Jacquard's broadcast message experiment workflow snippet was created in partnership with Adobe. Adobe Solutions Consultants recommended this method to minimise the steps you need to interact with to set up each experiment and conform to platform best practices.

It's important to note what we mean when we say workflow snippet. The meaning is twofold. First, it means the snippet requires some configuration from you in order to function properly. You can read about what those items are in our Getting started article.

Second, it means you can in theory add additional activities before and after the ones we provide as long as you leave the activities we provide untouched. The snippet is designed to replace your standard audience selection and email delivery activities.

We've included Start and Stop activities for your convenience, but you can remove those and add your own steps to the beginning or end of the snippet if you wish.

Before we go further, let's have a look at the whole snippet.

Although the workflow may appear intimidating, there are only a handful of steps you need to actually modify for each experiment, and you can read about that process in detail here.

But before you do that, let's go over each step in order and talk about how it's contributing to the overall optimisation.

This activity does pretty much exactly what its name implies. With some simple initial configuration in your workflow snippet template, this step will call the Jacquard API to retrieve your language variants. All you have to do each time is paste your Jacquard experiment ID in the URL field under Execution panel.

You'll want to paste over {campaign_id} so your end product will look something like this:

This workflow snippet contains a few forks. These are simply used to keep the primary data set from Jacquard's API along for the ride while the workflow snippet carries out other necessary activities needed to automate the process of testing, deployment, and results sharing.

For context, the top path passes the API data to the portion of the snippet that formats results to send back to Jacquard.

The middle path passes the variants along to be assigned to subscribers.

The bottom path is responsible for pulling the correct subscribers into the workflow. This path will merge with the middle path after subscriber selection.

We'll take a look now at the activities on the bottom path.

This is one of the activities we mentioned earlier where you'll need to do the configuring. This is a standard Query activity where you'll choose what subscribers will be a part of this experiment.

Whether you typically use Audiences or Profiles, you can use the click and drag interface within the query activity to select your subscribers. The workflow snippet uses Email as the unique identifier, so ensure your subscribers have a unique email address.

This Segmentation activity will parse out your test population from the remainder.

This is another step you'll need to modify to ensure it's selecting the population size Jacquard asks for when you're generating your language.

The limitation set for the Test outbound segment should match the percentage shown under Jacquard's split calculation as shown below.

This Enrichment activity will randomly assign each of your subscribers a number corresponding to one of the language variants called from the Jacquard API.

This Enrichment activity will match the appropriate language variant to each subscriber based on the number assigned in the previous activity.

At this point, the middle and bottom paths of the Fork from earlier have reunited.

This Email Delivery activity will deploy your email to the test population.

You'll need to select the HTML for your experiment in this step and ensure you've included the Jacquard variable in the Subject field.

This Wait activity tells the workflow snippet how long it should allow your test to continue before pulling results. The default is four hours, but you can consult with your Customer Success representative to see if a different wait time is right for your audience.

This query activity pulls unique opens, clicks and unsubscribes related to your experiment from your Adobe Campaign Standard tracking logs.

The Fork allows the subscriber data to remain in the workflow while it fetches the logs.

This enrichment joins the data pulled from the tracking logs with the main set of data from the delivery: your list of subscribers. It is set to make this match on Email, which is why it's very important Email is a unique field.

This Update Data activity will record to a custom Phrasee Experiment Deliveries table in your Adobe Campaign Standard instance. This table should have been imported with the workflow snippet from the package Jacquard provided for you.

Essentially, this is purely for record-keeping purposes. Once a workflow completes, it's temporary data sets are dumped. This step ensures you will still have a record outside of the workflow of which subscribers received which variants.

This data will not be passed back to Jacquard. So, there are no PII concerns there.

Yet another enriching Enrichment activity is included here. This step aggregates all of the tracking data pulled in the previous steps using a Sum formula. This helps ensure Jacquard will receive the accurate numbers for each metric.

Remember that first fork in the workflow from earlier? This step joins that original top path to the middle path, combining the variant data set from the initial API call to the results data set we just grabbed.

The Enrichment step here is doing it's part to take the engagement data and put it into a format Jacquard's API can ingest.

This activity also handily calculates open, click and unsubscribe rates using the aggregated metrics from the previous activity.

This Fork allows the workflow snippet to simultaneously pass along the test results to the Jacquard API and choose a winning variant at the same time. We'll tackle the top path first.

As we alluded to above, this step is going to pass along those formatted test results to the Jacquard API.

The one thing you'll need to update in this step is very similar to what you had to modify in the first step of the workflow: adding in your Jacquard experiment ID and token. This will help the API identify which experiment in your project these results should populate.

This completes the top path. You'll notice another arrow pointing to this activity. That arrow is coming from the Remainder path. The workflow efficiently uses the same API activity to add your final send results to your Jacquard experiment.

Onto the bottom path. The results formatted for the API conveniently also work with this handy Deduplication activity.

This step is programmed to check the results and remove all of the variants except for the one with the best open rate.

That Fork up ahead is going to keep the variant data safe for us on the top path while we head down the bottom path to deploy the winner.

This Enrichment activity is going to add your winning variant to the Remainder population we created earlier.

This delivery activity is the same as the prior one, except this one will be deploying your winning variant. You'll still need to add your HTML and place the Jacquard variable in the Subject field.

This activity waits one day to allow your winning deployment's results to flow into Adobe.

This piece of the workflow snippet should look familiar. These are the exact same steps you saw earlier in the workflow that will aggregate your results and format them for Jacquard's API.

You'll note the Update Phrasee final results API activity is configured ever so slightly differently in order to ensure Jacquard knows these are the final results. During initial setup, you'll need to add your Jacquard experiment ID and token to this step as well.

You've been fully briefed on what each activity does in Jacquard's Adobe Campaign Standard workflow snippet.

Feel like you're ready to get going? Make sure you've completed everything in the Getting started article and added personalisation fields if you need them. Then, on you go to Deploying an experiment.