Hi guys, we meet today for this special session of the Marketo Office Hour by Merlin/Leonard to showcase Sypher, an AI scoring app that will go beyond the limit of the scoring that we have in Marketo. That is great, but that has some limits. We show you that right away. Hi, hi to all. Thank you very much for being here with us, guys. I'm very happy, delighted to welcome Paul, Quentin and Thomas from Sypher.ai this morning. This session of this Marketo Office Hour follows the one from last week on the standard scoring from Marketo. We reviewed all the best practices in Marketo around the scoring, but we know that in Marketo we have a static scoring. So we can do a lot of things in Marketo, but there are some limits that we saw last week and this is why I invited my friend from Sypher today to go beyond those limits and see what the artificial intelligence could help us tackle with those limits. Guys, I'll let you introduce yourself and introduce Sypher and if possible, if you have a demo, that would be great. Yes, of course. Thank you so much, Sylvain, for the opportunity you give us to present our solution here. Quickly, my name is Quentin, I'm the CEO of Sypher. We have built algorithms and SaaS, who are able to identify the leads and actions most likely to convert. So you have the chance today to get the full team. So myself, the CEO and more on the marketing part. You got Paul here, more on the business part and Thomas, the CTO. Maybe a first question, how did you guys had the idea to create Sypher? Of course. So for me, I begin, I basically worked and coached some marketers for four years now with my agency, and during my missions with all those marketers, I was all the time in their tools, in their CRM, in their marketing automation tools, and there was a huge problem of doing things manually with the data. So now I got the power to do it technically and we wanted to focus. So we have a focus on lead scoring and I made a ton of mission of services and advising about lead scoring. And now, technically, it's possible to go more further. So Paul, you can go with our demo. Yeah, let's make a small demo. OK, you guys, do you see my screen? Yeah, we see your screen. Thank you very much. Yeah, perfect. OK, let's start first with, what is lead scoring? Lead scoring is based on three types of criteria. You can, if you want to do even a manual lead scoring, you need to get some demographic data, who is my contact, firmographic data, in which company is he working, and behavioural data, how my contact, this company is interacting with my product or my company. Well, and how do we do a good lead scoring based on AI at Sypher? Well, it's easy, we connect to all your data sources: CRM, like HubSpot, Salesforce, and all the other data sources like Marketo, Brevo, LEMlist, all these kinds of tools, and this is our algorithm. This is Sypher, that will identify the right characteristic for each kind of criteria. So, for example, on this demo account, we have seen that a good contact is someone that you met at the seminar, that is working on the innovation department. So it's someone that you have met at the seminar that is working in the department, the innovation department, and whose job title is product manager. So it's the demographic part. And for the firmographic part, we have the other kind of intent, like you have the company size, you know, if he's working with already a competitor and the kind of industry, of course. These two first tables will help you as a SDR or as a marketer to better generate leads with your tools to get a better accuracy in the list of leads you generate. But what is important is at each stage of the funnel, according, of course, to your specificity, we have identified, well, Sypher has identified the right actions that really make the leads progress through the conversion funnel. So, for example, if you are a lead, a good lead is someone who has opened two emails, visited my website and read an article of my blog. But this kind of behavioural actions can be different. They are different at each stage. What matters at one stage is different from another stage. So, for example, if you are an opportunity, it's good to have been contacted one time to have replied to two emails and to, for example, have seen the pricing pages, the pricing page. On the other side, there are some criteria that Sypher has identified that they have, that he has found, that they have no influence for the for the lead, to help, they are not helping him to move through the next stage of the funnel. So, for example, if you are an opportunity, being contacted is good, but subscribing to the newsletter has no impact on becoming — on going to the next stage, for example. So, for each stage of funnel, you see what is working and what is not working. That's super good intent for marketers or sales. You will see what action you could push for your leads. If I understand correctly, your Sypher tries to find patterns based on the fact that the person progress into the funnel. So, I assume that the success here for the lead is the fact that he switched to MQL. Exactly. And if he doesn't switch to MQL, that's a failure. So, that's how you educate your model, right? There is three types of data. So, demographics, firmographics and behaviourals. But, of course, we are based or first analyse and how we update the model. We focus on your signed contacts. So, the person who already buy your products, the people who are never going to buy your products and the prospect who are putting some, making some action in your strategy. Yeah, maybe, Thomas, you can add some precision. So, yes, they are saying it's machine learning. So, you need to learn from the existing data. So, usually, what we are going to do is first, we're going to collect all of this data from your different softwares. So, we're going to connect to most of the available softwares in the market using the API calls. And then what we usually do then is that we have a short meeting with people from the different from the company where they explain to us how they use their CRM, what matters, what doesn't matter. And what is important for us is to identify some condition where we can define that you have lost the lead. So, it's usually like a 30, 45 interview with the manager of the CRM usually. And what is really important is also that we need to split the pipeline in different steps. So, depending on the clients, those may be different. So, for instance, for you, Sylvain, because you don't have much data, we've mainly focused on the lead category. But for someone that has a decent pipeline with like a few, a few hundred lines or thousands of leads, we're going to be able to split it in the four different stages and at each stage we're going to define, OK, this is why this person is one and this is when this person is lost. OK, so what's great here is that, if I understand correctly, you push all your data from the person, the company and the behaviour to the machine and you can find patterns that you wouldn't even think about. For instance, if I had never the idea to score, let's say, I don't know, the industry, because maybe it will detect that the industry is crucial in the fact that you sign a deal or not, right? Yeah, so that's one thing that we can detect what is working or not, even if you didn't know about it. Another thing that we do is that we do some, what we call feature engineering. So we take some features like, let's say, the company size and then we're going to reprocess this feature. So, for instance, for the company size, we're going to bin it in different company size bins, so like 30, 50, 10, 29, as it is here. But we can do different things. Or, for instance, for the city, rather than using the city name, we may be able to use if it's a big city or a small city or for the country, if it's just a European country or another country or if it's a big or a small country. And we're going to reprocess all of those features to give you even more insight than you will have if you were just to take one of our competitors' solutions where they would just take your data and use it as is. As we're going to reprocess it and we're also going to clean it a little bit or let you know if it's not clean, well, we're going to tell you, OK, this feature is very important, probably you should work a bit on it, clean it a bit more, make sure your sales are fulfilling those categories. So it's a lot of different things we're going to add to your data to make it even more meaningful. And is it a one-shot calculation or do you do that regularly? So there are two different things. So first, why there is the setup phase, as I said, when we arrive, we just talk to when we see how it works and we set up the algorithm at the very beginning. Then what is powerful with our algorithm is that it updates daily. So the algorithm is re-optimised daily. So if your ICP changes a bit or if you have like new kind of data that goes in every day, it's going to look at it and see if it changes. If it changes, it will actually let you know that it changes. So you will have a notification and you will also be able to go here and see the differences. And the other main thing that it does is on top of updating the algorithm daily, it updates the scores of your different leads hourly. So it's going to recollect all of your data hourly and then update the scores. And then if, let's say, one of your lead has done behavioural action, in the meantime, it's going to be updated. So you're going to know that this lead maybe has increased a lot in score because of this section. And you're also going to get a notification from that. OK, so it's almost real time for the score. But what's great is that every day you recalculate the algorithm. So if a new factor arrives, for example, if I do new videos and those videos have a huge impact, I will not wait like with my standard scoring to review that every year and lose one year of business, for instance. Yeah, and it's then very fast as when we installed it on you, Sylvain, you had like people from Singapore that have a very good score because you've signed 100% of people in Singapore. But if tomorrow you have a new lead from Singapore and you lose this lead, then the next day the algorithm will be more like, OK, well, maybe Singapore is not that great. And then their score will decrease. OK. What are the other menus? I'm curious. OK, yeah. Now that we have all the characteristics at each stage of the funnel, well, we can do our job and score the integrality of your lead pipeline. So here we have the vision by lead, but we also work by account if you have no contact inside. So at each stage of the funnel, you will find the score of each lead, which represents the probability for the leads to get through the next stage. You have small indication here showing in a small, in a fast way what is on red and what is on the green part. So here is the demographic part, firmographic and behavioural, of course. As Thomas said, the score are updating every hour, so I can see if it's increasing or decreasing, of course, I have the justification just below where I can see. Well, this, for example, Olivier, he liked a LinkedIn post. He shared one of my post also of the company. So it's increasing the score, of course, but he's a CEO. And we have seen just before in the ICP part that we were more most likely, it's better for us to work with product manager. And he didn't book the demo I sent him. What can I show you is that Sypher and all the tools are, yeah, we, they are communicating together. So of course, if I'm clicking on this link, I will, it will automatically open my Salesforce and I can, of course, have my score. I can see a few justification about this score. If I'm going on this part, I can have the full view of my lead pipeline, of the scores of Sypher. I can see the trend, of course, and all the justification, every— OK, that's great. We can push data back into Salesforce, right? Yeah, everything is communicating with the two products are interacting together. So, yeah, as a marketer on the first step of, as an SDR on the first step of the funnel, I can do my own list. And as a SDR, of course, at the end of the funnel, I know exactly who to call and why, because I have the justification. So I can do my own list on Salesforce, but if I don't want, of course Sypher is creating some AI based list based on the revenue, based on many things that I'm going to show you, of course. So here on the homepage, you have the average amount, which is the amount of each leads multiplied by the probability for you to sign them. Of course, it's not the basically the amount… It's revenue. Yeah. Yeah. It's what we call a weighted deal amount. Yeah, it's a multiplied by the probability. So, of course, I have four type of list that we are pushing now. We have the list reactivation, which which are people that you haven't been contacting for a long time. You have the list pushback. You have contacted them for too many times and their score is not that good. So you need maybe to adapt your marketing strategy. The marketing actions needed are people that really fit with your persona, your ICP. But there is a lack of behavioural data. The score is really low. So you need to push them some marketing actions or sales actions. And the activation list, of course, there are people ready to move to the next stage, to the sales stage. So, of course, when I'm clicking on one of the list, I have the detail. I have the amount, of course. And as I showed you before, it's opening in Salesforce. I can also have the export function, so I can do my own stuff if I want. Yeah, on this part. Well, I think it's it's approximately everything. Yeah, we also have some statistics about your conversion funnel. If you want to get some characteristic about that. So we give you some intent about the conversion time, conversion rate, the average number of actions you take at each stage. Okay, great. Do you have any idea of the customer benefits for your customers? Yeah, we just started the commercialisation in December. So for the moment, we have a two month look, about two months. But yeah, marketers are winning two days per month, approximately, because they are not spending time on useless and manual tasks. And for the moment, the first metric we have is more than 3% in the conversion rate, increasing in the conversion rate with our customers. Our customers are startups, a small medium enterprise and more in the tech. They are digital services and software companies. Yeah, so I assume that for your product to work well and have the maximum impact, there are a number of prerequisites here, for instance, have a good data quality first, because otherwise bad data in, bad data out, I think. Definitely. You have to calculate somehow a funnel. And so, in Marketo, everybody in the audience in Marketo has usually a lead lifecycle programme that makes the lead advance in different statuses and you saw mine And so, we can simplify it a bit so that we fit with lead, MQL, SQL and opportunity. For companies with a lot of data, could we have more steps in Sypher or not? It's limited to four. No, no, so we can have less. If you have less people, usually we recommend at least 100 signed clients. So you have at least 100 successes and you need 100 failures. Usually you will have more failures than successes. So it's usually fine. Then the more you have, the more steps you can have. So currently, we show with four steps because it's usually what companies do. If you have more needed process with more steps, you can have more steps. Also, if you have different products, what we can do is to have a different algorithm for different products because a product might be signed with different criterias. Like, for instance, we had people doing insurances. If you insure your kid, of course, it will be very different than insuring an adult. So you have different properties that matters. Okay. I was thinking about my customer in higher education, like ESSEC or HEC, they have thousands, dozens of thousands of leads because everybody wants to apply to their programme. So that's the kind of company where they have a lot of products, a lot of business lines, a lot of universe. And yeah, they have, I think, enough data to split a bit more the funnel. So, yeah, okay, that's great that you can estimate that. That's one of the differences with competitors. Our competitors usually will just give you one score for the whole pipeline, which is not really meaningful because as Paul said, several times, criterias are really different at different part of the stage. So if you take like Einstein or this kind of solution, they will just give you one score for the prospect when they get in, while us really take you along the different steps and tell you exactly what works for each of those. And you can also like tailor which many algorithms you want because we have other people using HubSpot lead scoring and you have a limited amount of algorithms. I mean, manually defined lead scoring you can set. So with Sypher, you are not limited by that. Alright. So question, especially for Europe, what are the data? Where do you send the data? So we store all of the data on our servers that are private and that don't share data with any third-party programmes or software. You don't send anything to ChatGPT, right? We don't send anything to ChatGPT. So your data is safe. It's not going to leak next time there is a leak in ChatGPT. We're just going to connect well to your different marketing tools via API. And then, the front end will also connect with our server. That's all that we connect to it. So it's safe. It's in the United Kingdom at the moment. We plan in the future to grow our servers when we have more demand and we probably will move them to France. I think you told me that you can make the data anonymous or to use some cryptography. That's the thing. If you really don't want us to have access to any private data, then you just let us know and then we implement the software, the solution anonymously, which means we won't be collecting first name, last name, whatever you want. So you can just tell us, I don't want you to access this or that. So we will just actually store the criteria that are important for our scoring. So for instance, if you have the address of the person, we're going to say, okay, hide it. But if you can at least give us the city or maybe the department or whatever that may be useful. If you really don't want, well, we don't take it. And then we use our algorithm. We score your pipeline. We give you the front end as you have here, but you won't have the first name and whatever, but you will still have the link to your CRM. So actually, if you click on the link with the anonymous version, then you will just end up on your CRM with the name of the person, because this is your data in your CRM. And that makes it fully anonymous. And at the same time, you can easily just go back to the real person without us knowing who's the person. But what's the reception with the salespeople who use this product mainly? Yeah, well, salespeople are more directly using their own CRM. So they appreciate the fact that we are sending back all the data in the CRM. It's more the marketers that are using Sypher directly, our platform, because they can manage the lead pipeline. They can do their own list. They can track what action is working or not. But yeah, definitely the sales team, they just want to call, they just want to sell. So few of them are only using directly the platform. They appreciate the fact that everything is available on their Salesforce. One version of Sypher is totally different from another version of Sypher, because it depends on the data you collect. So one version is totally different from one of our customers, totally different from another. Okay, so yeah, you told me you have this setup phase. So what, that takes three days, 10 days? Approximately, yeah, approximately a big week, let's say 10 days. The first step is getting the access to your data sources. So we collect, as Thomas said, the API keys. And then, we start cleaning the data. This is our job, so you have nothing to do. And a few days after, like three days after, we make a big meeting to understand how do you use your CRM? What do you call an MQL? What do you call an SQL? And with this, we can settle well the algorithm at each stage of the funnel. And then, let's say, like five days after, we deliver the platform and all the data is sent back to the CRM if it asks. And down the road, after the setup phase, so if everything was perfectly that, that's great. But let's imagine, I have something to change, because I don't know, I have new fields I've created, I have new campaigns, new behavioural data I want to be taken into account. Am I self-reliant to do the changes or should I call you? First, we set up the phase when we deliver the product, we, of course, ask you what you think and have feedback about it. And usually we need to tailor it a bit more because you may wonder, okay, why is that data or not that one? So we might adjust it a bit. But usually, directly after delivering it, then I think every two weeks or months, depending, we're going to do a check up with you, see how everything is going, how you're using the platform and adjust depending on this. And if you really want to tailor your score because you really don't want exactly what the algorithm is giving you, we can then apply different weights on different leads just to get the algorithm to give you slightly different results. So all we can do this for you and then we do everything or we can also implement a new column in your Salesforce where you will be able to give some different weights to different leads. And then, we will just, as soon as you change this data, one hour after, it will just update in the software and you will get different results. And the next day, the algorithm will adjust to it. So you can do that or you can just send us an email and say, hi, guys, I've added this and that. Can you check it in Sypher and then make it work? And usually, we just add it so you get it the next day or the day after. Okay, I think that's very impressive what you did, guys. Maybe to finish, how does the pricing work? Yeah, the pricing is based on the data we need to score, of course. And it depends on the number of leads you ask us to score. Just a few examples, for 30k leads is based approximately at 10, at 1k, 1k per month. 1k per month. Okay. And now we are finishing our early adopter list. So we just push some proof of concepts and it's 300 by month for three months. Yeah, if anyone wants to try, want to get for this kind of, yeah, early adopter programme, so it's still available. Okay, so it's a perk for three months for 3k per month or? 300 per month. 300, sorry. 300 per month. Alright, so yeah, easy to start and to test if that makes sense for the company. Yeah. Any last words you want to add, guys? Oh, I think it's just to show a bit more about the ambitions of these projects. So now we are basically able to tell you which leads is going to the next stage. But the goal is to go into the next best action markets and tell you, okay, it's this lead who had to go to the next stage, but there is the logic of action you have to take to push him to the… To the next stage. Yeah, the final stage. So you will tell me for instance, you should call this lead or you should send a LinkedIn message or you should push the infographics xxx to this lead, something like that? Exactly. Exactly, yeah. So you can just take the list and go through them and then say that, okay, I send him an email, the next one, okay, I send him a new book, the next one, okay, I'm calling him or whatever you need to do. So really to just make it as easy as it can be for you to do your job. Alright. So it's your project for 24 or is it more for.. Yeah. 24? Yeah, we are at the part of the go-to market for sure. And it's a huge ambition for Sypher. We see our company even in the years. So yes, it's a more ambitious project for what's following. Okay, guys, we will follow you up and certainly organise another session a bit later in the year to see all the innovation you will bring in 24, okay? Yeah. Thank you, Sylvain. Thank you. Thank you very much. Bye. Bye.