Optimizely Opal... what it does actually do?

At Opticon 2023, Optimizely announced its first AI product Opal. AI is definitely the new tech buzzword in 2024 and with promises that AI will be everyone's productivity saviour, its hard to argue otherwise. One of the issues with the term AI, is that it can mean different things to different people, so my big question is what does Opal actually offer?

If you have the same curiosity then this is the article for you. Regardless if your interested in purchasing an Optimizely product, your a current developer, a marketer or a product owner, read on to learn exactly what Optimizely Opal can and can not do to help you in your job🔥🔥🔥

What Is Optimizely Opal?

The first question to answer is what is Opal, is it a standalone product, is it available now, and maybe more importantly how much does it cost? The first thing to straighten out is that Opal is not a standalone product. This means that you can not purchase it directly from Optimizely and it does not have a set price. Instead, it is better to think of Opal as an initiative around AI. Optimizely Opal defines a suite of capabilities that are peppered within the entirety of the Optimizely platform. This means that you can find Opal capabilities in content creation (CMP and CMS), experimentation (Web and FX), and data analysis (ODP).

The next question to answer is what specifically does Opal do for us? AI is definitely a vague term but it conjures up images of the terminator sat in front of a keyboard trying to code some JavaScript! With Arny trying to steal your job, will Opal actually replace you?

Initially when I heard about Optimizely Opal, I assumed like a lot of other companies promising AI, Opal was just a fancy name for basically a ChatGPT wrapper, however, this is definitely not true. While, Opal does have a ChatGPT wrapper (which can be switched to other models and currently Optimizely supports 2 generative AI providers) it is much more.

As Optimizely Opal is a suite of features, this means in terms of AI it does more than one thing. This is why Optimizely Opal is a combination of different capabilities and technologies. Opal combines the latest generative AI services, like ChatGPT, combined with a bunch of proprietary machine learning algorithms that Optimizely have been building for over 15+ years.

When were talking generative AI, some companies might have a concern about privacy, data loss, security, GDPR, PII, etc.. So some useful things to note around Opal is that Optimizely uses ChatGPT with an enterprise license. This means that your searches will not be used for global training. Additional, Optimizely does not have a global learning model that combines all its customer data, this means a specific customers data is not shared with anyone else and no one company or individual can access it.

With Opal explained, lets see how you can use it to make you more productive and get a competitive edge! As someone who spends the majority of their working lives working with AB testing, lets start with Opal within the Optimizely experimentation platform.

Optimizely Opal Within Optimizely Experimentation

The Optimizely experimentation platform is composed by two experimentation modules, Optimizely Web and Optimizely Feature Experimentation. Optimizely Web is a client-side tool that comes with a visual editor that is aimed at marketers who want to run a CRO program on their website in order to improve conversion. While feature experimentation is known as server-side testing, more aimed at developers and program managers, that are looking for feature flagging and code based experimentation.

In terms of what Opal does within the experimentation platform, you will get access to mix of generative AI including content creation, some patentedmachine learning algorithms to help you get to statistical significance, and a chat. Below lists all the current Opal capabilities in both Web & Feature Experimentation.

Content and variation suggestions: This feature will allow an editor to highlight any text on their site and then Opal will suggest 5 different suggestions on how it could be reworded. The benefit here is time saving, instead of spending hours trying to guess what text will convert the best, instead, create an A/B test using auto-suggestions in seconds and prove it instead!

In the future this feature will be improved, so you can automatically create new variations based on the automated suggestions rather than just replacing content within the current variation!

Multi-arm bandit: A multi-armed bandit (MAB) algorithm is useful for any company that spends money on PPC. A MAB uses machine learning to optimize which variation to show a user in real-time. The learning is based on real-time popularity. In essence, Optimizely will dynamically adjust the distribution of resources based on the performance feedback received, balancing the exploration of new options with the exploitation of known successful ones. This adaptive approach maximizes overall campaign effectiveness by continuously learning and improving allocation strategies in real-time.

Stats Accelerator: Optimizely Stats Accelerator uses a patented statistical method that combines sequential analysis mixed with a low false discovery rate. This patented way of calculating significance was created by Optimizely in partnership with Stanford University. Stas accelerator uses machine learning on your data to provides faster and more accurate insights. It works with multiple metrics and also has a feature called stats reset built in. This was added to ensure your results are not impacted by any changes you make on your website while your test is running.

By accelerating the analysis of experiment results using machine learning, it enables marketers and developers to make data-driven decisions more swiftly and efficiently, ultimately enhancing campaign effectiveness and ROI.

Opal Chat: Within both tools you now get access to an AI powered chat, to ask experimentation related questions. You can think of Opal chat as a ChatGPT wrapper, although Optimizely can swap the API that powers the chat to any of the popular generative AI providers. The Opal Chat will allow you to engage in meaningful and contextually relevant conversations that provide you answers using a vast knowledge base (including the Optimizely knowledge base). This means you can use the Opal chat to ask about any experimentation terms you might not understand. In the future, you may even be able to ask the Opal chat to get ideas about what to test next. One thing to note is that no customer data is shared here and customer data is never shared!

Optimizely Opal within Optimizely Content Marketing Platform

The next Optimizely product where Opal can be used to save people a bunch of time is within the CMP (Optimizely Content Marketing Platform). CMP has three big benefits for marketing teams, better collaboration, easier marketing oversight for managers and a reduction in software costs. Basically, if you work within a marketing team and you need to create content, campaigns, events, or everything in between this is where CMP might help.

In terms of Opal within the CMP, you will get access to the same Opal chat that was mentioned above within the experimentation platform. I found having access to this type of chat within CMP a little more handy compared to experimentation because if you need some 'inspiration' on what to write next, or, the idea for a blog post, you can do all your research directly within the tool, without needing to have a second ChatGPT tab open in your browser. Optimizely also uses enterprise ChatGPT which will give you access to better data than simply using your own free account.

When it comes to creating content within CMP, you will find some Opal options here as well. For example, within the text editor you will find buttons to auto generate content and pick images.

I would say Opal is especially handy for helping teams to work with images. In case you didn't know CMP has an image marketplace module and an asset library module. Having an asset manager is handy to ensure team members follow brand guidelines by giving governance about what images and assets they can use within their marketing campaigns. Additionally, an asset manager can help ensure your team members can easily find the images, PDFs, PowerPoints and videos the company has created for them!

When it comes to images, Opal can be used to automatically tag your assets with relevant keywords. Lets say you pick an image with a picture of an office. Opal can automatically add tags like team, events, office, desk etc.. to that asset. Adding tags is boring but useful. Automating this process means your team has the benefits of using the tags to find assets easier, additionally, they could also use these tags words within your website to improve SEO.

Another Opal image capability is automated cropping. For example if you used the same image on desktop website and mobile, your images might look strange due to the differences in dimensions. Opal can add auto-crops to any image so you can ensure its optimized correctly for its intended display channel.

One big issue with automated cropping is the focal point. Sometimes simply cropping an image canter might make the image look strange. A better solution is to use an auto-focal point. This ensures when the image is cropped, the most interesting thing within that image is still centre. For any editor that has to do this regularly, I'm sure you can appreciate how boring and tedious, image cropping can be. Anything that can automate this process is a win in my opinion!

Data

Next up is the Optimizely data platform, ODP . ODP is the product that you would use to automatically track and store all your customer interactions within a database. Using this data, you can then create segments and run reports for analysis. The benefit of creating segments is that you can run personalisation campaigns to create targeted experiences for your customers, hopefully improving your conversion rates!

Opal has only recently been added to ODP and currently, it will generate automatic profile insights for you on a per individual level. Within the profile view, you can ask Opal to summarize all the interaction and information about any customer.

This is currently step one, in the next phase this automation will be added to the segment level. So you might ask Opal to generate a report of all your customers that have brought something this month, or, to create a segment for all your big spenders!

Optimizely Recommendations

Next, we can find Opal within Optimizely recommendations. Optimizely has a bunch of patented machine learning algorithms for both its content and product recommendation products.

Content Recommendations: If you run a website that is very content heavy with lots of blogs and articles and you have low site engagement and high-bounce rates, Content recommendations can be used to help make your site more sticky. Instead of just showing generic, one size fits all links, menu items and related articles, using personalised content has been proven to increase engagement and increase the odds of people sticking around.

Content recs has been a standalone product within the Optimizely tool stack for about 7-8 years. Content recs will automatically spider your site on a regular basis, it then uses AI and machine learning to create topic clouds of keywords. Based on your visitors, their interests and these topic clouds, content recs will then generate better recommendations for your customers. All you need to do is install a snippet of JavaScript on your site and then update your HTML to call the recommendation API and off it goes!

Product Recommendations: Not all websites will be content heavy. If you run an online store, it is likely that your site will mainly be composed from products. When your main KPIs are average order size, increase number of orders, order size, etc... content recommendations will be less useful and instead product recommendations will be more useful.

Product Recommendations is not new for Optimizely and has been around for 6-7 years. Product recommendations works in a similar manner as content recommendations. You provide Optimizely with a product feed. You install a snippet of JavaScript on your website and then using machine learning Optimizely will figure out better product recommendations. Within your site, you will then display a recommendation widget that talks to the Optimizely API and voila!

Both these recommendation tools fall under Opal as they both use proprietary machine learning and automation to provide better data for your customers!

Optimizely CMS

Finally, we can find Opal within Optimizely CMS. I won't go into the details of what the CMS does within this article. As I've written over 500 articles on Optimizely CMS on this site alone, use the search above to learn more!

The CMS has some big changes coming this year like Saas CMS, however, Opal will be a key part of the content creation journey for editors. Expect to see the same benefits of content creation within CMP here as well.

Currently, the main Opal use case within CMS is for translations. If you manage a multi-language website, you can use generative AI to convert your copy from one language to another. Translating copy exactly is hard and will still need a manual review, however, using a tool to do the bulk of the conversion first is a massive time saver for any editor!

That covers everything Opal can do at the start of 2024. I expect this will change over time but hopefully this gives you some good insights into what it does and how it can help. Happy Coding 🤘