Skip to main content
All CollectionsSelf-service: Generate & Predict by YourselfGenerate & Predict - by Yourself
7. Upload and View Self-service Results
7. Upload and View Self-service Results

Learn how to flow results of your self-generated content into Portal and how to use this centralized reporting.

A
Written by Amy Blakemore
Updated over 5 months ago

Once your content is in market, we want to track the content’s performance for two reasons:

  1. This data will empower you and your team to communicate ROI, conversions, and more to your internal stakeholders.

  2. Even more, the more performance data our AI can process, the more it will be able to feed your language model to improve its content recommendations for your specific brand and audience. Simply put: results are key to making the machine smarter!

Results ingestion happens differently depending on whether or not you are leveraging one of Persado’s integrations. If you'd like an integration, your Persado team can help review these options with you.

If You Have an Integration…

Results will populate automatically if an integration is installed. This is one of the key benefits to setting up an integration - we’ll set up a regular cadence where results flow back into Portal without any need for manual intervention.

If You Don’t Have an Integration…

We recommend using our template drop method if you don’t have an integration. This is an easy manual process for you to upload your results on a per campaign basis.

With this option, Portal will prompt you to fill out a results file template and then upload it back into Portal. This way, your results will be in the format we need.

We recommend you follow your standard observation windows prior to uploading results. After your initial upload, you can re-upload results as new data comes in. Simply click ‘Upload New Results.’

Reviewing Results in Portal

Once your results are available in Portal, you’ll be able to view Variant-level performance against your KPIs as well as insights into the winning Variants. You can review results any time by clicking the ‘Results’ step on the left-hand toolbar. Once on this page, you can also toggle results by your desired KPI (opens, clicks, conversions, etc.) depending on what matters most to you and your team.

At the top of your results, you’ll be able to see how the winning Variant performed - and, if applicable, how it performed compared to your control message. You’ll also be able to see insights into your winning Variant, such as the emotional language it used that contributed to its success and how it performs generally in your industry. You can apply these learnings to other campaigns you work on (whether Persado is involved in them or not).

When you scroll down, you’ll also see results on how each Variant performed in market, including your control (if included).

Once you have results in for a deployment with at least 3 Variants, you’ll also have access to aggregate insights and aggregate performance for your account in Portal. Each deployment will feed your brand’s unique language model so our AI can generate even better results.

How do I build more insights?

You will build insights by testing varied language. Keep the following in mind:

  • If you use “direct swap” (where you’re simply substituting Persado content for your control), you won’t build insights.

  • Persado recommends testing at least 2 Variants against your control. Consider selecting Variants that are different from one another in some way—that is, that use different Emotional, Narrative, and Descriptive tags. By intentionally curating Variants like this, you will begin to see trends in what works overtime.

  • Remember to upload your campaign results so you can have the most relevant, up-to-date analysis.

Understanding Statistical Significance

If you're curious about whether or not a message has reached statistical significance, you can reference the level of confidence vs. control column in your reporting.

The “level of confidence vs. the control” represents the certainty of a Persado Variant that will perform the same way, if we run the same campaign multiple times under the same conditions (texts, images, number of Variants, target population, audience segments, seasonality, etc).

For example, if a Persado Variant is winning vs. control and shows a 95% level of confidence, this means that if the same campaign runs 100 times, in 95 of them the same Variant will win again vs. the same control.

The same stands for losing. For example, if a Persado Variant is losing vs. the control and shows a 90% level of confidence, this means that if the same campaign runs 100 times, in 90 of them the same Variant will lose again vs. the same control.

In summary, the level of confidence vs. the control tells us how statistically confident we are for the winning and losing Variants.

Last but not least, you may sometimes notice a green badge ✅ next to the % level of confidence for some of the Variants. This means that the specific Variant has reached or exceeded the statistical significance level we need for analysts to be certain (the same result will occur again and again) of the specific Variant’s performance. In most cases, this has to be above 95%.

The level of confidence vs. the control is mathematically the same as the 1-(p-value), represented as a percentage.

Note that level of confidence vs. the control can only be calculated for a comparison between a minimum of 2 Variants, and one of them must be the control. Additionally, it cannot be calculated for estimated data (e.g., if we have estimated the performance of the control).

Leverage These Learnings

So, what are some ways you can leverage these results to your advantage?

  • Increase Stakeholder Awareness: You can easily share performance metrics with key stakeholders to articulate ROI.

  • Learn along with our AI: Humans and models both learn by testing. The best way to generate content today is with a human + AI partnership.

    • After enough testing, the Insights page in Portal will surface top-performing language with tags that will tell you what type of language performs best. For example, emotional language tends to perform strongly, and some emotions will work best for your brand, or for certain types of campaigns.

    • Since you’re able to see which emotions are resonating most, you’ll understand the type of language that encourages your audience to convert and, ultimately, order more. Emotions/sentiment may change over time and our platform enables you to monitor how these trends shift.

  • Inform Variant Selection: Employ your knowledge from the insights when deciding which Variants to deploy.

    • Look at the tags on the different Variants that tell you what language patterns are present. Look for high performing tags. Or look for any other patterns that you see – similarities to what is presented on the insights page.

    • Use your human creative intuition when authoring or generating your controls (see best practices here).

    • Let the AI generate Variants based on your controls. Then you make the decisions about what content to use. Persado has created a number of indicators to help with this (e.g., Performance Prediction scores, Brand Alignment scores).

If you have any questions about your results, how to leverage them, or how to read them, feel free to reach out to your Persado team member for help!

Did this answer your question?