Custom Ad Creative Predictive Model to Increase ROAS

Akkio's self-serve predictive modeling feature simplifies ad creative optimization by enabling marketers to build custom predictive ROAS models for ad creative instantly. Try this recipe to analyze historical campaign data to determine the most high-impact visuals, copy, and messages to boost ROAS. Uncover actionable, data-backed creative insights to guide development with just a few clicks, skipping the lengthy guesswork of additional A/B testing. Akkio bridges skill gaps with an simple workflow that places intuitive advanced predictive business intelligence into marketers' hands to drive performance.

Gather Historical Data

Ensure your historical paid media data set has the proper variables to run this creative prediction. The data variables used in this particular recipe included Ad Name which describes creative variables tested within the name of the ad, campaign name, ad body, campaign objective, CPA, Clicks, CVR, and ROAS. You can include as many different historical creative variables as you would like to analyze in aggregate, as long as you have corresponding ROAS outputs for each.

Import Your Data

Import your data (via table or Akkio-supported integration of choice) in this categorized creative element format

Ensure All Labels of Data Columns Are Correct 

All columns labeled as “Number” should contain only numeric inputs, and columns labeled as “text” should contain only text inputs.

Use “Chat Data Prep” to Prep Your Data

Use the prebuilt Data Cleaning options to quickly perform basic data cleanliness steps (e.g. Standardize Date Columns) In this case, we removed outliers from the ROAS column, and removed empty cells as it will negatively impact predictions.

Create Your Predictive Promo Strategy Model

• Click “Predict” in the top navigation bar
• To train your model, click “Predict” not “Forecast” since we are creating a predictive promo strategy model. We are using “Predict” in this instance because we want to predict outcomes for data based on categories and numerical values. Examples are lead scoring, housing prices, employee attrition, etc.
• We would use the Forecast (Time Series) feature only for time-dependent outcomes based on historical data with time information, which we don’t have in this instance. Use “Forecast” to see sales projections based on time of year, temperature ranges based on season, energy usage, etc.

• Target the column of interest - in this case, ROAS, in the left-hand “Predict Column”

• Click the blue “Create Predictive Model” button
• Wait for the model to be generated. This may take a couple of minutes to a half an hour, depending on the size of the dataset 

Key Insights

The report generated once the model is complete is called the Key Insights report. These insights highlight correlations and driving factors that impact the outcome, which will help you understand where to act next

• Click Expand Key Insights
• Look at Top Fields 

• Click on creative-aligned fields such as “ad body” to view specific insights about which creative elements were predicted most likely to drive ROAS increases. You can see here that creative focusing on product features such as “pencil” or coin types like “Adjudicator” has a positive predictive impact on ROAS. From these insights, adjust creative strategy accordingly

##Conclusion

Repeat this process across all of your tracked historical ad data. The model will adapt to the fields within the data set as long as historical ROAS data is present for each creative variable tested. Reach out for a demo if you have more questions. Connect your own data and try the recipe yourself with our free two-week trial.