Monitor your experiments

Once your experiment is running, you can begin viewing data for your experiment within your campaign and compare it to your control data. You'll find arrow icons in your account performance data that'll help you figure out how certain you can be that the change in data is due to the changes you've made. You can also download your data or set up your account so that people you choose can receive regular emails with the experiment report.

Viewing your campaign data during experiments

As soon as you start your experiment, we'll collect data for your control and experiment results.

View your overall results

  1. Sign in to your AdWords account at http://adwords.google.com. Sign into your account
  2. Click Campaigns. Campaigns tab
  3. Click the date menu and select Custom date range. Edit the date range
  4. Select a date range. Select your dates
  5. From the Segment menu, click Experiment to show the control and experiment details for each campaign. Data for your experimental changes will appear in the "Experiment" row and data for your control group will appear in the "Control" row. Display both your control and experiment data

See how individual keywords, ads, or ad groups performed in your experiment

  1. Sign in to your AdWords account at http://adwords.google.com. Sign into your account
  2. Click Campaigns. Campaigns tab
  3. Click the date menu and select Custom date range. Edit the date range
  4. Select a date range. Select your dates
  5. Click on the campaign you're using for your experiment. Select your experimental campaign
  6. Click Keywords, Ads, or Ad Groups depending on what you'd like to see results for. Choose what you
  7. From the Segment menu, click Experiment to show the control and experiment details for each keyword, ad, or ad group. Data for your experimental changes will appear in the "Experiment" row and data for your control group will appear in the "Control" row. Display both your control and experiment data

Try it out

If you can't see your experiment data

Make sure:

  • Your date range includes time when your experiment was running.
  • The ads in your experiment have been approved to show.
  • That any experimental bids are above first page bids.
Understanding the experiment report

If your experimental data is statistically significant, meaning that it's likely that any differences in performance aren't due to chance, we'll display an Arrow up up arrow or Arrow down down arrow next to that data depending on whether your performance has increased or decreased. As many as three arrows can appear in the same direction, and the more arrows in the same direction, the more statistically significant the results are. One arrow means that we're 95 percent certain that the change is not due to chance, two arrows means we're 99 percent certain, and three arrows means we're 99.9 percent certain. Two gray arrows Arrows in opposite directions mean the results are not statistically significant.

The more statistically significant the results, the more likely that the results you see in your experiment will continue if you apply these experimental changes. The more traffic you have on a keyword, ad, or ad group, the faster you're likely to get statistically significant results.

Understanding your experiment's results

How you read the results of your experiment directly relates to what your advertising goal is and how you operate your business. For example, a higher clickthrough rate might not be worthwhile for you if you haven't seen an increase in conversions on your website. Conversely, a higher average CPC might be worth the extra cost if you're attracting more high-quality clicks.

Here are some general tips for evaluating your experiment's results:

Even if the performance of particular keywords, ads, or ad groups declined, you might be doing better overall in your campaign – or vice versa. For example, you could have been using keywords that were competing with each other for the same traffic – having multiple keywords with lower clickthrough rates could mean more traffic for other, perhaps less expensive, keywords in your campaign. 

For example, you can have a clickthrough rate in your experiment that has three upward arrows next to it – meaning its performance compared to your control is almost certainly not due to chance – but the rate might only be 0.01 percent higher, a trivial amount to you. Moreover, maybe your advertising goals center more around average CPC or impressions than clickthrough rate.

Try to look at larger time periods in one-week chunks, instead of one day at a time. This will help you get a bigger picture of the data that's less affected by the fluctuations in traffic caused by the day of the week. 

Some things just take more time to develop. That's why making conclusions based on results that are not statistically significant can be misleading or just plain incorrect. Because larger date ranges include more data, you're more likely to see subtle variations the longer you wait.

  1. Evaluate your campaign as a whole
  2. Statistical significance doesn't necessarily mean it's important
  3. Choose a longer time frame
  4. Wait for experiment results to become significant
Understanding differences in impressions between control and experiment groups

Even if you create an experiment that uses 50 percent of your auctions for experimental changes and 50 percent for your control, these groups might get different amounts of impressions. This can happen because the performance of keywords, ads, ad groups, and bids might be causing fewer or more impressions.

For example, if you create an experiment with an experimental bid that's 50 percent higher than your bid in your control group, that higher bid might result in your ad receiving many more impressions because it's winning more auctions.

Downloading your experiment report and scheduling regular email reports

You might want to download a report from a current experiment or have emails sent to people regularly with your experiment results:

  1. Sign in to your AdWords account at http://adwords.google.com Sign into your account
  2. Click Campaigns. Campaigns tab
  3. From the Segment menu, click Experiment to show the control and experiment details for each campaign. Display both your control and experiment data
  4. Click the Download download button. Download your experiment data
  5. Give your report a name. Name the data file
  6. Select what format you'd like for your report. Choose the format of your data report
  7. Choose which segments you'd like to apply to your report. You can use this to evaluate data from a certain time period, or to look at just a subset of your data. Select which data you
  8. If you want to email reports, click Email and schedule report. You can then specify who should receive the report and how often they should receive it. See the Email and schedule report section
  9. Click Create. Create your report

Try it out

Ending your experiment

You can stop your experiment at any time:

  1. Sign in to your AdWords account at http://adwords.google.com Sign into your account
  2. Click Campaigns. Go to the Campaigns tab
  3. Click on the campaign whose experiment you want to stop. Select your experimental campaign
  4. Click Settings. Click Settings
  5. Under "Advanced settings" near the bottom of the page, click Experiment if the experiment section isn't open. Select whether to keep or discard experimental changes
  6. Click Stop running experiment. Click Stop Running Experiment
Try it out

When you stop your experiment, your campaign will keep your experiment data unless you delete or apply the experimental changes. However, your experimental changes will no longer affect your keywords, ads, campaigns or bids.