FAQ about Onsite Analytics

Do Onsite Analytics include any data from our Google Analytics?

No, the Onsite Analytics are based on Insider data.

Would using Insider Object (IO) affect the accuracy of numbers?

Yes, it would. Without IO, system rules can stop functioning with a change on your website. Let’s say you changed your cart page’s design, this can cause us to stop tracking revenue and cart counts if you are not using IO.

Can "incremental conversions" or "incremental revenue" be negative?

No, the incremental values display the increased values. 

How can the incremental revenue or conversions of a month be higher than those of 3 months?

This is due to the nature of incremental revenues and conversions. If you have positive values for 2 months and receive a negative value for 1 month, the overall value for 3 months can be lower than the incremental value of a 1 month-period. Let's say you have 300.000 incremental values in April, 450.000 in May, and 0 in June. As you receive a negative incremental metric in June, it is displayed as 0. If you decide to take three months, you will see a value less than 750.000 as the June has negative incremental value.

Why do I sometimes see a summary metric (incremental revenue) as identical to a campaign's incremental revenue?

The summary metric for the incremental revenue is a formula. Since the onsite campaigns do not have an attribution model, this formula is used instead of the raw data with attributed revenue. That's why Insider provides the campaign with incremental revenue as a summary metric if a summary metric is less than a single campaign's incremental revenue.

What does a 0% significance mean?

It means that the conversion rates of the variant and control groups are close to each other.

What does an 85% or above significance mean?

It means that the conversion rates of the variant and control groups are not close to each other. A variant can have a greater or less value compared to the conversion rate of the control group, but not be equal or similar.

Can the Bayesian probabilities change?

Yes, they can. In low sample campaigns, the likelihood to win can change.

What happens if a variant has 0% significance, but the bayesian is almost 100%?

Since the Bayesian is a separate significance calculation, it does not have to correlate to the significance values.

Is there a relation between incremental metrics and Bayesian?

No, there isn't. The Bayesian is not related to the calculation or importance of the incremental revenue. They are mutually exclusive metrics.

Why cannot I see our campaign on the Onsite Analytics page?

First of all, you need to make sure to remove all the applied filters. If you still cannot see your campaign, it might be due to the number of impressions and conversions. Only campaigns with at least 10 impressions and 1 conversion are displayed on the OnSite Campaign Analytics page.

• If a CG (A/B) campaign has fewer than 10 impressions and 1 conversion for both variant and control groups, it appears under the 'Without Control Group' tab. When the campaign reaches 10+ impressions and 1+ conversions, it is displayed under the 'With Control Group' tab on the Onsite Campaign Analytics page.
• For campaigns under 'With Control Group' analytics, you can use the 'show 0 conversion campaigns' filter to list those with 10+ impressions and 0 conversions.
• Campaigns without a control group and 0 conversions are not visible on the Analytics page in either tab.

Why is the incremental revenue of the Purchases From Clicks goal greater than that of the Purchases goal?

The incremental revenue of the Purchases From Clicks goal might be more than that of the Purchases goal in some cases. For example, if your control group does not have any corresponding clickable element, your conversion rate would be 0 in this case. According to the formula of the incremental revenue metric, if the conversion rate of the control group equals 0, then the revenue of your variation group will be equal to your incremental revenue. For this reason, the incremental revenue when you select the Purchases From Clicks goal seems higher than when you select the Purchases goal.

I created a campaign with a traffic allocation set to 80%-20%. However, I see the campaign listed under the Campaigns Without Control Groups. What could be the reason?

If the campaign has less than 10 impressions and no conversion (both variant and control groups), it is listed under the Campaigns Without Control Groups. A campaign should have more than 10 impressions and at least 1 conversion to be listed under the Campaigns With Control Group.

If a customer purchases 2 items at once, is it counted as 1 conversion or 2 conversions?

It is counted as 1 conversion no matter how many items are purchased.

When can I report Bayesian values?

You can report campaigns along with their probability of win. However, it's essential to note that if a campaign has a low sample size, these values may still fluctuate. If the incoming data remains consistent, the probability to win will stabilize at X%."

Is there a relation between incremental metrics and bayesian?

No there isn’t. Bayesian has no relation to the calculation or importance of incremental revenue. They’re mutually exclusive metrics.

Should the sum of incremental revenues for each day in a month equal the total incremental revenue for that month?

It is normal for the summation of daily values not to be equal to the total incremental revenue due to the incremental nature of the metrics. Here are three points to keep in mind:

• Negative incremental revenue values are displayed as 0, but they contribute to the monthly calculation, potentially resulting in a lower total compared to the sum of daily values.
• The summary metric calculates unique revenues. Incremental revenue considers daily unique revenue, but for a monthly view, it's unique for the entire month. This difference contributes to a discrepancy from the total of daily values.
  • In the summary incremental revenue formula, Unique Revenue x ΣIncremental(revenue) / ΣVar(revenue), the summation considers campaign-level incremental revenues within the selected date range. This is calculated with the formula Var(Rev) - ((Var(Imp) x CG(CR)) x CG(AOV)).

    For instance, if a campaign's CG conversion rate for a single day is 3, this rate can increase to 10-15 within the month, causing the monthly incremental value to decrease. This calculation logic explains why the total of day-by-day incremental revenue may not equal the monthly value.

• Campaigns listed on the OnSite analytics page can change daily. A campaign needs at least 10 impressions and 1 conversion to appear. For a monthly calculation, a campaign may reach these thresholds, influencing the overall value calculation.

Why do I see different revenues for Smart Recommender campaigns on the OnSite Analytics page?

The attribution works differently for the data shown on the OnSite Analytics and Smart Recommender Analytics pages:

• Smart Recommender analytics attributes revenue to both campaigns and products. For example, if a user views a campaign for product X and makes a purchase within the attribution window, both OnSite and Smart Recommender analytics will display the revenue.
• However, if the user views the campaign for product X but purchases product Y, OnSite analytics will show revenue, but Smart Recommender analytics won't.

Where can I find unique revenue and unique conversion metrics for my OnSite campaigns?

You can obtain these values from the exported report summary tab. The generated reports, based on date selection and applied filters, include unique revenue and conversions for the relevant results.

Can I export 1+ year data for my OnSite campaigns?

Data is available on the OnSite Analytics page for up to the past 1 year, and the exported data is also limited to this timeframe.

How can I include the clicks data on the file while exporting the OnSite analytics?

Before exporting the OnSite analytics, make sure to select "Clicks" as your conversion criteria. This way, clicks will be displayed as a conversion criterion along with the conversions column in the exported file.