Weglot

The Weglot story: The journey to building a website translation solution

The Weglot story: The journey to building a website translation solution
Rémy Berda
Written by
Rémy Berda
Rémy Berda
Written by
Rémy Berda
Reviewed
Rémy Berda
Reviewed by
Updated on
July 11, 2023

Chapter 1: “The Prelude”

Let’s start the story in July 2015, two months before Weglot was born. I’m alone in my studio apartment and I’m feeling very depressed. After 2 years working on Spothers, Alexis (my business partner) and I just decided to give up on our startup.

Knowing when to stop

Spothers was a website I co-founded with my childhood friend Alexis in 2013. It was a classified ads website with geolocalisation (you could see what people were selling in your street through a Google map interface). It never really took off, but we had still quit our jobs during the summer of 2014 to work full-time on it. At that time, we were very optimistic (and probably, looking at the wrong metrics) and we thought it would grow as soon as we focused 100% on it.

For a year, we worked full-time on it. We participated in the Microsoft Ventures program for 6 months and tried to grow our user base. We tried a lot of different things like participating in second-hand clothing events, web scraping, buying items from people on our own website to kickstart things. During that time, I remember we were very positive and always thought it would go up at some point.

Sometimes we would get an article in the newspaper or be mentioned on TV. The traffic would go up on our website and we would get very excited.

Unfortunately, though, our actions were not paying off and in July 2015 we felt like there was nothing more we could try. When we stopped looking at the traffic or the number of ads on our website but rather at the number of transactions being made on it, we realized it was close to zero. At the same time, Facebook was launching its ads market which would add some competition in an already ultra-competitive market. As the walls started closing in, we realized Spothers would never become what we wanted and that the dream would never come true.

On Thursday 30 July, I remember it was the end of the afternoon and we were sitting at my apartment with Alexis and realizing we had no ideas left.

Previously, we had done many things that didn’t work but that afternoon we just had no more ideas and it really felt like the end of the road. We decided to think about it during the evening and regroup the next day. Alexis went back to his place and I went to visit my family. I remember being in my thoughts all night, tormented, out of options, and with a feeling of having lost a full year for nothing.

During that year, I had seen many great talks and I had read so many articles of great entrepreneurs and somewhere in their success story there was always a moment where they had nothing but they struggled and finally came through it. Maybe this was the struggle in our story, maybe I should persevere and not quit, I thought.

But I really didn’t see any path to success that evening and I went to bed thinking it was time to move on.

The next morning, Alexis came back to my place and also had the same realization. There was nothing left to do and we needed to stop. We spoke for about an hour and he left. Alexis and I stayed on very good terms and we are still convinced that we were a good team but we worked on the wrong project. The funny thing is that Alexis also created a startup after Spothers, called Cyclofix that is still going today!

Make or break

So here I was, sitting at my table alone with nothing to work on. The entire focus of my last year was gone. Spothers was more than a project. It was my first startup so I had a strong emotional connection to it, it was kind of “my baby” and losing it was devastating.

We didn’t sell the website, we just stopped. Back then I thought it was a pure loss of time without any value. Even if I was sure it was the right decision (I knew that any time spent on this project would not make it work, there were stronger forces that we couldn’t fight anyway) it was a depressing feeling. Now I would have to announce it to my family and my friends and it would be hard (turns out it wasn’t).

The good thing is, the end of Spothers was 2 days before my summer vacation. During my 2 weeks of vacation, I tried to relax and think of what I would do next: look for a job or try to start something else. I weighed up the pros and cons during that time.

  1. Look for a job: Safer approach. I was worried I would lose interest in my diploma/studies if I didn’t get a job soon. I didn’t want to waste another year for nothing. I only had 1 year left of unemployment benefits which meant that I would need a paid job within a year (I had 0 savings). Oh and also, I had no idea, and I had no cofounder.
  2. Start something new: I could have my own rhythm. I could work on something I like and do some coding alone (which I enjoyed doing). I was sure I would find something. After all, I still had 1 year of unemployment benefits. I should give it a try for 6 months at least and try to find an idea. Yes, I’m alone but I can find someone later.

Searching for that big idea

I came back from my vacation on 16 August and started looking for ideas. For me, that didn’t mean getting up early, doing some kind of sport, and being super productive like I was used to reading in some blog posts from other entrepreneurs. Since I had no obligation, I would sleep until 12pm, cook some pasta, and surf the internet all day.

As it was the summer and my studio was small I felt oppressed after a few hours indoors so every day I would grab my flip-flops and walk around my block for an hour.

I remember clearly that I would walk very slowly and make imaginary scenarios with imaginary ideas as I stared at the ground. These walks were the most productive moments of the day.

After a few days, I had my first idea of a mobile application to make donations. I was thinking, when we make donations to charities, we always do it on our desktop, not on our mobile. Why and how could I change this?

This time, and compared to Spothers, I didn’t want to lose time. I wanted to have feedback quickly and see that people were interested before I spent time on it. So instead of developing something, I went on a dozen charity websites and sent messages through the contact forms to offer my help as a developer for free to better understand what was lacking in the donation world. Here is the email I sent at the time.

Translation:

Hi David,
I am a young active person and I would like to help. I have no money but I have time and coding skills. I saw that you don’t have a mobile application. I wanted to know if you were interested in me developing a mobile application (to make donations and allow you to push notifications to your members). Obviously, I’m not asking for anything in return.
Just let me know if you’re interested
Best,

I wasn’t sure if it was a good or a bad idea, so I would still go on my walks and think about other ideas. I wanted to test several ideas in parallel. I felt very pressed by time so I couldn’t risk only pursuing one idea. I had to find more than 1 idea but also some kind of validation also.

The Weglot idea, a translation API, appeared on such a walk. On Monday 31 August, I was still trying to think of ideas and because good ideas didn’t come naturally (at least not for me) I would just think of random words and then think if it was a good idea. At that time, API was a big thing and many successful startups were “API of SMS” or “API for emails” etc…

So I was joking around putting random words after “API”, sometimes it meant nothing. Like “API of cars”, “API of supermarket”, “API of pasta”, “API of families”, even “API of API” etc… At some point in that mixture, I threw in an “API of translations” but didn’t really think much of it on that day.

The next day, again on a walk, I tried to think again of ideas, this time trying to focus more on my strengths. I was alone so if I were to succeed in an idea, it was probably in something I liked and had an advantage compared to others, like code.

At that time, the charity project was at a standstill. Few people had answered me and I was thinking I was not the right person to do it. I had no advantage as I was not particularly involved in one. I gave up that idea after 10 days. Then, I tried to think of ideas that were more developer-oriented.

I developed the website for Spothers alone for one year so there were many problems I encountered. But each time there was a good API for it, like Sendgrid, Stripe, Twilio etc… However, when I tried to translate Spothers in several languages 3 months earlier, it wasn’t so easy.

This is where my “API of translation” idea came back again. Whilst out on my usual daily walk, I tried to picture what it meant and how it would work. I could use Google Translate to translate and connect it to people’s websites. Ok, but what would be the value then compared to just using Google Translate?

People need to control translation quality 💡. Ok, then I can save the translation in a database and give people the option to change it. Ok, but then I would save translations from all the websites from potentially many users!

It’s scary. It won’t work, I thought. Well maybe…at least for one website it would be fine so I can try.

Building on the idea

I didn’t need to think yet of the problems I will have if I’m used by many websites. That would be a “good” problem to have by then. Would I add a Javascript in the site to make the API call and try to change the words dynamically? Or maybe I should do it serverside and use a PHP library instead of a JS snippet?

I wasn’t sure of many things: was it possible to connect to Google API, is it free, etc… I quickly stopped my walk and went back to my apartment to search for this. I was very excited to see if it was possible. I searched only for feasibility and after 20 minutes I had the feeling it was possible.

So I started the code right away because I wanted to see this in action as I wasn’t sure it would really work. All I needed to do was a small PHP code that connects to Google Translate and saves the translation and returns it. Then I did a small javascript code that I put on the Spothers website (the site was abandoned but still online, which was good for testing) that would parse the HTML and send words to my API.

I developed this for a week. On September 2nd, I did the PHP code that connected to the Yandex API (like Google Translate but with a free limit, so better for testing). Then I tried to put it on the Spothers website but I needed a public domain name and SSL to make the communication possible.

So on the 3rd of September, I thought of a good name that would still be available for me to buy the .com domain of. I tried “polyglot.com” but it wasn’t available so after a few tests of domain availability, I came up with weglot.com. It was short enough so I just bought the domain right away – I just wanted to test anyway.

GoDaddy email after purchasing weglot.com

And so…Weglot was born!

Chapter 2: The First User

On the same day Weglot was born, I created my AWS account to bring the code online. I needed it online to be able to call it from another website (spothers.com). As I rushed towards an answer of whether it would work or not, I actually never searched on the internet to see if this idea already existed.

It turns out there were already other products that were doing this, but I didn’t know at that time. I found out 2 or 3 weeks later. I’m not sure how I would have reacted if I had found this out before coding.

Getting feedback

I finally decided it was better to translate server-side because translation is always done server-side when done manually. I really had the Algolia PHP library as an example. I really wanted to make something that integrated the same way they did in the PHP code because I thought it was simple. After a few hours, it turned out that it was working. I got the spothers.com website translated automatically through my code. This was the first page translated by Weglot!

After that, I wanted to quickly see if I was again going to lose my time or not (I still had it in the back of my mind that I needed to get paid in 1-year max or I would run out of money).

I wanted to make people use it and see if they needed it. Of course, my code was too “raw” so I coded for one week something that was just about ok for people to see. No design but just a simple interface. There wasn’t even an onboarding. People signed up, then I contacted them manually and gave them a PHP library code (as ZIP) to add in their website, and then they would have translations and could edit them through an ugly form.

I thought that even if the Weglot website was ugly, it would be a good test to try to make people use it. If people were not using it, adding design would not change this. But if they would use it as it is, it meant there was an actual need.

I didn’t dare post in startup Facebook groups because I really wasn’t sure if it was good or not and I didn’t want to waste that “one-shot” opportunity and save it for later. Instead, I tried to contact a few people manually on Facebook.

So, it’s now September 9 and the first time I would reach out to the external world. My goal was to ask a few people if they wanted to translate their website and if yes, tell them I can give them a code that would do it in 5 minutes. I knew I needed to speak to the developer of the site because there was a PHP library to install.

I wasn’t sure it would work and not make the website crash so I tried to contact people that had a website but not even a startup yet. Many people didn’t reply but one did. The conversation we had was totally unexpected and the value he brought to me that day was infinite.

Jérémy Dana was the founder of Early adopters (a platform that connects new products and early adopters). He instantaneously wanted to test the library and was very excited when he saw this working. He suggested that we do a Skype call. What was awesome is that he was telling me exactly what he wanted and I was making the modification to the code on the fly because I didn’t want to “lose” him. And then I was telling him “It’s good, what you asked is online ” and he was still online. We spent 2 hours online.

The conversation continued 5 days later. I was so motivated by the fact I had one beta tester that I had improved the website and dropped the PHP library and replaced it with a JS snippet because I felt it was too hard to install. So I showed him the changes and he was again very excited and kept testing and asking for features. At that point, I was basically developing the service for him alone.

I kept emailing websites through contact forms or email. I wasn’t directly telling people I had something to test, I wanted first to get an answer. Most of the time I didn’t and when I did, many people were not interested.

My email was :

Hi Théo,
I wanted to know if your website “website.com” was only in French or if you had planned to also have it in English, German etc…
Thanks,

And the email I usually got was something like:

“Hi Rémy, sorry your message went to spam …
For now, the website is still in development so I don’t think it’s useful but thanks for your message.”

After a few days of sending messages, I was able to have 3 people install Weglot on their website. The important thing for me was not really if they would keep using it, I just wanted to make sure of 2 things:

  1. Whether it installed quickly, no longer than other SaaS that had libraries/ JS snippets.
  2. Whether it was doing what it was supposed to do: was the website translated instantly when people clicked the flag icons.

From 9 September to 17 September, all I did was contact people and try to get them to install the service. And when I wasn’t sending emails, I was improving the design to make it less scary to people. It was not beautiful at all but just “OK”.

I was able to jump on Skype with another tester on 17 September. He installed the service and was also very enthusiastic. He offered to promote the tool to his business school fellows and asked for a slide. This is what I gave him at the time: the first presentation of Weglot!

The first presentation of Weglot, slide 1/2
The first presentation of Weglot, slide 2/2

At that time I was clearly enthusiastic. Having people already installing the solution and using it boosted me and I was thinking that it could be a good idea to actually keep focusing on it. But not alone. I knew I couldn’t do this alone.

Running a startup is just too hard. I remembered the “down” feeling I had with Spothers when it wasn’t working as expected. Having someone else is not just good for complementarity skills, it’s also someone with whom you can share the burden. Someone you can rely on. Someone who can pick you up when you’ve hit a wall.

I needed a co-founder…

Chapter 3: Meeting Augustin

Back in August, I called Balthazar from “The Family” – an entrepreneur network we were in – to let him know that we were stopping our startup “Spothers”. His reply was “No problem, do you want me to put you in contact with other entrepreneurs looking for new ideas in case you launch something new?”. I said “Sure”.

And on 9 September, Balthazar called me back to ask me if I wanted to meet someone, someone more “business-oriented”, who was looking for a co-founder and ideas. I accepted and 1 week later, I met Augustin at a café.

Putting 2 minds together

We talked about our journey and how we got to where we were right now. He had just come back from New York where he worked in M&A but was missing something. He wanted to start an entrepreneurial adventure.

I had just failed my first startup and was trying to rebound in one year. He also had a deadline and like me, he could only sustain 1 year without an income. Then, we started exchanging all the ideas we had. I talked to him about the donation app and that I wasn’t really sure about it, he talked to me about an idea in accounting (kind of like PayFit) and then I told him about Weglot, and that I already had users and believed we should work on it.

He immediately liked the idea and the fact it was a problem I had encountered myself as a developer. He asked me how he could help and I answered I needed people to install it on their website and give feedback. At this stage, we never talked about creating a company and how we would divide shares. It was pure execution and making sure people were interested in this tool.

So we began to work together the day after we met. Augustin was sending emails and Facebook messages to anyone with a website to make them install the JS library. He was also going to the NUMA, a coworking space for startups in Paris, and directly asking people to install it.

Thanks to Augustin, I was getting a lot of feedback and improving Weglot every day. But at this stage, the most important thing was still to get people to try it, so I was focusing more than 50% of my time on also finding users.

By the end of September, we got consistent feedback from users, meaning it was converging to several key points. They liked how easy it was to install and the “wow effect” when they saw their website translated. However, many people were on WordPress and wanted a WordPress solution (a WordPress plugin). People were also concerned about translation quality so we insisted more on the fact that all translations were editable when we presented the tool.

Discovering WordPress

At this time Augustin and I didn’t know WordPress at all but because people didn’t want to install the Javascript library directly, we decided to create a WordPress plugin so that we could have more people install it.

I created a plugin that was only adding JS code in the HTML pages. There was a small admin page in WordPress where you entered your API key and the languages and that’s it. This version of the plugin had no complexity in it and was created in an afternoon.

After that, we started to get WordPress users organically as we were listed in the official plugin repository. But it was maybe 1 person a day downloading the plugin and usually not keeping it. So our testers/ users were still coming from our manual emailing.

Every time someone installed the plugin we were following his or her every move. Were they using it? And, how? We were sending emails after the user subscribed (manually, of course, nothing was automated), then on day +1, day +3 etc… Basically we were asking if everything was ok, and what feature they would want.

The most useful feedback came from people who stopped using it after a few days. The main reason was SEO. Indeed, as we were translating using Javascript (client-side) our solution would not work for Google. Meaning that the translated versions would not be indexed in Google. And that was a big problem.

Email from a user about our SEO problems

That’s where the dilemma of JS vs PHP came back. Should we keep working on the JS library that was easy to install but had no multilingual SEO (so not a long-term solution) or should we focus on the PHP integration that was too hard for people to integrate? The solution came from WordPress users.

Indeed, as I had just developed the WordPress plugin, I noticed that plugin developers can access server-side code of the website so it was possible to wrap our PHP library inside a plugin and combine the two advantages of our newborn solution:

  • A very easy setup: people will only need to install a plugin
  • SEO compatible: the plugin will use server-side translation and ensure best practices

So by the end of October, I rebuilt the plugin from scratch, without Javascript, to have a full PHP solution.

After a week, we had an easy-to-install and SEO-compatible product for WordPress users only. So then we made an important decision. We were going to focus only on WordPress for a while, as it was where our product was the best at that moment.

We contacted all the WordPress users that had previously wanted a plugin or rejected us because of SEO to let them know we had a new solution. Along with WordPress users that discovered us directly from the WordPress public directory, we had now a constant flow of registering people (5 per day) and a product that was buyable.

We were now waiting for our first customer to finally validate there was a business potential…

Chapter 4: The First Customer

Augustin and I had read on a blog that if you can get 10 unrelated people to buy your product, you might have something valuable. I really think it’s true and that you shouldn’t rely on people saying “you have a great product” or “what you are doing is amazing”. Indeed, they have no interest in saying that it’s bad, so they will always try to please you. That’s human, and that’s perfectly fine.

The best way to know if you have a potential company is to get people that you don’t know (not your friends) to pay for your product. If you get 1 it’s awesome but getting 10 is a more consistent signal that you have something.

Finding paying customers

So Augustin and I embarked on that quest starting in October 2015. At that time we had a few people who tried the product and kept it on their website for several days/ weeks after the initial test. It was obviously useful for them and translating their website every day. But it was also totally free so we were wondering what they would do if it wasn’t free. Would they pay or uninstall it?

So in that same month, we changed our speech for new users and talked about a “trial period”. So it was free for now but you would have to pay or stop using it after 2 weeks. We integrated a payment form on our website and added a pricing page with a unique plan: 15€/ month, all unlimited.

Then we kept talking to users making sure they had no problem with the current solution and waited to see if someone would decide to take a subscription

First pricing page of Weglot

On the 5 November, it happened, our first client. I clearly remember the excitement when I received Stripe email notifying me of the new subscriber:

Email from Stripe after our first payment

We felt like those 15 euros were the beginning of something. It meant that the product had some value. We really thought that we could reach those 10 paying customers. We thought, if we could get 1, we could get 2 by doing the same thing. And then 10 is not that far.

In November and in December, we kept that only goal: get 10 paying customers as fast as possible. We focused on the 2 following tasks:

  • Get a maximum of people to sign up – keep sending emails and contacting people on Facebook and also improve our ranking in the WordPress Directory to get more inbound traffic.
Registered users per day from November 2015 to March 2016:
  • Try to convert registered users into active users that would still use it after the end of the 14 day trial period. We thought that if a user was still using it at the end of the trial period, he would have a higher chance of becoming a customer.

For this, we kept manually emailing the customer separated emails. No automatisation at all. Basically, we would go on the client website and give him personal advice like:

“Hey it seems you registered but didn’t use it on your website traveltoberlin.com, maybe it didn’t work? I can help because it would be nice to have your website in German also! Just tell me, I’m the developer of the plugin so I can help”

“Hey it seems you have integrated Weglot but it’s not very visible, would you like me to make some CSS code so that it fits your website’s design”

By doing this we were able to convert a few more people and on 31 December, we reached 7 paying customers. At that time, we also added annual payment as an option (before it was only monthly). We got our first yearly payment that December, a €150 yearly payment. It boosted our confidence and we kept working hard to get as many people signing up as possible.

We progressively shifted from outbound cold emailing to inbound. Meaning that we stopped emailing people but focused more on improving the number of people who signed up by themselves.

In December and January, we focused on reaching the first page on the WordPress Directory with words such as “translate” or “multilingual”. For this, we improved our listing by doing a demo video, clearly explaining what the plugin does, and providing stellar support to users to try to get positive reviews. The strategy paid off and the number of users signing up increased.

In parallel, we contacted many WordPress blogs in the French community to present what we had built. Many didn’t answer but one did.

Fabrice was the owner of WPFormation, a very influential blog in the WP community. He agreed to jump on a call with us so we were able to show him what we built. He was very knowledgeable of WordPress and on the multilingual problems in WordPress. Many plugins already existed but the most used at that time had many problems. He gave us valuable advice and published the very first article of Weglot in December.

This gave us a huge boost and some legitimacy in the WordPress community. At that point, we decided we would participate as sponsors at the next WordCamp Paris – a physical event in Paris around WordPress in early February 2016.

By the end of January 2016, we had reached 20 paying customers and were convinced that we had the beginning of a business.

Conclusion

This is when we decided with Augustin to create the company. Weglot was officially created on 3rd February 2016.

We were very confident that we had something to work with and that could grow a lot. Of course, the struggle would continue and there were many issues we would have to face progressively. But compared to my previous startup, Spothers, this time we had the “market” with us.

That’s one of the biggest lessons I learned from doing a failed startup and a working one.

Satisfying a need is the most important thing to look for in an idea. If you do, you will be swimming with the current and even if you’re not a good swimmer, you will be faster than an Olympic champion going against the current.

Discover weglot

Ready to display your website in multiple languages?

Try Weglot on your website for free (no credit card required).

Icon blog

In this article, we're going to look into:
No items found.
Try for free