A high-level overview of the Dynamic Yield platform

Read the full transcript

Thanks for watching this video demo of Dynamic Yield’s end to end experience optimization platform where I’ll be walking though a high level overview of our many capabilities, as well as highlighting the overall usability of the product. Outside of what I’ll be covering here, we have many additional features that we’d love to show you. So if you’re interested in learning more about them, please be sure to contact us for a one-on-one demo.

Dynamic Yield supports multiple data collection methods to build a comprehensive omnichannel touchpoint history for every user that engages with your brand. Our client-side JavaScript plugin is the primary method that our customers use to integrate Dynamic Yield with their website. It records each user’s browser behavior and combines that with important contextual data about where the user is browsing on the site and what they are interacting with. We then leverage this data to build a holistic view of each customer and quantify their level of interest for specific products and services. Right now I’m browsing the website as a new user. The content is fairly generic and the product recommendations are seemingly random. As I begin browsing, Dynamic Yield will build an understanding of my affinities in real time based on where I show the most engagement. Using only my activity from a short browsing session, we were able to identify that I’m primarily interested in women’s clothing and less interested in other product categories. We can immediately leverage this understanding to personalize the site experience and serve product recommendations that cater to the users’ unique affinities. Individual affinity modeling is just one of many capabilities that differentiate Dynamic Yield in the market. Customers engage with brands on multiple devices, channels, and web domains. They expect a personalized experience that’s also consistent across these different touch points. Brands require a personalization engine that can support this detailed omnichannel customer profile, but without fully exposing the complexity of its underlying identity graph. Dynamic Yield strikes the right balance by automatically identifying users at the device level and then mapping those ideas to customer level identifiers, like a hash email address, customer loyalty ID, or whatever ID is most relevant to the business. We call this the customer unique identifier or CUID. As soon as I landed on this website, Dynamic Yield generated a unique ID and stored it in my browser’s local storage, session storage, and as a secure first party cookie. We call this device level identifier the DYID. It allows us to track the user’s behavior across multiple browsing sessions and personalize their experience from the moment they land on the site. We do this for anonymous users on both the website and mobile app as well. Once I identify myself and provide my CUID, which in this case is my email address, we can retrieve additional information about my omnichannel activity through Dynamic Yields unified customer profile API. The response from this REST API gives us a fully up-to-date record of the customer. The products they browsed on our website, what they purchased in the store, and so much more. Our clients leverage our unified customer profile API to give them a single view of the customer that they can integrate into their call center, POS, or any other enterprise microservice. In addition to our client-side script, we also provide mobile app SDKs for iOS and Android as well as a REST API that completely abstracts our platform from the device altogether. We call this our personalization API. And we use it to interface Dynamic Yield with devices in the physical world that were previously walled off to personalization, whether it’s a menu board in a restaurant drive-through or a self-service kiosk in a store. Our flexible integration options give our clients the power to take personalization anywhere. Customer data takes many forms.

So in addition to tracking event-level interactions like page views and clicks in real time, Dynamic Yield also provides a helpful admin interface to upload data in a batch file format. Our clients leverage this capability to upload their product catalog and integrate offline purchase events from their store POS for both anonymous and identified users. They can import additional customer attributes from their CRM, CDP, or any other kind of user data store and integrate content from their digital asset manager or CMS through what we call a variation feed All of the data that we collect about your customers is made available in the Dynamic Yield audience explorer. Analysts can profile their user data and build customer segments in real time. They can define complex logical statements in a simple interface using out-of-the-box segmentation conditions that include things like the user’s onsite behavior, geographic location, device information, referring source, and their offline transactions. Dynamic Yield can also ingest additional user data from your CRM, CDP, DMP, or any other system of record to further enrich your data set. This allows our clients to consolidate their disparate data silos into a single view of the customer.

In addition to rule-based segmentation, Dynamic Yield also provide support for what we call predictive audiences. These are segments defined by the user’s likelihood to complete a specific action, such as their probability to make a purchase in the next seven days. Our customers leverage predictive audiences to optimize their conversion funnel by personalizing it to the customer’s predicted purchase behavior. The Dynamic Yield admin UI uses an intuitive workflow to guide users as they set up a new experiment. I’m currently at our site personalization module and I’m about to walk through the steps that one would take to deploy a totally new personalized message to their site and then optimize that message to achieve the best possible uplift in their KPIs of interest. In this example, I want to increase the number of users that opt in to receiving marketing emails from my brand. I think an exit intent overlay will be a good way to capture their email address before they navigate away from the site, and I want to test my hypothesis. I’ll give this experience a name and then I’ll define how this message should be triggered. I’ll leave the frequency kept at once per session just so I’m not annoying the user by triggering the pop-up on every page. I can also configure some additional advanced settings like the position of the modal on the screen, the opacity of the backdrop and other minor tweaks related to the user experience. I’ll leave them at their default values for now. After I configure those initial settings, I can set up one or more targeted experiences to define how the message should be personalized based on who is viewing it, where it’s displayed on the site, and when the message is shown. If I leave these conditions blank, then I’ll effectively target every user with this message. For my use case, however, I only want to show this message to new users that haven’t already opted into our newsletter. I can use and/or statements to refine my targeting logic further if necessary I can choose to only target certain pages with this message, but in this case, it makes sense to trigger the exit intent overlay from any page on the site. I can also choose to display a different message based on the current date, day of the week, or time of day. After I define my targeting conditions, I can set up my test variations using the template editor. Dynamic Yield provides an extensive library of templated experiences that our clients can customize to fit their use case. Each template is a block of HTML, CSS, and JavaScript code, all of which can be modified by a developer if they wish. Dynamic Yield differentiates itself from other personalization platforms by allowing developers to insert custom variable tags into their template code using a dollar sign bracket notation. Different types of variables can be configured depending on what kind of value should be inserted into the code. Variables give non-technical users a simple interface to adjust the underlying code of a generic skeleton and customize it to fit their specific use case. Our clients reap tremendous value from our template architecture because it empowers their non-technical users to deploy new content to the site whenever they want without needing support from IT. Every Dynamic Yield account has one or more sites that each map to a particular domain, mobile app, or API implementation. Our fictitious Blueberry brand has 10 sites that map to its dev environment, production websites, mobile apps, and API implementation. We also have a site for the outlet as well. Right now I’m looking at the admin UI for our web implementation. The dashboard gives you a high-level overview of your campaigns, site performance, audiences and more. The audience explorer lets you visualize all the customer data that Dynamic Yield is collecting across your marketing channels and enterprise systems. You can profile the data set in real time and specify logical conditions to create new user segments. After you create a new audience, it will be saved to the audience manager. This is where you can monitor your key segments, report on their performance, and compare them with one another. Audiences can also be exported out of Dynamic Yield and into the client’s other marketing platforms. This allows them to leverage their Dynamic Yield audiences anywhere, whether it’s for retargeting and their display ads, or for generating look like audiences in social platforms. Site Personalization is where you can manage all of the campaigns running on your website. Each campaign falls into one of seven different categories. Dynamic Content allows you to replace static web components with personalized content. An example use case would be a hero banner on the homepage that targets different types of users with different types of content. Moving on to recommendations. This type of campaign is used to recommend products and content that are personalized to the customer’s unique profile. A single campaign can test multiple recommendation strategies as well as different widgets styles. There are dedicated campaign types for serving different forms of messaging like overlays and notifications. And a custom code campaign type for users that are comfortable with making changes to the site using pure JavaScript. We offer a visual editor where you can make simple changes to the website without writing any code, as well as our multi-touch campaign type, which wraps all of these capabilities into one and lets you test multiple touch points at once.

We power email personalization using three different modules. Our triggered email campaigns can support classic use cases like cart and browse abandonment, or they can be configured to trigger off of any customer event that occurs on your website. The Dynamic Content and recommendations modules let you serve personalized blocks of content and product recommendations in your email campaigns just like the campaigns that run on your site. Our acquisition modules let you test and personalize your prospecting and retargeting efforts. Dynamic links are used to optimize the click-through URL for an ad placement. You define which landing URLs you want to test and then we generate a link for you to use in the ad placement. When the user clicks the ad, the dynamic link will route them to one of the landing URLs that you’ve defined in your campaign. Later, I’ll elaborate on how you can use our dynamic allocation method to monitor performance in real time and automatically allocate more traffic to the landing URL that proves to be the most effective. Display ads lets you target your users with personalized ad creative while they’re browsing other sites. You define the targeting conditions and provide the creative that you want to test. Then we give you an Iframe to serve through your DSP. The Iframe populates with personalized ad creative by leveraging the DYID stored in the user’s browser. Landing pages let you test, personalize, and deploy entirely new landing pages to a sub-domain on your site. The pages themselves are hosted by Dynamic Yield, which allows teams to easily deploy new landing pages without any integration with their CMS. High-performing brands know that testing and optimization are critical to delivering a delightful customer experience. Once they begin to personalize that experience, however, they find that the number of possible user journeys becomes more diverse. That’s why Dynamic Yield is built to help brands scale their optimization efforts and ensure that every personalized touch point is also optimized to the maximum degree. Every experience in your campaigns can be A/B tested using our Bayesians statistical engine. Analysts can define the percentage of traffic that should be allocated to the control, as well as each test variation. They can also configure more advanced settings related to the significance level, test duration, and attribution. The variation stickiness defines whether the user should see the same test variation across multiple browsing sessions or just within the current session. Analysts can also provide one or more email addresses to receive an automatic notification once a winning variation is declared. Dynamic Yield chooses a winner by evaluating the performance of each variation as measured by the primary metric of success. Key metrics like click-through rate conversions and revenue are available to select by default. Custom metrics can also be configured using any kind of onsite activity or event. For my use case, I want to know which variation captures the most newsletter subscriptions. An A/B test is ideal for experimenting with changes that will continue to impact the user experience long after the test concludes, but sometimes you don’t have two weeks to monitor an A/B test and manually declare a winning variation. One example is on Black Friday every year when retailers are required to change their entire customer shopping experience and simultaneously optimize its performance. A/B testing in this scenario would require too much overhead and result in decision making that’s too slow to keep up with traffic. In this scenario, we encourage marketers to leverage our dynamic allocation method. With this approach, each variation is exposed to an equal amount of traffic when the experience is first published. We then monitor their performance in real time and use a multi-armed bandit algorithm to shift traffic away from variations that perform poorly and reallocate it to the variation that performs best. Dynamic allocation lets marketers harness the power of machine learning to scale their optimization efforts and automate their decision making using real time performance data. Dynamic Yield provide several interfaces to visualize your data and report on campaign performance. The dashboard summarizes your personalization efforts into one single view. I can see important metrics like the conversion rate, revenue per user, and AOV. Different calculations are shown depending on the length of my lockback window. Over here on the left, I can see the recent activity that took place in my Dynamic Yield campaigns. It shows every time a campaign was created, modified, or deleted. I can also view a detailed change log that includes the users and timestamps associated with every change in my Dynamic Yield account. Every time a user creates, modifies, or deletes a campaign, it will show up here. Even though I selected a 30-day lockback window in our dashboard, this log extends all the way back to when your account was created. It’s not limited to the duration of your lockback window. In the middle of our dashboard, we list all of your AB tests that are currently running, as well as those that have recently concluded and declared a winner. If I want to take a deeper dive into a specific campaign then I can click into a more detailed reporting view right from the dashboard. In this case, Dynamic Yield has declared a winner. I can scan the detailed test results in my report and then pause all traffic allocation to the losing variations with a click of a button. This will end the test and allocate a 100% of traffic to the winning variation. At the very top of this list, we identify any tests where Dynamic Yield has identified a personalization opportunity with its predictive targeting feature. We’ll talk about predictive targeting more later on in this video. At the bottom of the A/B test widget is where you can download a full log of your A/B test history. This includes a summary of every test that was created in Dynamic Yield so that you can easily format this information into your own internal reports, like an upcoming quarterly business review. Just like the activity log that we saw a minute ago, this file contains the full history of your A/B tests since your account was created. It’s not limited to the duration of your lookback window. On the right, we have a dedicated widget to highlight the impact of recommendations across your site. Direct revenue is for revenue from products that are purchased no more than 30 days after clicking on a recommended item. Assisted revenue is revenue from products that are purchased after clicking on a recommended item and then buying any product within the same session. We show the percentage of your total revenue that comes from direct recommendations revenue, as well as the total number of clicks on your recommendations and the page types in which at least one recommendation widget is live. Below the recommendations widget, we list all the experiences that are using dynamic allocation. These are different from A/B tests because we’re using a multi-armed bandit algorithm to allocate traffic in real time instead of allocating a fixed percentage of traffic to each variation.

In the bottom left of our dashboard is our audiences widget. The top 15 audiences from your audience manager are displayed based on their conversion rate and the average order value. Audiences at the top right are top performers. Their conversion rate is high and the revenue they spend on each purchase is high. The size of the bubble reflects the size of the audience. You can hover over the bubbles for more information. You can also click the audience manager and audience explorer buttons to drill down or see all of your audiences. To the right, I can see recent platform updates and click-through to more detailed release notes on the Dynamic Yield website. At the very bottom of our dashboard, we have our inspiration widget which features relevant use cases from our inspiration library. Users had access to more than 80 detailed use cases on our website to promote knowledge sharing and help enhance their personalization program. Our dashboard centralizes all of your personalization efforts into one single view, giving you informed look of how your program is performing and where it’s headed in the future. Right now I’m looking at a campaign for clothing retailer Land’s End. Campaign reports show high level performance metrics as well as the performance breakdown of each targeted experience in the campaign. You can modify the columns displayed in this table and export the data to a spreadsheet if you wish. In this campaign, we’re targeting users with a different visual navigation based on whether they have a stronger affinity for women’s, men’s, girls’, or boys’ clothing. I’ll click into the experience for boys’ clothing shoppers and take a look at the report for the variations that I’m testing. Right now I’m looking at the report for the current test version. This timeframe can be adjusted to any date range though. If you wanna see a full log of the conversions that were recorded by Dynamic Yield during this test, then you can use our revenue events log to download and analyze the full transaction history. I can analyze different performance metrics over time in this graph and see more detailed metrics in the data tables below. The primary metric for my test is shown by default, but I can also add any other secondary metrics that are relevant to my analysis. The probability to be best is the probability that a variation will outperform all other active variations based on the performance data that we’ve gathered. You can also break down these metrics by audience to see how each variation is performing at the segment level. If a variation performs particularly well for a specific segment, then it may be more optimal to target that segment with that test variation and personalize the experience. This is why Dynamic Yield gives marketers the ability to proactively identify personalization opportunities through what we call predictive targeting.

Every A/B test and Dynamic Yield has its variations performance automatically broken down by every segment defined in your audience manager. If predictive targeting sees that your campaign can benefit by exclusively targeting a segment with only one of your test variations, then our platform will notify the user via email and give them the option to add the new targeted experience with a single click. In this example, predictive targeting has notified the user that they can achieve an additional 7.8% uplift by targeting users in the search traffic audience with the control. All other users, however, should be targeted with the winning variation that’s being served through DY. When evaluating the results of your A/B tests, we also give you the option to filter out any outliers from the underlying data. This allows the analyst to confirm the validity of their test and ensure that their results are not being skewed by a handful of users. In this way, the campaign reporting view simplifies your analysis while also giving full transparency into the underlying data. Dynamic Yield uses recommendation strategies as the decisioning components of your recommendation campaigns. A single campaign can use multiple strategies. Like in the case where you want to test different strategies against each other. A single strategy can also be used in multiple recommendation campaigns. Therefore, it’s helpful to have a more global report of strategy performance to supplement your campaign level reports. The recommendation strategy report shows key performance metrics broken down by each individual strategy to help you understand which recommendations are the most effective across the site. Dynamic Yield continues to differentiate itself as a recommendations engine that not only provides advanced algorithms, but also flexible merchandising and filtering capabilities. I’ll define the page type where I plan to serve these recommendations since only some algorithms are valid on certain page types. For example, viewed together recommendations totally makes sense on a product page, but if I’m serving on the homepage, then viewed together recommendations wouldn’t be appropriate and therefore, shouldn’t be available to select as your algorithm. Dynamic Yield offers a wide variety of recommendation algorithms that are trained by customer data from both online and offline sources. This allows us to make better recommendations on the website based on what customers are purchasing in the store and vice versa. I can use something we call algorithmic fusion to power a single widget with multiple recommendation algorithms. For example, if my widget is a single row of four products, I can use the popularity algorithm to choose the first item, then collaborative filtering for the second, user affinity for the third, and recently viewed in the fourth. We also give you the option to shuffle the recommendation results. Instead of showing the product that receives the top score from our algorithm, we’ll randomly select one of the highest scoring products. This promotes product discovery and exposes your customers to more of your product catalog.

For my use case, I’ll use Dynamic Yields deep learning algorithm, which uses neural networks to identify patterns in customer behavior and then predict which product recommendations will lead to the most purchases. I’ll also choose to exclude recently viewed items, and then I’ll add some custom filtering logic for users in the men audience. I can implement rules that only includes some items or totally excludes some items from the recommendation results. I can also pin specific products to the top of my results and I can even choose to only apply the rule to certain slots in my widget. In my case, I want to only include items from the men product category and I want to apply this rule to all slots. I can search my product catalog and manually select the items, or use logical conditions to select them dynamically. I can set targeting conditions if I only want to apply this rule in a certain context or for users that fit a specific profile. In this case, I want to target users that fall into the men’s segment. I can schedule this rule to only apply at a certain time if I choose, but I’ll leave these conditions blank. I can add more rules to target my other key segments if I wish, and I also have the ability to pause a rule at any time. Once I saved the recommendation strategy, I can use it in a recommendations campaign and serve them as a widget through our client-side script.

Alternatively, I can choose to retrieve the recommendations as a JSON object through our API. This gives marketers the ability to merchandise and curate their sites’ recommendations, and developers, the flexibility to render those recommendations however they like. Now that I’ve configured my recommendation strategy, I’ll create a recommendation campaign to serve as the front end component of my homepage recommendation widget. I’ll target all users with this experience. And this time, I’ll use a template for my own library. I’ll use the variables tab to customize the template and select my underlying recommendation strategy. This makes it easy to test a new product recommendation strategy with the same widget, or keep the strategy the same, but make some aesthetic changes to the widget. After I click save and publish, the recommendations will be ready to serve. Emails can be triggered from Dynamic Yield to handle common use cases like cart and browser abandonment or as a response to any other kind of custom event. The cart abandonment email that I triggered during the front-end demonstration was powered by this campaign. The email is triggered once the user adds any product to their cart, but we delay sending the email for another three hours and cancel sending altogether if the user ends up purchasing the product or starting a new session. These rules are pretty standard for any cart abandonment email. But they can be easily configured to fit your use case. I can also set frequency capping rules to ensure that I don’t send too many emails to the customer too frequently. I can target a specific audience or group of audiences if I wish and then set up the variations that I want to test. I can easily change the variables just like any other DY template. I can also embed multiple recommendation widgets and choose the recommendation strategy that powers each one. You can set up granular merchandising rules in your email recommendation strategies, just like recommendations on your website. In the next segment, I’ll show how you can build these email recommendation strategies. For now, I’ll choose the user affinity strategy and populate the recommendations at the bottom of this email with products that cater to the user’s affinity profile. To make sure everything looks good, I’ll send a preview of each variation. Dynamic Yield can integrate with the client’s ESP to send triggered emails or send them through our default integration with SendGrid. Our clients also have the option to send blocks of personalized content and product recommendations through any ESP as well. After I click save and publish, the campaign will be live and ready to trigger based on the events we are collecting. In the previous example, we created a cart abandonment triggered email campaign. We displayed the items that the user abandoned at the top of that email and then selected a recommendation strategy for a second widget at the very bottom of the email. These strategies are defined in our email recommendations module. They can be embedded within a Dynamic Yield triggered email campaign, as we just saw a moment ago, or they can be copied and pasted into any other campaign, such as the daily batch and blast email.

You can use templates to customize the look and feel of the widget. Choose a recommendation algorithm and apply merchandising rules to the results. I’ll target this rule so that it only applies when the user belongs to my price sensitive audience. Once I hit save, this widget will be available to serve recommendations in both my Dynamic Yield triggered email campaigns, as well as any other email campaigns that I currently send through my ESP. Our dynamic content module is very similar. But instead of embedding a product recommendation widget, you can embed a block of content that’s personalized to the user. Notice that the campaign setup is just like a block of content on the website, whereas before we were personalizing a homepage hero banner and now we’re personalizing the banner at the top of an email campaign. Through flexible integration options and an intuitive workflow, our email modules demonstrate the cohesive, unified nature of the Dynamic Yield platform and enable our clients to scale their personalization efforts across all channels. Right now I’m in the app personalization module of our Blueberry iOS mobile app. I’ll add a new message to target users that abandoned their cart on web. I’ll trigger the message on page load and increase the size to cover the whole screen. I’ll target the web cart abandoners in men’s audiences. With DY’s native targeting, we can target across channels including web, mobile app, email, and display. I’ll create two variations to test my own template against the Dynamic Yield template. Once published, the messages will be served in the app without requiring any updates or approvals from the app store. In this next example, I’ll use Dynamic Yield’s visual edit tool to personalize the site navigation. I’ll provide the URL for the page where I want to preview the changes I’m about to make. I’ll target users in the women audience on all pages of the site. And then I’ll use the Wysiwyg editor to make changes. The tool makes it easy to navigate through the structure of different elements on the page. I can replace images, modify text, change colors, and even rearrange elements on the page. After I click save, I can preview the variations if I wish and then publish my test in the status page. Variations are marked with the default status of draft when they’re in development. They can be saved in Dynamic Yield and previewed on the website, but they won’t be deployed to the production site until their status is set to active. If I publish a variation to the site and then discover that there’s an issue, no problem, I can change the status from active to paused and seize traffic allocation to the problematic test variation. Once the variation is fixed, I can redeploy to the site by simply changing the status back to active again. In addition to personalizing individual touch points in the customer journey, Dynamic Yield gives you the ability to test and optimize multiple touch points across your entire conversion funnel with what we call multi-touch campaigns. In this example, I’m targeting users in the men audience and I’m testing to see which version of the user journey yields the most revenue. Think of each test variation as its own unique version of the customer experience. It has multiple touch points. It may include different types of campaigns. The variation displayed here has five touch points, three of which are dynamic content campaigns. One is custom code and the other is a notification. All the touch points, however, are created using the same simple workflow and leverage the same tools like our template editor and Wysiwyg editor. I’ll demonstrate how a new touch point is created by adding a welcome overlay to this user journey. I can target specific pages on the site. I can choose a template and modify its variables. Every type of campaign in Dynamic Yield can be tested, and multi-touch campaigns are no different. In this example, I’m allocating 10% of traffic to the control and 90% to variation one. Well, let’s say I want to test a second version of the user journey. I can easily duplicate a variation and modify its touch points. This makes it easy to scale and create as many variations as I need for my use case. Dynamic Yield clients leverage our platform to show users relevant items and guide them during their shopping journey. We track the performance of each product and then adapt to the context of user’s journey to make the most effective decisions.

Dynamic Yield is the Customer Experience Optimization Platform built for agility, providing all of the tools necessary to quickly build relevant digital experiences at every touchpoint under one roof. Through artificial intelligence, teams are then able to maximize the performance of each campaign, improving engagement, lifetime value, and revenue to help businesses gain a sustainable competitive advantage. In this high-level product demo, we walk through capabilities that help marketers quickly build and deliver better digital customer experiences at every touchpoint. If you are interested in seeing what else is under the hood, schedule a call to see an in-depth demo.