Dude, where is my order? Behind the scenes of the e-commerce or the story of one DEV and QA

This talk explores the evolution of e-commerce from its inception in 1994, examining various business models such as customer-to-business and business-to-consumer. The speaker discusses the technological challenges faced during the early days, including the lack of standards and manual testing processes. As mobile technology emerged, new issues arose from the need to adapt content for multiple devices and the limitations of automated testing frameworks. The presentation highlights the shift toward headless commerce and the significance of integrating third-party services. Finally, it delves into future trends like personalization and the impact of the pandemic on e-commerce operations, predicting a move towards seamless and automated purchasing experiences.

Uh we have years of hands-on experience debugging, validating and fixing lots of releases and today we are here to share it all with you. >> Hi again for me or if I was my colleague would say hello world. Now we know what it's ecommerce but let's see how many types we have. First we will start with customer to business. When individuals selling good and services to companies

for example we have influencers. Uh the second one that we have is business to business. When one company selling to another companies for example in Bulgaria we have metro. Third one customer to customer. When individuals selling good and services to another people for example ox. And the last one, business to customers. This should be our topic for today. When online stores selling directly to the end users.

Let's start our journey. The first period that we define we named the dark ages of the ecommerce. The moment that it all begins was in 1994 when one sting city was sold. Yeah. one small city for delivery but one giant leap for the ecommerce. Why? Because this was a time when technologies started growing and opening new possibilities. In terms of the ecommerce, yeah, the scar was the

the scale was small back then, but the businesses like Amazon and eBay emerged. In terms of tech, our orders were not in a good place. There were no standards. There was no W3C. CSS was premature. JavaScript was just Mocha and also the HTML tables were the go-to option for styling websites and layouting. There were only two browsers and they fought a war for dominance. Netscape and Internet

Explorer both implementing standards differently making lives of developers in QA a real nightmare. Also, SSL was introduced, but it was really slow to take place on the scene and people were still emailing photos of their checks instead of making real online payments. Okay, let's say some words for the testing. For this period, entirely testing was only manual. No automation frameworks development following waterfall models that was expected.

Also, if we have any QRS, they were really rare. What to say about the tools? We don't have proper tools for debugging, for testing and etc. Please imagine how we measuring pixels to different devices. And when I say different devices, most of the time we have just one top or one desktop machine that we can use to test our products. Unfortunately, with the new millennium, new problems.

All this shows how big was the system this time. And the most important that you will not see on the slide but me and Biswaf was just a children that have no idea what it's e-commerce and web. If we have to move to the next era, it's about mobile revolution. Phones and tablets were and know the computing power that they have were now in our pockets. So

suddenly there was an urge to target those users. Still around 90% of the transactions were made through desktop but the shift has begun and we had to adapt. This is the era where we handled click and touch. On the tech side we needed to serve different content for mobile, tablet and desktop. This all had to be premputed based on devices, user agents and browsers. And believe me,

in the start there were millions and they exponentially growth. So it was really hard for engineering teams to catch up. Features were fragmented. You had 20% of rakia on mobile but then you have 15% only on shops casawat on desktop. So content and marketing had to adapt also and this all impacted no functional metrics. For example, caching was a real big problem because server servers needed to

know what kind of device to serve. Also, there was an SEO problem because you had to maintain two domains, one for mobile and one for desktop causing content duplication issues and of course the surface for attackers was doubled and this impacted security. What was interesting for this period? We have one of the same functionalities but we have to execute different actions to see customer journeys for QA

team this means we have to cover all devices all browsers to see the final result and this means we have fragmented user experience. Yeah most of you will say man we have automation it's easy no worries it's not needed to do everything manually. Yes, but most of the frameworks were focused on the desktop. This means everything that is on mobile on tablet should be covered manually. Uh

also we have to say do we have proper tools for debugging or to use our code to work more properly. Unfortunately we have device specific issues that was for this period and please imagine if I raise one iOS issue how boy will debug on his Windows machine. And as we saw that adaptive actually brought more problems to take business in configurations a new nature of server server

was uh the responsive design. This was the time when we shift to mobile really happened and let see what we have here. We'll start with the problems. For example, we have mobile performance issues. Why? Please imagine you pay $20 to receive 50 megabytes to your from your mobile provider. Of course, when I open just one, two, three sites and I reach my limit. Why? Because the back

ends not only fast needed to front end to mobile but also for desktop. This was absolutely necessary and it was uh in the beginning of the period also. Yeah, here we have automation we have to use not only for desktop also for mobile. Unfortunately, we have to add additional logic. We have a problem like how I I will automate pinch swipe, how I automate Apple Pay for

example was really difficult. Again we continue with the problems that we have with different devices. The that I remember was iPad and only this device has specific size on the screen and the last one was the content. Why in this period we expect everything to be one of the same to different devices to resolve the problem from previous period. Unfortunately, this wasn't happened. Why? Because the mindset

was different. The business thinking was previous period. For example, I want when I open my mobile phone, phone and one category, I just want to scroll to see the products. Easy. But for desktop, please guys add pagination. I want to click the number and after that to see the product. This was the problem for in the beginning of this period. And that mindset was normal because at

the start of the period 90% of the of the transactions were still going through your desktop machines. This changed at the end of the period but there was no mobile first design in the beginning and the lack of e-commerce front and the libraries was severe because every company started with a stack of jQuery and bootstrap and there was no nothing like carousels that were implemented to go

on the fly. I just don't even want to remember how many versions of and variations of carousels I did. And the game was redesigning every three months because designers went wild and they start popping up content left, right, sideways. And as you can see on the picture, this is the holy grail design. It was actually an elusive thing in e-commerce. All those describe the responsive design pitfalls,

but there's two more. We said about the performance issues, but there is also an SEO problem because even if the content was hidden, it was still indexed or not being indexed or nobody knows what happened with it. And the other thing is even if you hide something, it's still accessible. So in terms of security features that were hidden were not really hidden because they were just not

displayed on the storefront leaving that to be exposed by any attackers or even worse you can show features that were for mobile on your desktop so people can exploit that. Okay, let's see the next period. As the responsive design handle the UX now and the customer expectations from seamless website, mobile apps, social platforms was finally met, a new challenge arised. E-commerce was not not just a website

anymore. It becames ecosystem of back-end services where we have payment, delivery, tracking, also invoicing. This was really important. This was the world of customer 360 view where we moved beyond the simple online process to integrated customer journey via different channels. If we need to look into the integration period from tech perspective, we should most probably separate it into different streams. One is actually moving the existing business

systems to e-commerce like accounting and financing and the other side was a sea of third parties that all promised great features. Of course, they came with one case fits them all or they are just plug and play and everything works. As from based on our experience, this might work for small to mid businesses. It's never the case for large corporations. This uh actually brought a need for

a new orchestrator as things were happening previously on the back end. Now headless commerce arise and things started happening on the front end because server computations were not that powerful systems such as CRMs product inventory management or data asset management all had their market specific shares shares. Let's take an example for a complex conf configuration of a review system. You can have one review system for your

Asia sites and another review system for your Europe sites and also you can have various channels that those reviewers are coming from. So you need to aggregate those and give uh consolidated feedback for for the business and the customers. This showed how vital API scalability is and the uptime of our services because if one system fails then all our system fail and as Boris mentioned the integration

are as bridges. If one fails the system fails but was what was most important first before that we thinking only for our product how we can test how we can maintenance and etc. Now we have to read the documentation for all our third parties all integration but not just to read the documentation we have to know how to working with them how we can maintance how we

can configure and etc. This adding additional complexity additional test cases that have to be right and executed. Of course now we have automation we can use for different browser different devices also we have browser stack source swap platforms that we can execute our test cases everything it's fine but do we have test environment to do it most of the third parties no they have just production or

if they have for example QA environment fine please imagine how I can test my payment provider on the test environment my response is always mocked I see success good and when reach on production and we have a problem. Why? Because we have different request or response and etc. Most of the time the third parties do not inform when they make any changes and that was really really

serious problem. Next important thing is the skills. Before that we have to know how to write test case, how to execute them. Now we have to start thinking more for security, performance, accessibility and etc. This was really serious change for this period >> and we moved to a point where people became obsessed with their phones. They needed to be online all the time. Desktop was now that

60 to 70% of the transactions were going to mobile devices. Fast and reliable ways of tracking and notifications were actually needed. A new breed was born, a hybrid between the website e-commerce and the mobile app world. Tech needed to do a really serious choice. They needed to decide whether they choose a native or hybrid applications. Basically, native application promised all the features of the devices and promised

great user experience. But actually this came at a cost because you needed different skill sets and most probably different code base. Hybrid was the easy way out but it was like trying to fit your website in a mobile app or fitting an elephant into a box and it hurt the speed and the user experience. We need to also mention deployments because for web you can push fix

is straightforward and easy let's say but for mobile apps this needs to go through legal checks app store reviews compliance versioning and then you can push it to your mobile end users. Good. We have new period new technologies unfortunately the same problems. Why? Again we can see the fragmentation. Before that was device fragmentation. Now we have operation system fragmentation. What this means? We have issues for Android

or iOS. Of course we have device specific issues for both of them. Next that was important. We had to learn really quickly. Why? Because we have to know how to test this new architecture. How we can automate this and how we can develop this. I remember the time when I was one week to implement my first automation test where was really difficult. Why? No documentation, no chat

GPT and really simple examples just one page and click one but that was absolutely incorrect. Why? Because I did not know that my app should be in the debug version for example. Of course the most important thing for this period was push notification. really powerful tool, but we have to be really careful when we test it on our test environments because we can see something like K.

And this is the last period that we have the era of personalization shopping when every click, every touch, every purchase tells a story. Customers no longer wants just generic uh content. They wants recommendation, dynamic content and of course seamless digital experience. But before we go into nowadays, we should talk about the corona virus pandemic because this was a gamecher for e-commerce. It showed all the cracks that

we have into building, deploying and feature management on the production instances. It also brought the need of rapid features like unattended delivery where couers had to take photos of the delivered orders to your door because of contactless delivery and uh they needed proof that it happened. All systems were real life stress test. The traffic tripled, the share of e-commerce sales doubled and uh of course this showed

that our contracts with third parties were not sufficient enough. There was a problem with the infrastructure because it needed to scale and also there was a problem with the system communicating between each other. And the final problem that was most the most significant one was the people problem because we needed lots of people both on the ground in warehouses and and in it and as this is

a very complex ecosystem. They need to be trained real fast without having the proper knowledge. So uh it showed another crack of how we have prepared. And then the final one is that businesses weren't that we can really fast deploy features which is a problem till nowadays because they expect the same norm. And moving to what happens today I want to start this slide with a story.

Last winter I was casually browsing and my wife was shoulder sniffing behind me and suddenly in my feet there started to pop up gorgeous women in bathing suits and she just nodded her head and say shame on you. You're a married man and I turned around and I said I don't know who buys bathing suits in the winter. Basically this comes to show that data collection processing

and segmentation is really key. We now expect if we do something our content to be tailored right and immediately. But if we get something wrong, we get the example that I gave you. And tart experience is asian said the norm. But features like live shopping, try it on, where influencers actually do real shopping for you and put items in your bag or shop the look where I

want to try out the new tuxedo of James Bond and see whether it fits me and go to the next wedding with it are something that people expect. Real time a APIs, real time content generation is now the problem that engineering has to face. And our final destination of the order journey is what the future holds. My bold prediction is that we are going from an era

of personalized commerce to no commerce at all. You won't care where your order is. We'll have three types of shopping. One is impulsive shopping where I see that Charlie has a Ferrari. I really like it and I want to buy a Ferrari myself. The other two are actually necessities where we'll have subscription based shopping and once every two or three months you receive your toilet paper, you

won't care about it. And third one is I think that you won't care about stock. There are refrigerators nowadays with cameras that know what you have in them. They can actually suggest recipes. So, I'm pretty sure that soon we'll have to build versions of websites, APIs or sitemaps where your oven or your fridge will buy for you and you won't care about it. The only problem there

is whether you trust your oven or your fridge with your credit card. Okay. From my point of view, I see the things differently. I really like your prediction and you have to explain internet AI to my grandma but I'm not sure you do what I think what we have now the customers are foughted with content from all directions computers phones or all gadgets that we use really

good but what is the problem I think the people change their habits will be the same with the shopping now maybe we buy something that we not need we just I want this we'll buy it I don't care if I will use it what I mean I say habits and what what we can see people move from city to village like to be the same in the

also with the shopping where people will want to buy something with story products experience and etc. will not just to buy something they will prefer small producers go on the village and buy the shopping will be driven by the community when somebody recommend me product I will buy this product but how first we'll go we'll visit the village we'll try the product and after that we'll buy

and maybe next time we'll try something to see if they have social platform or something like that that I can order and this is us we really hope that we've managed to find your order or at least tell you where your order is going to be. Thank you. >> Thank you very much. >> Thank you. So, are there any questions? So based on your predictions um either

the e-commerce will disappear as websites and you know the way that we do it today or it will move to the physical world and be personalized and you know you'll have to just go to a physical place and ensure that what you're buying is what what you want. And how do you fit in all this? Because you are essentially part of the digital world. How do you

adapt? >> You first. >> Uh okay, if I can answer this one, I think that commerce will really change in a way that you won't have to go to a website to buy anything. The products will come to you. So basically, you have predictive searches, gener some feet on your phone or whatever where you can easily buy. You can click on a button and you have it.

So we all my expectation is that we will always have development in QA work in the e-commerce. It will be just a shift of how we build the systems. >> So it's going to be in the context of smart you know appliances but you are concentrated in the like the web world. >> Well it it it's not generally web. We you can go to a store on

a kiosk device and just click on a button. But the trend is that you want it now and you want it simple. So basically it won't be a huge checkout process like we're used to it now. But it will be very easy. So oneclick checkouts uh something from uh you talk to Alexa and say Alexa buy this or or Alexa will know what to buy for you

something like that. it will become simpler and simpler and we won't care about it like doing laundry. So basically people did laundry back in the days uh manual by hand. Now you don't think about doing quandry right you go to your appliance throw in and just take it out that's it >> and how the whip e-commerce will adapt to that >> I think >> all the all

the shops that there are out today how will they >> to be honest I think that this is already dead >> it's like 80% of the shares still people like me because I'm very conservative buy from those websites but all the young people actually use apps. So if you have your phone, you're going to buy this. This is my prediction. >> Yeah, but this is this is

just a cycle now. This is the trendy for me. This is just a cycle. It's as the is the fashion today. Yeah, this is trendy. You use your phone, but maybe after 5 years, how many people want to buy something just to have it? Maybe they will want to try first to to use natural products and etc. There are different customers and different for me stores that

can cover their expectations. >> Thank you for the question. Any more No questions. So, >> okay. >> Hey, thank you so much for the presentation. It was absolutely amazing. You draw interesting future and my question well with all this AI automation all this stuff. what your thoughts about the security and the personalization about all that because all the data is going to be generated by someone right

and um you definitely will provide your personal data to all these >> yeah thank you for the question that's a very good question first let me correct you sorry for that we've never used the word I and it's intentionally because currently we have machine learning we are yet to reach the AI era and even now you're tracked everywhere. Microsoft has cookies on your operating system. Google has

cookies on your phone on on Chrome or whatever. So basically you you have various channels where your data is integrated and I think that machines now know more about us than we know. The problem is that we don't aggregate the data in a proper way and the boom of machine learning will actually give us that and I my expectation is that in the next six to seven

months we really have this tailored personalized experience that was an enigma for the last year because e-commerce was pushing in that way but there wasn't the computing power to actually do that in my opinion. >> Thank you. So if you have more questions, you can find Bobby and Stillan outside or in the networks corner. Thank you.