Transcript:
Hey everybody! Earlier than we bounce into at present’s episode, I wish to remind you that registration for Opticon 2022, Optimizely’s annual flagship convention, remains to be out there. The occasion might be again in-person this yr in San Diego, CA from October 3-5. So please go to optimizely.com/opticon to save lots of your seat! Once more, Opticon 2022 registration remains to be open, so should you haven’t already, make sure to get your tickets now at optimizely.com/opticon and hope to see you there!
Laura Dolan:
Hiya everybody, and welcome to the Optimizely podcast. I’m your host, Laura Dolan, and we’re joined by Griffin Cox, who’s a Senior Technical Product Supervisor right here at Optimizely, and he’s going to speak a few main element of our DXP at present, characteristic flags. Welcome, Griffin. How’s it going?
Griffin Cox:
Hey, I am good. Joyful to be right here.
Laura Dolan:
Thanks a lot for becoming a member of us at present. Let’s simply dive proper in. I do know there’s loads to unpack. So let’s begin out by telling us slightly bit about your background and your historical past at Optimizely.
Griffin Cox:
Positive. So I have been with Optimizely nearly 4 months now, however earlier than becoming a member of Optimizely, I used to be really a buyer of Optimizely’s. I labored at a FinTech firm first as a software program engineer for about 4 years, after which I grew to become a product supervisor earlier than I switched jobs to Optimizely. And whereas I used to be an engineer at that FinTech firm, I wanted a software that would assist me show the worth of this expertise I felt it might be good for us to make use of when it comes to efficiency and conversions and all that stuff. And we had a homegrown software, like I am certain a lot of our listeners do for this form of experimentation. And it was woefully insufficient and really arduous to make use of. So I made the case for Optimizely after evaluating it to a lot of our opponents, together with Adobe, which we really already had, and it proved itself many instances over. After which I ran our experimentation program at that FinTech firm for a few yr. After which I figured, eh, I like this product a lot. Why not work for the corporate behind it? So right here I’m.
Laura Dolan:
That’s superior. That is like saying how a lot you like the iPhone. You are like, I feel I’ll go work for Apple. After which you find yourself getting a job with the corporate product you imagine in that’s actually, actually cool.
Griffin Cox:
That could be a little bit of a dream come true for certain.
Laura Dolan:
Completely. That’s an incredible story. You are one of many fortunate ones simply to get to work on a product you imagine in and nonetheless get to work together with your ardour on daily basis. In order that’s superior. Fast query for you, Griffin, to not go off subject, however are you a fan of “The Large Bang Idea”?
Griffin Cox:
I would not describe myself as an excellent fan, however I’ve seen it. And I did discover the identify of this episode, however I will allow you to clarify it.
Laura Dolan:
So yeah, so “Enjoyable with Characteristic Flags” is in reference to “Enjoyable with Flags”, which is a section on “The Large Bang Idea” the place principally Sheldon, Dr. Sheldon Cooper and his girlfriend, Amy Farrah Fowler do an online collection referred to as “Enjoyable with Flags”. And principally they speak about vexillology, which is the examine of flags and the completely different animals which might be on flags and completely different states and international locations and what completely different flags imply. So I observed we had Characteristic Flags as a part of our merchandise. I used to be like, okay, I’ve to do a podcast referred to as, Enjoyable with Characteristic Flags. So right here we’re. And we cannot be speaking about vexillology at present. We’ll be speaking a few very completely different sort of flagging. So I’ll flip it again to you Griffin, and inform us, please, what’s characteristic flagging?
Griffin Cox:
Yeah. So merely put, Characteristic Flag is a cut up in your code the place you both do one factor, a distinct factor, or maybe many different issues. You’ll be able to give it some thought conceptually a bit like a lightweight change, at the least in the simplest instances the place you’ll be able to management a characteristic or the way it capabilities, the way it’s exhibited to customers remotely, should you’re utilizing software program like ours. And it actually offers you management over the characteristic when it is in manufacturing. It allows experimentation on that characteristic, which I am certain we’ll get into slightly bit later. And it might probably additionally make it easier to within the improvement course of as effectively. So I am completely satisfied to leap into any of these matters. Simply let me know which first.
Laura Dolan:
Yeah. So if I am decoding this appropriately, it sounds slightly bit like a toggle change. Am I proper in assuming that?
Griffin Cox:
Sure. Yeah. It is in probably the most easiest case it is on or off. Within the extra superior case it is on and the way, versus on and the way a distinct means versus probably off. So you may get slightly bit extra complicated with it, however fascinated by it like a toggle, completely an effective way to go.
Laura Dolan:
Cool. That helps me perceive slightly bit extra. So what are the assorted approaches to characteristic flagging?
Griffin Cox:
Yeah, so characteristic flagging, even when now we have a developer listening, they might not know the time period. Although they might, it is type of an business customary time period now, however they undoubtedly know the idea. And so in its easiest type with no software like Optimizely normally takes the type of a continuing or a worth that does not change within the code. You set the worth to true or false normally. After which in your code, you reference that fixed. And also you say if present buying cart icon then, after which the code that exhibits the buying cart icon, else, do not present that code. Proper.
Griffin Cox:
And that’s the most rudimentary form of characteristic flagging instance. The issue with utilizing a tough coded form of Characteristic Flag like that’s that to launch that Characteristic Flag, it is advisable undergo your software program improvement lifecycle. So that you’d have to jot down the code, take a look at the code, get it reviewed, get it deployed, possibly do a post-release test, and bam it is out. And there are such a lot of factors of failure in that course of. Even should you do not run into some extent of failure, it is nonetheless normally a sluggish course of until you are an organization like Google, and you bought all of it found out. However then you definately in all probability have your individual characteristic flagging resolution.
Laura Dolan:
True. True. So what does characteristic flagging need to do with Full Stack? How are they associated?
Griffin Cox:
Yeah. So should you take that idea and also you principally make it distant, so that you need not undergo that software program improvement lifecycle, you’ve a pleasant UI the place you’ll be able to simply flip it on or off with a click on or change the way in which that the characteristic is configured by typing in a couple of phrases. After which once more, one other click on and growth, you are off to the races. We noticed that as integral to enabling our builders to essentially experiment actually, on their very own with out experimentation type of like formally in thoughts, such as you need not do metrics to take advantage of out of a characteristic flagging. You’ll be able to, for instance, restrict the chance of a deployment. So if the characteristic has a bug, you simply flip it off. It may flip off in minutes as a substitute of probably hours, or when you’ve got an issue even longer than that.
Griffin Cox:
And so builders actually just like the idea. And since we have carried out characteristic rollouts, that are primarily characteristic flags, we simply name them rollouts. We have type of expanded on that topic to not solely allow safer improvement and deployment practices, however actually additionally allow builders to roll out a characteristic to simply percentages, components of their audiences. So it is type of loopy now that I look again that so many instances I launched a characteristic as a developer. I launched it to 100% of site visitors. So if one thing was improper, 100% of my site visitors noticed it, proper?
Laura Dolan:
Oh my gosh, no strain.
Griffin Cox:
Sure. Yeah. Numerous strain. And so, even then I used to be on the lookout for the appropriate method to launch it to possibly only a beta group of customers that possibly are extra resilient to issues to bugs, and possibly that would not hate us. They’d simply ship us a message saying like, hey, one thing’s damaged, it’s best to take a look at it. So you possibly can roll out to beta teams or you possibly can simply say, you already know what? Let’s simply roll it out to 10% of the general site visitors. After which let us take a look at our errors, and see if that site visitors causes some errors, by which case I would flip it off earlier than I thought of turning it up. And in order that form of rollout, that gradual rollout over time has additionally been actually useful for groups as they work on experimental options.
Griffin Cox:
And to reply your query in full it is actually, we took it one more step additional and now do experimentation totally on Characteristic Flags. And so any flag may be arrange in with any configuration. It may be on or off, however then you’ll be able to add a layer the place you primarily launch the identical characteristic on with this setting on, with this setting on, with this different setting, or possibly off, all to for instance 25%, 25%, 25% of your site visitors. After which you’ll be able to let the shoppers primarily present you what the appropriate configuration is moderately than needing to rely by yourself form of intestine intuition. I may offer you some examples, however does that make sense?
Laura Dolan:
It does make sense. In order that leads me to my subsequent query. How does characteristic flagging match into Optimizely’s historical past of digital experimentation? As a result of that is actually what we’re all about is A/B testing, multi variant testing. So I used to be questioning how this dovetails into that?
Griffin Cox:
Yeah. So have you ever ever heard our story from 2008, about how actually Optimizely supported Barack Obama’s presidential marketing campaign?
Laura Dolan:
I’ve heard a couple of issues about that, and I must know extra as a result of that’s so cool.
Griffin Cox:
Yeah. It is one in all my favourite tales to share. So primarily in 2008 in Barack Obama’s presidential campaign, clearly he had quite a lot of curiosity in his insurance policies and from his celebration, but it surely was type of a query of what can we wish to current on the web page to get volunteers to enroll, to help it, or to donate? And it is type of analogous to what quite a lot of firms wish to do. And in most areas it is like, how do you improve convergence type of usually? And on the time, there actually wasn’t a simple means to do that and measure it multi function software. And that was actually the delivery of our first product Internet. However earlier than I type of get into that, we helped the Obama campaign with type of a fast and soiled resolution the place we had been in a position to take a look at three pictures versus three movies, every with one in all 4 completely different calls to motion.
Griffin Cox:
So one picture for instance, was a colour picture. One other one was black and white along with his household, the movies. After which the calls to motion might be like, “be part of us now” or “be taught extra”, “signal me up”, or “join”. And so there have been 24 form of combos of content material and calls to motion. And the query was which one would result in the best join charge? And the group Barack Obama’s digital group took a guess, which ended up being improper really, which we love to listen to right here at Optimizely. So long as you discover the appropriate reply, ultimately, I feel you’ll be able to be taught as a lot or extra from being improper. And primarily, what got here out of it was that one of the best mixture was him along with his household. So aww and to “be taught extra”. So it was a bit much less aggressive than, “signal me up”.
Griffin Cox:
Proper. So folks possibly wished to be taught extra than simply plop down their electronic mail deal with and dig in, decide to signing up, proper. And so, should you assume that the development had stayed roughly the identical for his complete marketing campaign, so this experiment was solely run for a section of it, however should you primarily assumed that it had run for the entire thing, then roughly 10 million folks signed up over that time frame. And in the event that they hadn’t run this experiment, if they’d run with the mix of content material and name to motion that they had been planning to, it might’ve been nearer to about 7.1 million signal ups. So it is a distinction of about 2.9 million electronic mail addresses. So the signup charge was a lot larger. And sometimes, volunteers really grew to become volunteers at about 10% conversion charge. In order that 2.9 million further signups primarily translated into about 290,000 extra volunteers for his marketing campaign.
Griffin Cox:
Nicely, that is lots of people. And I feel if we stopped there, that might be a transparent instance of experimentation actually resulting in some good selections and good outcomes. However even when it comes to donations, the common donation was about $21 in the course of the size of the marketing campaign. And so 2.9 million extra signups really translated into about $60 million in projected donations. So it is some huge cash.
Laura Dolan:
That is implausible.
Griffin Cox:
It is quite a lot of further volunteers. And in our personal small means, we contributed to the success of his presidential marketing campaign. It is type of thrilling.
Laura Dolan:
Actually thrilling, each little bit counts. And it is wonderful how the speculation was improper, and the way it simply goes to point out you the way it’s so vital to simply take a look at completely different variables, even when it is simply altering the colour of a CTA, or if it is within the first particular person versus the second particular person, it is wonderful the completely different combos and the impact that it has on folks since you and I are shoppers. You and I are on-line on daily basis. We’re getting emails. We’re getting CTAs, and what causes us to click on on various things. And it is simply so cool to consider after we’re on the opposite facet of it, like what’s going to get in entrance of our viewers, what is going on to get forward of them, and what is going on to have an effect on them probably the most? So that’s so cool that we had such a, I do not wish to say an enormous position, however at the least we had a task in that. And it is a case examine that we are able to leverage going ahead. So that’s actually, actually cool.
Griffin Cox:
Yeah. So within the grand scheme of Optimizely’s historical past, we took that have and turned it right into a product. We went zero to at least one in a way, and principally introduced a brand new expertise to {the marketplace}, enabling net primarily based experimentation with ultimately a Wizzy Wig editor or a what you see is what you get form of drag and drop editor. So not solely did we type of make it an ordinary, however we additionally enabled non-technical people so not builders to make these sorts of modifications with our Optimizely net product with out probably needing to contain builders. So we have continued to type of construct on that. That was our first type of main zero to at least one. After which after all, cellular was beginning to take off round 2010-ish. And we then began to help cellular apps with cellular experimentation.
Griffin Cox:
We had been the primary to try this. After which it grew to become more and more vital to help actually superior experiments that required server facet experimentation. So we had been additionally the primary to try this. So right here at Optimizely, and I used to be shocked to listen to this too, however we persistently ship new options to the market. And I am undoubtedly fascinated by what our subsequent one goes to be and what our subsequent main iteration of that server facet full stack product builds on our characteristic flagging capabilities, combines them with experimentation, so that every one experiments in full stack 2.0 coming quickly, all experiments might be primarily based on Characteristic Flags, which lets you roll out a winner without having to undergo your software program improvement lifecycle once more. So we proceed to innovate and type of ask ourselves what’s subsequent. And so it is an thrilling historical past and I feel an thrilling future.
Laura Dolan:
Yeah, it sounds prefer it’s saving quite a lot of steps. And I simply wish to circle again to once you had been speaking about Wizzy Wig. And we drink our personal champagne right here at Optimizely. So I really edit our model pages by way of Wizzy Wig, and it is such a visit since you principally simply hit the edit button, and also you’re modifying on the precise webpage itself. And then you definately click on out of it, and all of your modifications are saved. And it is so handy and so streamlined, and I simply, I find it irresistible. I hope extra firms begin adopting that going ahead as a result of it is such a cool sort of expertise that I’ve by no means even labored with till I obtained to Optimizely. So I like that.
Griffin Cox:
Yeah. And a number of the greatest tech firms on the market have been doing this form of factor with their very own proprietary options for a while to various levels of success. However what Optimizely does is basically make it accessible to everybody, notably prospects with quite a lot of site visitors, as a result of it does take tens of 1000’s, a whole bunch of 1000’s of customers typically to detect a statistically vital distinction within the land of experimentation characteristic. Characteristic flagging could also be slightly bit completely different characteristic. Characteristic flagging can definitely profit firms of all sizes, however definitely within the experimentation measurement, like the larger you’re, the quicker you’ll be able to attain an perception, and the extra you’ll be able to improve conversions when it comes to like experimentation velocity as effectively.
Laura Dolan:
Very cool. So the place are we going subsequent with this? Do any of the latest acquisitions relate to Characteristic Flags?
Griffin Cox:
Yeah. Good query. So in case your listeners have been following Optimizely’s latest acquisitions, you may see that we’re shortly changing into a number one DXP. So a digital expertise platform, which actually wants extra than simply experimentation, though we’re an enormous a part of it. So related to, I feel this dialog a few yr in the past, we acquired a CDP, Zaius now often known as the Optimizely knowledge platform or ODP.
Laura Dolan:
Yep, I am totally conscious. I got here from Zaius!
Griffin Cox:
Superior. I did not know that.
Laura Dolan:
Yep, got here from that acquisition.
Griffin Cox:
Yeah. So then as you already know calling Zaius a CDP alone I feel is just not fairly doing it justice. It is usually a connector platform designed to essentially pull and share knowledge with tons of instruments, tons of instruments, together with Adobe audiences, Shopify, MailChimp, HubSpot, plenty of standard platforms. So I feel instance of that is with the ability to pull in an already outlined like Adobe viewers, mix it together with your Shopify gross sales knowledge should you’re a retail ecommerce web site, and with the ability to mix it additionally together with your first-party knowledge. So what you already know in regards to the buyer all into one section. And so our leaders right here at Optimizely acknowledge that form of functionality, that superior segmentation, machine studying primarily based segmentation as effectively, as actually highly effective and core to our want to be like a full DXP. Early on, we knew that we wished to combine ODP or Zaius with our experimentation merchandise, together with full stack and Internet to essentially unlock some new focusing on capabilities.
Griffin Cox:
So whereas each net and full stack have viewers builders in-built now with out this integration, they’re comparatively restricted in comparison with the superior capabilities that ODP provides out of the field. And so by combining these two merchandise, we’ll allow our prospects to focus on customers extra simply with much less improvement effort, but additionally extra particularly. So you’ll be able to run some actually extra superior campaigns when it comes to personalization in net, extra superior experiments in full stack, for instance, after which even goal lists of customers. And people segments once more in full stack for Characteristic Flags actually is a strong integration and that’s coming this yr with bulletins to return in Opticon in October.
Laura Dolan:
Superior. Yeah. And I just lately had the chance to check a brand new section builder in beta mode with our CDP. In order that was really actually cool that I obtained to see her from the again finish and see the way it really works and issues like that, helped me perceive the product slightly bit higher. And this has helped me perceive what we’re doing slightly bit higher as effectively, Griffin. So I respect you breaking it down and type of making it slightly bit extra digestible for our viewers, as a result of I do know this may get actually technical actually fast.
Griffin Cox:
It could possibly. These ideas, they possibly on the floor appear actually easy at instances, however they get complicated actually quick. And so once you’re speaking about changing a picture on a web page, that we have streamlined principally as a lot as attainable with our net product, however once you’re speaking about principally experimenting on completely different algorithms that energy an API server facet and utilizing full stack to try this, will get slightly bit extra complicated and also you want engineers to be sure to’re measuring the appropriate stuff and that you simply’re doing it safely, proper? However we right here at Optimizely try to make it as simple to run as many sorts of experiments as attainable. And I feel most significantly to achieve insights or stats sig as shortly as attainable. And we’re undoubtedly the leaders in that with our multi-arm banded algorithms, our stats accelerator, which these might be episodes type of on their very own. So we’ll save these for an additional time. However we’re all about getting prospects to worth as shortly as attainable. It sounds possibly slightly cliche, however one thing that actually powers every thing we do.
Laura Dolan:
Completely. And it is undoubtedly a precedence of ours, for our viewers to get the outcomes that they need within the quickest time attainable. And our software will assist them get there. So it is every thing we are able to do in our energy to empower them. So I respect all of the work that you simply’re doing, Griffin, and that your group is doing. Excited to know extra. So if our viewers needs to know extra about these things, the place can they go?
Griffin Cox:
Yeah. So I feel the simplest means for them to be taught extra is to comply with the hyperlink I am hoping you may put within the description.
Laura Dolan:
Completely. I’ll.
Griffin Cox:
Cool. Yeah. It’s kind of of a mouthful, however primarily our developer docs are implausible. In case you’re trying to be taught extra, now we have movies on YouTube. Now we have podcasts like this. We attempt to attain our market by way of as many channels as attainable. So plenty of other ways, should you search Optimizely even on Google, you will see loads of assets for all of those various things and for the remainder of our DXP platform. We, after all, have not talked about commerce. We have not talked about CMS. And I collect, you’ve got talked about these on different episodes. However undoubtedly…
Laura Dolan:
Oh, yeah. There’s so many various parts of our DXP. We may speak about all of it day.
Griffin Cox:
Yeah, it is actually thrilling. And over this subsequent yr, one in all our highest priorities is to combine our merchandise. So we spent quite a lot of time on this podcast speaking about full stack and net, our experimentation merchandise integrating with ODP or CDP. However I am excited to see the way it suits into our CMS, the way it suits into our commerce choices, as a result of quite a lot of our prospects wish to experiment and have not picked up our experimentation instruments but. So yeah, I feel extra to return on this subsequent yr, it is one in all our high priorities.
Laura Dolan:
Completely. And so they’re adopting one in all our main values, which is rarely cease enhancing. And I am an enormous believer of that. In terms of something you do, whether or not it is work associated or life associated, I am all the time desirous to be taught. So yeah, I like that that is one in all our greatest priorities. So I am proud to be part of that.
Griffin Cox:
Yeah. And I feel, to not converse for our CEO, however I imagine you mentioned this earlier than should you’re not digitizing your organization and optimizing utilizing instruments like ours, you are actually lacking out. Particularly, one impact of the pandemic was simply the transfer to ecommerce accelerating quickly, the transfer to digital usually accelerating, and more and more folks shopping for their groceries on-line, proper? Consuming streaming content material on-line, not by way of DVDs, these had been type of out the door quickly anyway. However a platform like ours actually simply helps folks get probably the most out of their migration to digital. And I am actually excited to be part of it. And once more, completely satisfied to be right here.
Laura Dolan:
Sure. Thanks a lot for becoming a member of us at present. Once more, I’ll put that hyperlink within the backside of the outline. Each episode will get a weblog. So if you find yourself following together with the weblog to this episode, simply hit that hyperlink, and go be taught extra. I am undoubtedly going to go be taught extra as a result of I wish to perceive these things. I respect you being a useful resource for us at present, Griffin, and thanks a lot for approaching.
Griffin Cox:
Yeah, my pleasure. Actually completely satisfied to do it once more.
Laura Dolan:
Thanks, Griffin. And thanks all a lot for tuning into this episode of the Optimizely podcast. I’m Laura Dolan, and I’ll see you subsequent time.
Laura Dolan:
Thanks for listening to this version of the Optimizely Podcast. If you would like to take a look at extra episodes or be taught extra about how we are able to take your small business to the subsequent degree by utilizing our advertising, content material, or experimentation instruments, please go to our web site at optimizely.com, or you’ll be able to contact us instantly utilizing the hyperlink on the backside of this podcast weblog to listen to extra about how our merchandise will make it easier to unlock your digital potential.