Visitor Monitoring Connecting Your Business to Web Performance: The Philips Journey | Keynote
Webcast

Connecting Your Business to Web Performance: The Philips Journey

About the Webcast

Hear how the performance team at Philips improved the experience of their mobile website customers through:

  • Identifying high-impact metrics with Marketing and Finance stakeholders
  • Analyzing performance and business data together
  • Understanding the dynamics of the mobile environment
  • Benchmarking and applying best practice optimizations
  • Using a business case mind-set to validate the benefits of performance optimization

Webcast Transcription

Yasmina Greco:         

Hello everyone. Thank you for joining us today. I’m Yasmina Greco. I’m with O’Reilly Media and I will be your host for today’s webcast. We would like to begin today’s program by saying a big thank you to Keynote for sponsoring our webcast today and let you all know Keynote pioneered the online performance monitoring space nearly 20 years ago and we continue today as the leader in helping the world’s best known brands such as Philips to make every digital interaction count.

Our global cloud-based solutions running on the Keynote global testing and monitoring infrastructure supports more than 700 million mobile and website measurements each day, giving our customers the unparalleled ability to ensure the quality and performance of their digital experience across mobile apps and websites around the world. Thank you again, Keynote.

Joining us today is Hakim el Fartasi from Philips Electronics. Hakim is a business service management consultant at Philips Electronics, Nederland B.V. Eindhoven, the Netherlands. He started in 2008 as service delivery manager online B2C. During this time, he defined the online monitoring service, made it a business priority, and helped set up several online improvement projects. He also introduced Keynote mobile web performance monitoring into the organization, setting up internal reporting services and related support processes.

Today, Hakim will be talking to you all about Connecting Your Business to Web Performance: The Philips Journey.

And folks, at this time, it is my pleasure to turn the program over to Hakim for his presentation. Hello, Hakim.

Hakim el Fartasi:        

Hi Yasmina, thank you. Hi everyone. Welcome to my webcast on the mobile performance management implemented for Philips. As the introduction said, I have been working for Philips on this since 2008. I’m actually a CGI consultant who was hired to do this for Philips. OK, going to Philips, the company, Royal Philips, as many of you know, is a Dutch-based company mainly focused on healthcare and consumer lifestyle with an online presence in almost 62 markets. We want to ensure that the release that we did on 2013 with regards to the mobile platform would give the best focal user experience.

For this, we had to introduce the user experience performance management. So, this presentation is basically gonna tell you how we implemented the performance management and how we raised it to a more business level.

So, going to the next slide; like I said, this presentation will focus on the release we did on 2013 with regard to the global mobile pages. These were all dedicated mobile pages specifically created for the device that they were meant for. Our initial rollout was in three main countries, China, the United States, and the United Kingdom. Since we didn’t want to measure all our 62 countries, this gave us the best global impression of what the performance would be in the various regions.

Before continuing with the next slide, I would like to ask you a polling question. So, what I would like to know from you is how much do you think in percentage that the visits that you get to your websites is actually from a mobile device? And I will give you 20 seconds to answer the question. OK.

And subsequently, I also want to send you another poll, and here I basically would like to get an answer on what strategy you currently have for your mobile pages. OK, seems that all have answered. Let me just send the poll results – and the second, OK. So, the answer on the second poll was somewhat expected and the first poll is also somewhat like we expected in the sense that a lot of mobile visits are very limited, especially when it’s not optimized for the platform itself.

What you will see in the current presentation is when we improved that, we will see an increase in the amount of mobile visits that we will get. Now, to get the results of our project more on a business level and to get also the business buy in, we basically had to identify the three main areas with which business and analytics determine the success of their campaigns, one of them, the primary being the goal completion rate.

The goal completion rate is basically the funnel that marketing designs for a user to navigate through. And what you want to know in this area is whether or not the user was able to complete all the steps within the specified time or whether he broke off the navigation. The second part that we wanted to focus on was the increase in online traffic. So, with normal campaigns, you would expect some kind of increase in the traffic. In the past, we saw that campaigns were successfully implemented, but we did not see the resulting visits with regards to those campaigns.

So, as I was discussing, the three areas are the part that we focused on and how we wanted to make the results relatable to the business and the last one being the buy intention. So, we actually wanted to ensure that the functionality of the purchase on the website was monitored and reported on as this directly relates to the potential revenue that we would get from our websites.

Moving on to the next slide; so, besides identifying the area that directly relate to the business and how we can make it relatable to them, we also had to identify what areas we would need to take into account before going ahead with the improvement. First of all, being the first-time user, so basically what we have here is the amount of traffic that we get to the website and in general you can say that most of the browsing users are users that actually go through the funnel are basic first-time consumers or first-time visitors. So, part of our measurement requirement had to be that we would measure from a first-time visitor perspective, the second part being repeat visitors.

Repeat visitors usually identify themselves with a user ID to log on to backend systems and to perform something like warranty checks and that sort of thing. So, we also wanted to include the repeating visitor in our measurements.

And lastly, we also had the first-party content that we needed to take into account and see whether or not improvements would also affect that area or not and how much we would be able to influence them. Now, being as the intermediate part like the ISP, the Internet Service Provider, the content, the third party and eventually also responsive design, these were all part of taking into account when designing our measurements criteria.

Moving on to the next slide, identifying our measurements criteria also meant that we had to identify our main area of influence and the parts that are basically user preference and Internet in general. By Internet in general, I basically mean the Internet Service Provider, the network state and that sort of thing. Especially when you look at the Asian countries, you can see that the network becomes a significant factor in the performance management part.

And of course the mobile platform and device basically meant that we had to focus on market research to provide us with what are the most popular platforms and what are the most popular devices. The areas that we could influence are basically covered by the normal IT processes as in frontend development and backend service.

Moving on, so having identified all these areas of what we could influence, what we couldn’t influence, what we wanted to have measured and how we wanted to make it relatable to the business, we basically created a mobile KPI proposal using the user preference parts, i.e., the mobile OS and the provider ranking but also how long people are on average willing to wait on a mobile page before navigating to a different page or to even a competitor. So, using this information, we basically created our eight seconds KPI based on a 3G connection. Please bear in mind this was in 2013, so 4G was not in play at that time.

Having identified all these areas and having KPI defined for IT as well as business, as this was an end-user KPI, we basically could go into the improvements areas or the improvement parts of our project. The most important insight that we got when performing the improvements parts, and one of them being the analysis part, is the fact that some elements have basically the most weight on your website performance. Starting at the top, you basically have your base page. That’s the code that is serviced by the backend and is mainly managed by managed operations.

All the other elements that you see in this waterfall; firstly, the style sheets and the JavaScripts and then at the end, the blue area where it says the image and graphical elements. So, we knew when we want to make our improvement, we had to focus on the frontend part and do the optimization on those elements specifically. So, using this insight and our monitoring tool, we were able to create this trend line. What you can see in this trend graph is the three countries as specified at the beginning, China, U.K., and the U.S.; China being the dark blue one, the U.S. being the orange-colored one.

In the six months that we implemented this project, we were able to reduce the loading time to below our set KPI, which basically meant that from an IT perspective, we achieved our goal and we could basically move on to the next improvement. But as I said, this project was not specifically done to improve the IT value as it was to improve the business value. So, we went on, and we first of all compared our page weights with Alexa and the Keynote Retailer U.S. Index. Using this information, we were able to determine that our mobile retailing sites were well within the acceptable page weight limits without actually affecting the user experience in that sense.

After these improvements and the things that we basically achieved with this project, I want to share two more polls with you. And these polls will just give me an idea of what you expect the improvement to be within the business basically. OK. And then I would like to ask you the last poll. So, push poll to audience. OK, so I think everyone answered. Let me just – where’s the poll results? OK. And the last one, OK. Well, I think I can surprise you with the average answers because the actual improvement that we did was 50 percent and the improvement on the statistics, on the visitor statistics, is actually 90 percent. OK, let me go back to the presentation.

So, what we did was once we achieved our KPI, we went to the Analytics Department and basically asked them what are the visitor rates, the unique visitors and the repeat visitors. And this is basically what we got from the various countries so you basically have from left to right, the U.S., the U.K., and China.

With the U.S., we see the most significant increase also because the average user in the U.S. is more likely to purchase something via the mobile platform as opposed to say the Chinese average consumer. On average what we saw with China is that they mainly use the browsing to navigate to different products, but the purchase is mainly done by local retailers….being one of them or being actually rather a large one of them.

So, what we then did was basically superimpose our performance graph on, in this case the U.S. visitor and unique visitors’ statistics. Now, the line that I would like you to focus on is actually the green line, and the green line shows the unique visitors. So, basically what we can say by this graph is that at the end of our project, we saw a dramatic increase of almost 90 percent in the unique visitors count and 50 percent in reduction of the loading time of the individual pages on average.

When we show this graph to the business stakeholders, we basically achieved the goal that we set out at the beginning of our project, making it more relatable to them and to also give them sort of a return on investment. Without being able to give the firm numbers, the return on investment ratio was about 1 to 50 for every dollar that we used in this project. We basically got $50.00 back in potential revenue. So, this project we basically used also for sort of lessons learned because we didn’t want to restrict this improvement only to the mobile platform.

So, the next few slides will just cover the various areas that we had to involve in our project to achieve the improvement that we were hoping for, first one being the performance optimization. Like I said in the previous slides, this basically means frontend optimization in the sense that most of the gain in the performance would be met if addressing the different frontend elements, one of the main areas being the JavaScripting.

If they are not optimized, then you basically had a very negative impact on the user experience as he would get a white screen before the script is actually executed. The next areas were basically the web content optimization. Our content management chain needed to have safeguards. That’s certain content specifically designed for mobile would also be used for that specific platform. What we did see in some cases was that content was not optimized and still pushed to the mobile platform, which resulted in a very heavy page.

The next area that we had to involve is, of course, the code optimization. Developers creating code for websites need to be aware of the impact that performance has and need to also basically optimize their coding to that end. And lastly, we had, of course, the web transformation, i.e., the project, the improvements, that were done on web pages. So, within the release management, we had to embed the performance part in it. Performance from release management was usually just seeing if you can get the actual coding to work. What we did was we included actual performance monitoring on the development area and the prerelease area.

So, that’s basically what we did with the various agencies within Philips. Outside of Philips, we were not able to make this improvement visible due to actual improvement without our two partners, Keynote and Key Performance, Key Performance being our most direct partner for the Europe-based locations. And that’s actually the end of my slide so I would like to thank, again, Keynote for giving me this opportunity to share this webinar with you and I hope you find something useful for your organization. Thank you and that’s it.

Yasmina Greco:         

Thank you so much Hakim for presenting an outstanding webcast for us all today. We do have a couple of questions that have come in. OK Hakim, a question, how did Philips come up with the IT and biz KPI? What data led to the decision for example?

Hakim el Fartasi:        

OK. Well, actually, that’s a pretty interesting question because we did not really focus on the IT KPI. With regards to IT KPI, you can say OK, if we ping a web server and we get a reply, we can say OK, the availability is 99 percent. What we actually wanted to do was, we wanted to have a KPI that represents more of a business value. So, our focus was on the end user experience and the KPI that we set was actually defined as a business KPI, which basically meant that we were counting on a collaborative effort from IT operations as well as development to achieve that KPI basically. Hope that answers the question.

Yasmina Greco:         

Thank you, another question here: How do you come to the eight seconds as the threshold in the KPI proposal?

Hakim el Fartasi:        

Yes, the eight seconds KPI is a difficult value to determine. Basically, what you’re talking about is how much are you willing to invest and where do you see the major drop-off points being when users are basically navigating through your site. For us, what we saw was that eight seconds was the ideal combination between the effort required to get the eight seconds and the drop‑off point of the average user on mobile by the way because for .com, it’s actually three seconds or two for retail sites.

Yasmina Greco:         

Thank you. Let’s see, another question here. Is the monitoring done in production live systems or preproduction?

Hakim el Fartasi:        

This is done in production live systems since that’s the area that we actually were interested in. The pre-launched or the pre‑live area was still considered an IT part so they basically did their performance monitoring at that end, but not as we did it with the end user being the main focus point.

Yasmina Greco:         

OK. What did you use for the performance monitoring?

Hakim el Fartasi:        

For the performance monitoring, we used the Keynote systems to monitor the various mobile websites. We are currently also using those same systems to monitor our normal .com websites.

Yasmina Greco:         

OK, another question here. Did you do any specific improvements for China, or was it the same as U.S. and U.K.?

Hakim el Fartasi:        

No, China did show us several areas that we had to take into account, one of them being we have sort of a unified platform globally for Philips and the China project actually showed us that we had to make more specific pages for China. Certain providers were not going through like YouTube, so we had to remove all that third-party content that were getting blocked by the firewall using local, Twitter, local Facebook equivalents, that sort of thing and also with regards to the content size itself but also where we hosted our data. In the past, we hosted everything from the U.S. With China-specific, we created our own datacenter to achieve the best performance in that country.

Yasmina Greco:         

Thank you, another question here. How did you get the visits data?

Hakim el Fartasi:        

Well, that’s actually pretty straightforward. Since analytics is one of the – yeah, well, most embedded processes within marketing, analytics can tell you based on the amount of visits, the amount of buy button actions, what the, how do you say it, potential revenue would be. So, basically just linking our performance metrics with the analytics gave us the insight that we wanted and gave us the graphs that you saw.

Yasmina Greco:         

Thank you folks, great questions coming in. Are the business and technical improvements that were seen correlational or causal?

Hakim el Fartasi:        

Hmm, interesting question. I would say causal because that’s basically what I’m showing in my presentation. Since we saw the increase in visits basically at the end of our project; on the other hand, we did have a campaign launched at the same period. It intended to get more traffic to our mobile so yeah, I think it’s a difficult question and I think especially when you talk about analytics and the result of performance, it’s yeah, I think it’s a difficult question to answer, but I would say causal.

Yasmina Greco:         

Let’s see here, another question. How did you verify test results in nonproduction environments?

Hakim el Fartasi:        

OK. So, we actually did not focus on the internal test results. We were basically mainly interested in what the end result would be, although we did go through the standard change process before releasing it. But again, we weren’t that interested in what the internal test results were.

Yasmina Greco:         

OK. Let’s see, next question: what were some of the improvements you made on the front end that were the most impactful?

Hakim el Fartasi:        

I think I can divide them in two areas or two main areas, one being the content, talking about the visual content, i.e., the images but also eventually YouTube, films and that sort of thing. So, we had one of our major improvements being that the reduction in the content size.

The other improvement was specifically on how we implemented our style sheet and our JavaScripts, moving them all the way up to the top of the page and making them as minified as possible, especially when they’re only specifically for the mobile platform. Those were actually the two main areas that we found resulted in the best performance improvements.

Yasmina Greco:         

Thank you, looks like we have one more question. OK, a couple more coming in. What monitoring tools is Philips using?

Hakim el Fartasi:        

That’s Keynote. So, for the synthetic user monitoring, we use Keynote’s infrastructure.

Yasmina Greco:         

OK. And what kind of detail in the reporting of the tools was particularly useful for improvement?

Hakim el Fartasi:        

OK. What we found very useful was the object trending graphs. The object trending graphs basically showed us what individual elements took the longest time. So, certain JavaScripts came out very clearly using those graphs and also, of course, the scatterplot. I think those were basically the most useful in helping us determine what area to address next basically.

Yasmina Greco:         

Thank you. Let’s see another question. Are there measurements, synthetic tests from specific node or taking real user monitoring into account as well?

Hakim el Fartasi:        

OK. I think the term real user monitoring is used in this case as the tagging solution. So, basically, you tag every site and then you can measure how the individual user is actually experiencing the site. That’s not the way we monitored this project. What we did in this project was basically synthetic monitoring of predefined flows and having them performed from a datacenter by Keynote, which always was the best-case scenario. So, we basically just measured the mobile site from Keynote infrastructure system.

Hakim el Fartasi:        

And I assume – I hope I made the correct assumption with regard to the real user monitoring part.

Yasmina Greco:         

Thank you, next question. There was a performance improvement that increased the number of visitors. Was there a significant improvement in the conversion rate as well?

Hakim el Fartasi:        

Yes. That’s basically what we got confirmed by our analytics teams and also by the financial controller so yes, we did get the confirmation that it’s still a potential increase in revenue, but yeah, we got their confirmation, yeah.

Yasmina Greco:         

Perfect. And looks like our final question, you said you moved the JS to the top of the page. Did you embed the JS in the HTML or still load it from an external file?

Hakim el Fartasi:        

It still loaded from an external file that basically is the result of the way of coding within Philips.

Yasmina Greco:         

Excellent. All right, I don’t see any additional questions coming in. So, with that, Hakim, we are gonna say a very big thank you to you for spending your time with us today and for sharing The Philips Journey with us all.

Hakim el Fartasi:        

It was a pleasure, thank you.

Yasmina Greco:         

We’d also like to thank all of you that attended our webcast today, and we had people from all over the world. So, for those of you that got up late or stayed up early, we thank you so much for coming and joining us today and hope you benefited from the event.

And in closing, we would like to say a big thank you to Keynote for sponsoring our webcast today and let you all know Keynote pioneered the online performance monitoring space nearly 20 years ago and they continue today as the leader in helping the world’s best known brands such as Philips to make every digital interaction count.

Their cloud-based solutions running on the Keynote global testing and monitoring infrastructure supports more than 700 million mobile and website measurements each day, giving customers the unparalleled ability to ensure the quality and performance of their digital experience across mobile apps and websites around the world. Thank you, again, Keynote. Folks, this will conclude today’s webcast.

Duration: 39 minutes

Back to Top