Optimize Tool


This tutorial covers the Optimize tool in the Visual Toolbox from Campaign-Genius. You might be using this tool inside an email platform or as a standalone or within the Campaign-Genius app. It'll work the same way everywhere despite the relatively simple interface. It's a fairly complex tool. This tutorial is going to take awhile because the aim is to give you an idea of all the things that can be done to make use of the tool at heart.

What Optimize is for is live split testing of content. By content we mean an image and a hyperlink destination for that image. The Optimize tool will handle sets of content and it'll automatically rotate, randomize the views of that content and keep track of impressions, clicks and the click through rate. The option that's available when you use this is getting data about what content people actually like.

You could test buttons, you could test hero images, you could test different products all by putting the images and setting the hyperlinks in this tool. And then based on the statistics, you can manually select a winner and switch all subsequent opens to that winning content or, and I'll walk through this in the tutorial, use any of a number of automated methods to handle switching that content. This demo is going to use some simple colored buttons just to keep it fast, but it's not a, it's not a constraint of the tool. You can send, as I said, product and hero images just as well. So this is a sort of rudimentary demo set up here so that I can walk you through what the thing does. As I said, content is two pieces in this tool; an image that someone's going to see an email and the hyperlink destination for that image on the web.

We actually keep track of coordinating those things so that the person seeing the gold button is going to go to apple.com if they click and yes, all parameters such as UTM codes on that URL will be passed through the person. Seeing the pink button is going to go to Google in this, in this demo config here, one last one there for the sake of demonstration. So while there are three images and three destination hyperlinks in that set inside the email, you're only going to end up inside the email HTML, you're only going to end up with one image source tag and  one anchor tag destination link. The switching and randomization is done on the in the cloud based on requests for those things. You'll notice that there's a rather long ID for that particularOptimize set and the ID will be in common to both the image link and the hyperlink, the image source tag and the hyperlink. 

In this particular case when I pasted the HTML into the upper box there, our backend randomly said this particular IP address is going to get gold. I'm going to go back to that set and explain the randomization a little bit. 

There are actually three different ways to handle randomizing content. I talked about IP address a moment ago. We will automatically randomly assign one of the available rows to an IP address when the request comes in. If we don't see email or user ID as parameters on the image request URL and or the hyperlink request URL.  The benefit to providing email or user ID is that there's some stickiness. A person seeing the gold button in their desktop is going to see the gold buttonon their mobile phone if gold is available and if we've got their email or their UID.
 that would look like this standard query parameter UID equals something or standard query parameter email equals campaign genius.com.

Again, the two parameter names are email, lowercase or UID, upper case. We do not store email addresses or user IDs when the randomization is done. There's a hashed representation - an encrypted representation - that we save to keep track of that assignment, but we don't actually save email addresses or user IDs at all.

One thing that I flipped through quickly getting there, but it's important and useful with the Optimize tool is the history tool that sort of goes with it. And as part of the toolbox history simply allows you to get back to things that you've already made in the case of Optimize because there are a lot of possible config choices and because you might want to change content over time, that history tab is particularly useful on every time you make a change and save it, thatthat particularthat particular item is going to go to the top of the history heapso that it doesn't end up where you can't get to it.

So I just added an image that's alsofeasible with the Optimize tool. If you want to test other content or test other hyperlinks. nothing about this is, is fixed. So in this config I'd have four images each with their own separate hyperlinked destination. And if I say that configuration, if I paste thatif I paste that HTML into an email template, the recipient's going to get a random image and when they click on that image, they're going to go to the coordinated hyperlink for that image. Gold is going to go to Apple. Pink, as I said, is going to go to Google. I'm picking a winner in this config is up to the marketer. If you decided you wanted to go with green for whatever reason, could be statistics, could be designed, could be that you've decided to say 20% off and add a button that says 20% off.

That's simply controlled with the checkboxes. Um, if I save this config, everybody's going to get green and everybody's going to go to microsoft.comwith the email that's already been sent because this is live contact content. Let's put those back in the available set. And I will explain the statistics based methods for picking a winner. We've got a number of them. First one is high click through rate in relation to sample size. This is probably the most complex. So we put a calculator in here. If you're going to send a campaign to a hundred thousand people and your average open rate is 20% you're looking at 5,000 opens. Statistics formula says at 95% confidence, 5% margin of error. a minimum statistically viable sample is 356 so I'll get the calculator out of there. Let's say we're going to be conservative and up that number to 400.

With that configuration, we're not going to switch to the winter until each of those rows has had at least 400 clicks until we've hit statistically significant sample for each possible piece of content. We're going to leave them all available and continue to rotate through them. Once we pass that, once the very lastleast liked piece of content gets its 400 click, we'd automatically switch to the winter, clear the checkboxes on everything else and only deliver that piece of content from there on out. Naturally you'd want to know about that. So that's what the notification set up is for there. We will incrementally notify as impressions come in and will automatically notify when a winner is auto selected. So once again, first the first algorithmic selection is high click through rate. Once we've passed sample size and that's what the sample size calculator is for.

The second option is a little simpler. Um, it's simply switched to the high click through rate after a set number of open. If we were expecting 5,000 for a campaign and we just wanted to get the most out of that campaign, we might say, you know what, after we get 2000 opens, go ahead and switch to the high click through rate. Go ahead and switch to the winner. Um, similarly we could set a date. We might say, you know what, on Friday, no matter what else has happened, go ahead and switch to the high click through winner. So that's just date based. We could also say on Fridaylet's, let's automatically switch to row two. Look, that's a way of using the Optimize tool to do something like switch a switch to a sale or to a sale ended a graphic based on a deadline.

So it's not just optimization in terms of what the audience does, it's auto optimization and control for the marketer as well. And then the last option for tools like Zapier is to automatically send a winning RO fee, a web hook. So you could set a Zapier trigger that says based on actual sales, go ahead and update the content to the road that drove the most sales or the highest dollar volume or something like that. Whenever any of those changes are made, I'm saving them there. That doesn't necessarily require, by the way that I go back and update the content in my email HTML, that that chunk of HTML is identical to the chunk from the first time that I pasted it in. The configuration changes are saved in the cloud. They're not part of the URL for this particular tool. So shifts and changes about which rows are available or which method is being used to select, select a winner are, are automatically looked up for image requests and link requests that come in the door.

So that's a fairly long tutorial, but it's a fairly robust tool. It allows you to test visual content and coordinate that content with destination hyperlinks in order to get the most out of your email campaigns. Um, you could try different hero images. You could try different products with different landing pages, or you could try different images and go to the same landing page. We had a demo with a, with four different views of a piece of sports equipment. They all went to the same hyperlink, but that would enable us to test the hypothesis about which, which view of the product was most compelling to people. Um, we will continue to add documentation on the Optimize tool, but for now we hope you find this useful.