No matter the industry or the size of a company, the customer experience is often one of the biggest factors
in long-term success. With that in mind, Keatext has set out to provide next-level customer analysis
feedback. Keatext technology takes emails, surveys, chats, and all other text feedback to help companies
understand the customer journey of their clients, including what can be done better.
Keatext founder and CEO Narjes Boufaden recently joined startup coach Roland Siebelink on the newest episode
of the Midstage Startup Momentum Podcast. Their conversation led to a lot of insight about the importance of
the customer journey, as well as Keatext’s startup journey so far and ambition for the future.
- Why Keatext has a product that can be used for both SMBs and enterprise clients.
- How Keatext has eliminated many of the typical challenges of selling to multiple industries.
- What it was like going from working on projects to building a product.
- What Keatext has done to seamlessly connect itself to certain ecosystems.
- What it looks like for Keatext while creating a diverse team where employees fit its culture.
Roland Siebelink:Hello and welcome to the Midstage Startup Momentum
Podcast. My name is Roland Siebelink and I'm the founder and CEO of Midstage
Institute. We help fast-growing startups do even better and disrupt all their
competition with ease. Now, part of what we do is to talk to a lot of the
mid-stage founders out there, and we're very honored to have with us today
Narjes Boufaden, who is the CEO and founder of Keatext. Hello, Narjes. How are
Narjes Boufaden:Hello, I'm good. Thanks. How about you?
Roland Siebelink:I'm very good. Thank you so much. Narjes, you're dialing
in from Montreal today. Is your whole team in Montreal?
Narjes Boufaden:Yes, we are about 25 people in Montreal.
Roland Siebelink:Let's talk about Keatext. What do you folks do? What
difference are you bringing into the world?
Narjes Boufaden:Keatext is a platform that does the analysis of customer
feedback and text feedback. It analyzes all different types of feedback, whether
it's surveys, emails, reviews, or chats. And the intent is really to help brands
have an understanding of what's going on in their customer journey, which means
that we can provide them with insights on what customers like, dislike, what the
question is they are asking you regarding the product or service. Afterward, the
AI that we've developed allows them to get a certain number of recommendations
about the places or the areas in the business that would require improvement and
the impact of that improvement. It quantifies the ROI, the changes that are
recommended by Keatext.
Roland Siebelink:Oh, that's amazing. That sounds really attractive. Can you
lead us through a concrete example? You don't have to mention any names, of
course, but what kind of brand you are working with and what kind of insights
the software generated for them?
Narjes Boufaden:Yeah, sure. We do work with SMBs and big enterprises. The
range is a lot. We cover a bit more than 10 different industries. Just to give
you an idea, we work with American Express and we also work with Lenovo and then
with Orange telecommunications. These are three completely different industries,
and that's one of the things that probably differentiates Keatext is the ability
to address multiple industries.
Roland Siebelink:Just to double click on that a little bit, how hard is it
to serve several industries with one product?
Narjes Boufaden:That's a good question. Traditionally, it used to be
difficult, specifically when you have a text analytics technology because you
need to understand the vocabulary, you need to be aware of industry knowledge
like the concepts and the main ideas and things that are relevant to that
industry. Usually, people would address that challenge by either creating
ontologies or knowledge bases for that specific industry. And more advanced
products would train models per industry. You still have to acquire a sufficient
amount of data to annotate and train into your models.
Being in that space for a long time, we figured out that if we can find a way to
acquire that domain knowledge automatically, to make it easy and instant when
people are uploading the data, that would remove the burden of having to spend
quite a lot of time to train or create a knowledge base. And that's what we did.
The platform is capable of coping with different industries instantly because
among the things that we're developing in our AI stack is the ability for the
assistant to acquire domain knowledge without the need for any human
Roland Siebelink:Oh, that's amazing. That's really solving a key problem in
the industry, it sounds like a key bottleneck that your competitors were
struggling with before. Is this your special sauce, how did you acquire that
Narjes Boufaden:It is. That's definitely our innovation. When a company
comes in, they do not care about whether your technology is capable of
addressing another industry except their own. They don't really care about that.
But the fact is that because we were well aware of the issues related to the
type of technology - the text analytics technology - and the fact that we've
developed that type of innovative algorithm, the side effect of that is it
allowed businesses to be able to discover new things happening instantly.
And just to give you an example. In the pandemic, a lot of things changed.
People started to buy online. They stopped in-store shopping. They started
spending a lot of time looking and being educated on a product, looking at
reviews and then asking questions. The consequence of that is that a lot of data
was coming from digital channels. It's funny enough because that wealth of
information actually delivers a lot of insights on how customers or how the
consumer is changing his way of deciding or choosing a product. What are the
motivations? What are the important things or stuff that you would pay attention
to before making any buying decision?
And that wasn't included. CX analysts, they're not aware of that, because they
are discovering this - because the pandemic didn't happen before, it happened
now. Businesses need to adjust now. They cannot anticipate, they cannot wait
until they have data.
Roland Siebelink:Is this related to what they say in business that you only
have data about the stuff that's historic and that you've understood for a long
Narjes Boufaden:Precisely. Because we had the technology that does not rely
on historical data - it's really the way from the data that you upload that the
domain ownership is acquired - that meant also that new stuff happening that you
weren't aware of, that wasn't part of your customer journey - because you've
changed your customer journey, it's another way of purchasing stuff - then the
system was able to provide that insight. And that's quite interesting and
relevant. Things are changing and fast, and businesses need to adjust, so that
ended up being a very strong, competitive advantage.
Roland Siebelink:That's amazing. Very good. It sounds like you had a lot of
background in this field already to be able to identify such a competitive
advantage. Can you talk us a little bit through the history of Keatext? What
brought you folks to the idea? What made you decide to found a company? And
what's been some of the history since you founded it?
Narjes Boufaden:Yeah, sure. My background is technical. I did a Master's
and PhD in machine learning, applied to natural language understanding. I was
very, very curious about creating machines that will be able to understand human
And after doing the PhD, I had the chance to work a lot as a researcher at
different universities. And then I ended up working at the Research Institute of
Computer Science in Montreal. And it was actually a very good place to start
facing businesses, companies with real issues relating to - how can we structure
information? How can we extract insights? How can we pull relevant information
that would help our processes? Having other algorithms, other software, making
decisions based on that insight. After a while, in 2010, I decided that it was a
good time for me to move away from continuing my academic path toward
Keatext, as it is today, was launched in 2015 and it was actually the result of
a very interesting project I had with one of my former customers at that time
who was interested to get some insights from call transcripts coming from these
customers. And that person actually tests a lot of products - sentiment analysis
products - and then came to the conclusion that that wasn't enough. It wasn't
enough to know if the customers were happy or not. They wanted to have a deeper
understanding of what is - the root causes, if you will - what is making a
customer happy, and even more than that, what customers want to see? What are
the questions? What are the suggestions?
It was really an opportunity for me to develop a prototype. And it was such a
success that we ended up - the market was really great also - we ended up
switching, pivoting the company and moving away from custom development to
create the product Keatext today that you can see. And that product evolved. We
started having more and more customers and having much more data to work with.
We ended up providing this software today that addresses a broad spectrum of
Roland Siebelink:As you moved from a project into a product - that's a very
good journey that we often see - were you able to hone in on what the real
problem is that people are facing and what your software is trying to solve for.
Narjes Boufaden:Yeah. That's a good question. We cannot say necessarily -
there is a high level problem, but then there is also how people are ready to
address it. In what way were they ready to address it? That brings some nuances
into what type of problems we are solving.
For businesses, when we think about productivity, it's not about solving all the
problems that customers are mentioning but starting to prioritize problems. See
what are the things that should be fixed because that impacts most of your
customer experience. That's one end of it. And the second part, which is why I'm
talking about how customers were ready - how they see the solution and how they
were interested to address it - is that the customer journey is definitely the
single source of truth to be able to improve your business. That traditionally
has been addressed through surveys. Once a month, once in six months, once a
quarter, we'll have a company running surveys with an NPS score. You either like
it or you don't like it. Would you recommend it or not?
This is only the tip of the iceberg. If you have a bad score, it's already late.
The way we address this problem is actually by providing a representation of the
conversation from customers by explaining what matters for customers. Of course,
what they like, what they dislike. But more importantly, what they would like to
see, the questions. And try to map all of that to how they perceive your brand
today with that score, to be able to bring recommendations at the end of the
And that's the way for you to set the path for future products, future
improvements, future things to come.
Roland Siebelink:You're saying it's not just about things you need to fix
but also understanding where there are new growth opportunities, potential new
products you might launch, new services.
Narjes Boufaden:Exactly. And for sure the pandemic - one of the side
effects is that a lot of conversations are coming now through chats. Think of
that conversation, before it was mostly taking place in-store and it wasn't
captured anywhere. It was a chat between the customer and the person on the
floor, and that's it and then it's lost. It's such valuable information that is
lost. Today, it's there, it's in chat log, and you can start digging into the
data to be able to understand what needs to be developed, proposed, provided to
your customer and future customers.
Roland Siebelink:That sounds amazing. Let's move a little bit from the
product to the go-to market. You already mentioned a little bit about customer
readiness, but I'm curious to find out what you have learned about which
customers to target. What kind of buyers are your ideal buyers that you'd love
to be in touch with? And maybe even something about who are the people who stay
with you versus those who maybe try it for a little while and then move on.
Narjes Boufaden:It's interesting because we were a little bit following the
marketing in the sense that we had to pay attention to the companies that had
achieved a certain level of maturity in terms of CX practices. And the spectrum
is really large. There's a whole bunch of companies that really thrive to
develop that process and that activity around being customer centric.
We were really focused at the beginning on big enterprises because more
customers, a larger customer base means that you cannot sustain or it's harder
to sustain a CX practice that would be a little bit more organized. I'm thinking
about having a Salesforce, a Zendesk, or whatever platform that would organize
it, all the information that you're gathering for new customers. And since
Keatext requires to have that data collected somewhere, naturally we've been
more keen to work with big enterprises.
Now we're seeing much more mid-enterprises coming into our segment too. And I
think the pandemic has played a strong role here because a lot of businesses
needed to adjust and start opening the door to online things, acquiring
infrastructure to support that have your contact center. Truly, we were really
opportunistic about the businesses and the need of businesses, considering the
maturity in CX processes. That was the main driver here.
In terms of industries, as I've mentioned, there wasn't necessarily a specific
path or a specific preference for an industry. But what we've seen is that the
market or the segment of the market that tends to have a lot of competition is
more keen to pay attention to the customer experience. Everyone pays attention,
but it would cost much more for some businesses or industries than others
because of the high competitive level of that industry. That was another one of
the other dimensions that we used to work with when it comes to thinking about
the go-to-market strategy.
Roland Siebelink:I was just wondering, Narjes, when you are so dependent on
streams of data being available already, people having their customer cases
logged, chat services, to what degree have you been depending on partnerships as
your go-to market? Maybe working with the people that install those systems and
positioning as an add on, for example?
Narjes Boufaden:That's a great question. We were very well aware of the
position of Keatext as an add on. And one of the early strategies was actually
to partner with server platforms, and we do have three partnerships going on.
And then in a more proactive way, we started to position Keatext in different
market places, so the Surveymonkey marketplace and Zendesk marketplace. Wherever
we could position Keatext as a solution that would connect easily, seamlessly,
that was our way to actually position Keatext in that ecosystem.
Roland Siebelink:To be discovered more easily by their existing customers,
Narjes Boufaden:Yes, absolutely. It's a way also to start validating some
of the things that we need to validate in terms of the volume of customers or
feedback, if there is any infrastructure that is already there to collect the
data. That was a natural fit for us to position Keatext in that ecosystem.
Roland Siebelink:Very good. Can you talk a little bit about the team? You
said you're about 25 people, all based in the Montreal area. I know listeners to
this podcast are always interested when you get to that stage, how have you
divided it up between the more technical folk and the more go-to-market folks?
Narjes Boufaden:We have a sales team, marketing team, and dev team. In the
dev team, we do have folks that are solely focused on AI development. This is
the core of our technology, so we need to make sure that all the data is
processed in a way that you can increase the type of value that you can leverage
from that. And there is a lot of potential there, so it's just about how
imaginative you can be in terms of creating value from data.
And then there is the customer success team that is making sure that our
customers are engaged and that we also are listening to what they need to make
sure that we can set the right path of the product in terms of future versions.
Roland Siebelink:Okay. Very good. And how would you say is the ratio at the
moment between the technical departments versus the go-to-market departments,
including customer success?
Narjes Boufaden:I guess, 60-40.
Roland Siebelink:Yeah, that's pretty common. What is, in your mind, the
perfect Keatext employee?
Narjes Boufaden:That's a great question. For sure, culture is the number
one factor. That comes way before any competencies, so there is a fit.
Roland Siebelink:Can you expand a little bit on what does fit mean for you
folks? What are some key characteristics that make someone fit or not fit in
Narjes Boufaden:Absolutely. One of the things that we really value a lot is
the ability to communicate, to be able to share thoughts, and be open-minded
about other thoughts. It's a community at the end of the day. And we want to
make sure that - the best thing we can get from people is when those folks feel
safe, feel listened to, feel respected, and appreciated. And that's a collective
effort starting, of course, from the management, but going toward also the
members team. That's definitely one of the important things that we share.
That's something that we ask right from the beginning. And you can see that when
you are conducting your own interview and a discussion with a potential future
For sure, as a woman, it's really important for me to try to get as much
feminine representation in the team. We are technical folks. It's not
necessarily easy, but so far we've been lucky. We have an amazing reputation
here in Montreal when it comes to supporting women, making sure that they feel
comfortable working with the whole team.
Roland Siebelink:Love to hear that. That's amazing that you can actually
push for more equal representation and diversity in your team as well. That's
How big can you see Keatext become? Let's say, 10 years down the line, where
would you like to be? How will you measure your success?
Narjes Boufaden:I would love to see Keatext in a position where it'd become
a must-have for any business. Despite the fact that your customers, the
happiness of your customers, and your reputation as a business is definitely the
critical component in the success of any company, I'm not sure we are yet in a
level of awareness that businesses would put that as a must-have solution. I'm
not criticizing. I'm mainly saying that maybe there is a certain level of
maturity from the technology perspective and value proposition. Brands can see
the ROI as a straight forward return on investment when they invest in those
types of technologies. That's definitely something that has moved us toward
recommendations to be much more proactive versus showing some insights and go
figure out how to make sense of this and how to use it to improve your business.
I think a part of it is this technology. Is the technology valuable enough to
become a must-have? I think we are definitely in a good position, in a good way
to be that type of technology. And once that is, I'm pretty sure that businesses
would tend to consider that type of software - and Keatext, hopefully - as
important as an online payment solution.
Roland Siebelink:Excellent. Very good. I fully expect Keatext to own 500 of
the fortune 500 companies in 10 years time because this sounds like an amazing
proposition. If people want to find out more about Keatext, Narjes, where should
they go and what should they download?
Narjes Boufaden:Keatext.AI. That's the place to go. Actually, there is a
free trial so that you can just upload your data and you are up and running. You
can start playing with the application and see all the values that we can bring.
Of course, there are interesting case studies from different customers with
different use cases. I hope that can shed some light on how they can get some
value and the type of return on investment. Of course, you can follow us on
LinkedIn, Facebook, and Twitter.
Roland Siebelink:Perfect. Okay. Thank you so much for all this amazing
insight, Narjes Boufaden, the CEO and founder of Keatext. Thank you for your
Narjes Boufaden:Thank you so much.
Roland Siebelink:And to our listeners, we will have a new episode next
week. Stay posted. Thank you.
Roland Siebelink talks all things tech startup and bring you interviews with tech cofounders across the