About simulations
Beta feature
Simulation reports show the estimated marginal cost-to-objective value, cost, number of clicks, and objective value that you can expect for a portfolio at various levels of spending (cost) and the corresponding daily budgets or other targets. You can optionally customize the view to see additional traffic metrics, simulation settings, and only a specific simulation type (Weekly or Custom).
Types of simulations
-
Automated weekly simulations
-
Custom, user-generated simulations
Automated weekly simulations
Simulation reports are run automatically each week using the current portfolio settings. Automated weekly simulations are available only for periods in which the portfolio is optimized or active.
Downloaded weekly simulations
Each downloaded weekly simulation consists of one workbook. Each workbook includes the target for each of 20 step levels and the projected cost, clicks, weighted revenue (objective value), and conversion metrics included in the objective, based on the corresponding target. For the first 20 data rows, constraints weren’t applied; for the remaining data rows, constraints were applied.
On-screen weekly simulation details
On-screen simulation details show portfolio-level visual and tabular insights. For data by campaign, ad groups, bid units, or device, download the simulation instead.
Graph view
The graph view shows the expected objective value or other specified metric (Y-Axis Metric) for the spend target for each of 20 spend levels. The target midpoint is identified, and you can optionally change the target midpoint to see the predicted data using that value. Hold the cursor over any point in the graph to see the data for that point.
You can view the data with and without constraints applied, with constraints applied, and without constraints applied. When you view data that takes constraints into consideration, the applied constraints are identified above the graph.
Table view
The table view shows the target spend for each of 20 spend levels. It also includes the corresponding estimated cost, marginal cost to objective value, clicks, objective value, and conversion metrics in the portfolio’s objective for each spend level. The target midpoint is identified, and you can optionally change the target midpoint to see the predicted data using that value.
You can view the data with and without constraints applied, with constraints applied, and without constraints applied. When you view data that takes constraints into consideration, the applied constraints are identified above the graph.
Simulation settings
The simulation settings are shown as read-only beneath the graph or table.
Portfolio settings
To see the read-only settings for the applicable portfolio, click Portfolio Settings in the upper right.
Custom, user-generated simulations
You can create a custom simulation report for a single optimized or active portfolio using the current portfolio settings or using custom portfolio settings, with or without bid unit-level constraints applied, to see the results that those settings would produce without actually changing them. For example, you could create a custom simulation to see the effect of using a different spending strategy or learning budget without considering active constraints on bid units in the portfolio. You can view estimated performance at the portfolio, campaign, bid unit, and device levels.
Downloaded custom simulations
Each downloaded custom simulation consists of one workbook. Each workbook includes one worksheet for each specified entity level of simulation (portfolios, campaigns, ad groups, bid units) when data is available for that level. When you specify device-level data, each worksheet includes a Device column. Each worksheet includes one row with data for each applicable entity and (when specified for the report) and device type for each of 20 steps (for example, one row per campaign). The data in each row includes the projected marginal cost-to-revenue, cost, clicks, weighted revenue (objective value), the device type, and conversion metrics included in the objective, based on the corresponding target. The portfolio-level worksheet also includes the target for the step levels, and the entity-level worksheet includes the ad network, account, campaign, and (when applicable) the ad group.
On-screen custom simulation details
On-screen simulation details show portfolio-level visual and tabular insights. For data by campaign, ad groups, bid units, or device, download the simulation instead.
Graph view
The graph view shows the expected objective value or other specified metric (Y-Axis Metric) for the spend target for the specified number of spend levels (steps) for the simulation. The target midpoint is identified. Hold the cursor over any point in the graph to see the data for that point.
When the simulation was built taking constraints into consideration, the applied constraints are identified above the graph.
Table view
The table view shows the target spend for each of the specified number of spend levels (steps) for the simulation. It also shows the corresponding estimated cost, marginal cost to objective value, clicks, objective value, and conversion metrics in the portfolio’s objective for each spend level. The target midpoint is identified.
When the simulation was built taking constraints into consideration, the applied constraints are identified above the graph.
Simulation settings
The simulation settings are shown as read-only beneath the graph or table.
Portfolio settings
To see the read-only settings for the applicable portfolio, click Portfolio Settings in the upper right.
The Simulations view
The Simulations view lists weekly simulations and custom simulations. Administrator and some other user types can see custom simulations created by other users. All other users can see only the custom simulations they create.
The data table includes the progress of each simulation; a Target Midpoint column populated for each row; and columns for cost/impression/click/objective value predictions, actual values, and accuracy. You can optionally add additional custom columns (including lift metrics, such as actual objective value divided by cost).
Available actions simulations-actions
-
Customize the view to include additional metrics, including the estimated impressions; the actual cost, clicks, impressions, and objective value; the cost-to-objective value; the cost accuracy, click accuracy, and objective value accuracy; and the difference (delta) between the predicted and actual objective value and cost-to-objective value. You can also include columns for most of the simulation settings and the simulation type (Custom or Weekly).
-
Generate or rerun a custom simulation for a single portfolio. You can either create a new simulation or regenerate an existing simulation in the list.
-
Download weekly and custom simulations as Microsoft Excel workbooks in ZIP files.
-
Using the Spend Planner button, open the legacy Spend Recommendation tool, which helps you to identify the optimal spend distribution across portfolios.
Best practices
Monitor simulation reports in the following situations:
-
Before you launch a portfolio to estimate the performance you can expect with the corresponding portfolio settings; use at least two weeks of data. If the simulation results indicate lower performance than you would expect based on historical data for the included campaigns, investigate and resolve the issues before launching the portfolio.
-
After any major change to a portfolio, such as adding a campaign or changing the objective. If you make changes to the portfolio’s modeling start date, the weight of a conversion metric, or the click value of an objective, then wait until after 17:00 PST the next day to run the simulation, when updated cost and revenue models are available.
-
Periodically to monitor performance trends at the conversion metric level.