Change initiatives usually don't fail because the change was introduced too early, but rather because it was introduced too late after problems had already manifested. Godin notes that change almost always fails due not to being premature but to occurring too late after issues have already surfaced and become harder to address. The key is recognizing the need for change before problems become deeply entrenched.
9. DOMContentLoaded
Parsing of the document is finished, the Document Object Model is ready, but
referenced style sheets, images, and subframes may not be finished loading.
18. “An artist's job is to show people what was
always there but they never noticed.”
-Andrew Hamilton
Editor's Notes
Who of you think WebPerf,speed, short reaction time is important?
Who of you have struggled with getting enough time/priority to actually do important stuff around it?
What made the difference?
Price Money Winner 1950: $57,459
F1 2013 winning team (Red Bull) $98.97m
You might be in it just for fun or to make the web a better place.
Deepest Respect
But in reality webperf is only interesting for a company because of the additional money it can make.
It’s not a hobby – it’s a business
In the end your salary is influenced by the speed of your sites.
Too often WebPerf is seen as a cost for IT.
Something that needs to be done, but other things take priority.
It is often put into the Non-Functional Requirement bucket.
Because no-one has connected the dots properly yet.
So you need to be able to do it!!!
You need to be able to build the WebPerf business case for anything you do on that topic.
Difficult - you need to prove speed increases revenue, user satisfaction , brand value
or whatever the goal of site is.
Many studies and examples have been published.
Famous Walmart, Mozilla, Amazon,…
Great case studies
But they are not yours and you can’t use them to build your business case.
You can and should however use them as 3rd party verifications to the topic and to get the right people in your organization interested.
More case studies and lot more tips
Lets say you’ve convinced management
you invest time&effort
site speeds up
And this is the result
Cool
But user behavior doesn’t seem to change - no change in revenue!!!
Wait! What?
This is something I see very often. “We’ve tried WebPerf - doesn’t do it for us!”
In almost every case the data they looked at was not the right one.
In this case the wrong timer – DOmContentLoaded as reported by Google Analytics has basically no influence on the end user experience.
In other cases the timeframe they looked at was too short, results were not clean enough to distinguish between webperf changes and overall site changes, …
So which timer should you choose?
All of them make sense - why else would W3C have defined them?
But which one has the strongest correlation to your business success?
No good way to give general answer.
Often it is not one of the navigation timings but some custom timer relevant to your use case.
You need to make correlations to your business success criteria.
If you can’t do it for yourself find a data analyst and make him/her your best friend.
Business Success?
How do you measure that?
How do your different teams measure that?
If you want to build a business case for WebPerf try doing to for every team.
Understand their success criteria - their KPIs.
Bounce rate, session length, conversion,CAC, PPC, CPL,…
People think averages are the norm and in their mind if you tell them the average speed is 4,3s and we have a bounce rate of 38% this is what they think of.
How do you explain why speed matters if they have such an image in their head?
Question: Does you site load like the 1953 pit stop or rather like the 2013 Melbourne pit stop?
Answer: both
In reality the load time distribution looks something like this.
Basic shape follows a lognormal distribution with lots of people faster than the median load time but also a very long tail.
So you will most likely have a lot of people with a 3s load time but you will also always have some percentage of users with a 60+ s load time.
Question is how many of these.
Important to communicate and explain not just 1 aggregated number but percentiles and bets of all a complete distribution curve.
And then overlay the bounce rate for every one of those specific load times.
This visualization always help to make someone outside of IT ( or even within) to understand the speed influence.
Do the same for other business metrics like number of pages in session, conversion rate, number of reviews written, number of registrations, …
And when you improve your performance the new shape wont just move to the left as is but it will change its shape sligthly.
You still have a long tail but less people have a really bad expererience.
And more users fall into the performance sweet spot of your business metric.
So get in touch with our local data analyst, collect the performance data ( that is easy – much easier than lot of other data they are already playing around with).
Correlate Business value across teams:
Example CDN investment:
Compressing images (or files in general) is good for performance.
Visualize the business impact as done earlier.
But also calculate the savings of your CDN budget.
How much money can IT safe if they are delivering x% less data or y% less objects via the CDN?
Often it is quite significant – so you end up with a win-win situation.
WebPerf can be an artform.
You have to decide though whether you want to be the poor poet who is doing wonderful stuff biut noone cares about it
Or
Whether you want to become recognized for what you do, change the world and actually rake in lots of money for yourself as well
Von Carl Spitzweg - 1. The Yorck Project: 10.000 Meisterwerke der Malerei. DVD-ROM, 2002. ISBN 3936122202. Distributed by DIRECTMEDIA Publishing GmbH.2. Wichmann, Siegfried: Carl Spitzweg, München 1990, S. 57 ISBN 3-7654-2306-83. Cybershot800i, Eigenes Werk, aufgenommen 17. Juni 2011, Gemeinfrei, https://commons.wikimedia.org/w/index.php?curid=159093
Jack Mitchell: Fotoporträt von Andy Warhol mit Dackel Archie (1973) https://de.wikipedia.org/wiki/Andy_Warhol#/media/File:Andy_Warhol_by_Jack_Mitchell.jpg