up with like answers on why a certain metric is down?
Speaker 1 42:36
Yeah, as a surface level, you can just send a narrative to your resume. And in an Earth, we don't just show the specific metric comes down or up. But also we drill it down, for example, we talk about conversions, we show what what channels drive conversions and what channels drive most conversion. If for this answer is not not not enough. We can you can just press on. Yep, LLM recommendations, marking. And often our AI will try to find what was the underlying reason behind this specific bike and will provide some some recommendation. Also, we have a feature called GPT Insights, where we actually utilize open API. To summarize, to summarize some metrics and wisdom are lost or GPT. Ties. We also tried to explain why potential anomaly or potential correlation happened. So I think, in most cases, it should be enough to all transfers, question one.
Philippe Gamache 43:38
Very cool. Well, we'll grab some some screenshots. And we'll add this to the the repurposed blog post version of this because I think this is like you're speaking to a pain that a lot of people have like having a dashboard and a good handle on your metrics is one thing, but figuring out how to tell your CEO why a certain number is down is a whole other nut to crack. I know, I know, TT feels that as well.
Speaker 1 44:04
So ideally, you just connect narrative at your Slack channel and send sincere insight directly to Slack. So the CEO will not bother you. Well,
Jon Taylor 44:15
one, one thing that I just find refreshing about the approach that you have, is it it comes back down to this knowledge worker component, as I think that seeing dashboards and seeing data, it becomes really easy to become kind of tuned out of all those metrics. And I think the key thing, and this is something that I've seen over and over my career is like the people who are really good with data aren't actually like, great at data per se, like I'm sure they have great skills of building dashboards and stuff like that. But that's secondary to actually pulling out things that you don't expect. And I think that's kind of what the promise of your platform provides. Is this idea that like, everything's humming along, I don't need to check my dashboard. Like if you're driving in your car, how often do you check your fuel Will you don't check your fuel a whole lot. But when you see that little ding, ding ding that you need to pull over and add some fuel or charge your car, you're going to act on that information. So I think this is some of the promise that you're bringing to the marketplace. So I think is a really, really unique proposition that you have.
Speaker 1 45:17
But yeah, our goal is to help marketers growth teams to be reactive, not proactive for where you need to spend hours on Google Analytics, or Google Search Console dashboard to find something unusual. If you're doing it, you're probably doing something wrong, it can, it can be easily automated. So instead of being proactive, you can act when one's really necessary when something happens. And a lot of interesting stuff Stuff happens when you have some unusual or unusual patterns. So that problem not reflected in your everyday dashboard, because your everyday dashboard, the static on the timely focuses since some time key metrics you track but sometimes, I could be Fritzie interests and especially like during some turbulent times, like COVID, which were brought a lot of unexpected market dynamics, a lot of a lot of fluctuations in supply chain, and brought, I believe, brought a lot of phenomenalism into markets and data.
Jon Taylor 46:20
This has been a fascinating discussion. I could talk business intelligence dashboards and analytics all day I did for most of my career, this has been an awesome, awesome discussion. I hope folks check out Nehru ba bi and then take a free trial. But one question we ask all our guests and I wanna make sure we ask it of you. You're a CEO, a co founder, a startup advisor, a prolific investor, former amateur soccer player, how do you remain happy and successful in your career? How do you find balance between all the things you're working on whilst only staying happy?
Michael Rumiantsau 46:51
A hard question, which implies that I'm I'm pedal? Well, the I think the the problem with startups is that many founders, as I said, paid their whole personality with their company, which, which makes people miserable. Because startups are mostly a binary outcome, you either succeed, or you or you either fall and not, not my main outcomes somewhere in the middle, like in traditional business where you can take some profits, you can get dividends, and especially if you're a venture backed startup, like it's a very predatory environment. And VC says, say, here in Silicon Valley, like Go big or go home, and pianist, I think the solution for this problem is to try to distance distance yourself from from your job, try to make some boundaries. I personally don't work on weekends, I like I reserved this time for hiking for some other activities. And also understand that it's, it's about it's about journey, not about the destination. Because if you succeed on the on the positive outcome, which is very unlikely in the startup world, because the chance of for IPO is like three to 5% might be so very slim, slim chance and chance for good acquisitions, also not not so high. So it's a crazy environment, whereas the positive outcome was very unlikely. But you still fixated fixate yourself on this outcome which makes people people unhappy and sounds sort of this is to focus on the Jordan find find purpose in what you do and not what your what can you achieve for how this company can be a multi billion company. I personally, I have a bigger reason I think how to make nursing dilate, successful company but I'm more focused on everyday stuff like helping a customer I personally do a lot of support, or delivering new feature I'm excited about or building a new Pro product brand new integration. This is what makes me excited. This is what makes my head
Philippe Gamache 49:02
beloved, such a great answer. I think it can be summed up as like startups 3% chance of success 100% chance of having a great time and going along for a pretty wild read. Yeah, I agree. Thank you so much for your time, Michael. This has been a super fun conversation. We'll put links to narrative bi and really excited to see the the future of the startup and wish you all the best. Thank you so much.
Speaker 1 49:27
Thank you so much.
Philippe Gamache 49:36
Folks, thank you so much for listening this far. We really appreciate you being here. Just wanted to call out two things before we go. Number one, the best way to support the show is by signing up for our newsletter on humans of martech.com. We send you a quick email every Tuesday morning letting you know what episode just dropped. We include our favorite takeaways. So if you don't have time to listen to that one, no pressure we have you covered with some learnings anyway, and number two proceeds from sponsors this year to have allowed us to venture into video. We recently launched a YouTube channel where we publish full length episodes. So if you want to see our radio faces, check that out. That's it for now. Really appreciate you listening again. Thank you so much
Transcribed by https://otter.ai