Skip to main content
All CollectionsTywin
Which hypotheses are supported and how do they work?
Which hypotheses are supported and how do they work?
Updated over a week ago

The following hypotheses are currently supported:

  • Channel Performance Differences

Investigates if specific channels drove the reported anomaly in the sales funnel.

Triggered: Always, for every new incident.

Retained: When the top channel or the two top channels combined (in terms of contribution to the total revenue change) has/have a contribution score of at least 60%.

Rejected: When the top or the two top channels combined (in terms of contribution to the total revenue change) do not reach a contribution score of 60%.

  • Campaign Performance Differences

Investigates if specific campaigns drove the reported anomaly in the sales funnel.

Triggered: The only case that this hypothesis is skipped, is when the "Channel Performance Differences" hypothesis is retained, and the contributing channels is Direct.

In case two contributing channels have been identified in the “Channel Performance Differences” hypothesis, and one of them is either “Direct”, then the “Campaign Performance Differences” hypothesis runs only for the other channel.

Retained:When the top or the two top campaigns combined (in terms of contribution to the total revenue change) has/have at least a contribution score of 60%.

Rejected: When the top or the two top campaigns combined (in terms of contribution to the total revenue change) do not reach a contribution score of 60%.

  • Product Performance Differences

Investigates if specific products drove the reported anomaly in the sales funnel.

Triggered: Always, for every new incident reported.

Retained: When the top or the two top products combined (in terms of contribution to the total revenue change) has/have at least a contribution score of 60%.

Rejected: When the top or the two top products combined (in terms of contribution to the total revenue change) do not reach a contribution score of 60%.

  • Product Price Changes

Checks if there was a change in product prices that could have affected the eCommerce funnel performance, for example, if any product was sold at a discount price.

Triggered: If the “Product Performance Differences” hypothesis has been retained AND one of the following conditions is met:

1. The actual revenue value is above the baseline AND the top contributing funnel step is Carts, Checkouts, or Orders.

2. The actual revenue value is below the baseline AND the top contributing funnel step is AOV.

Retained: If the price of the detected product(s) on the reporting date is lower than the average price of the previous 7 days by at least 10%.

Rejected: If the price of the detected product(s) on the reporting date is not lower than the average price of the previous 7 days by at least 10% OR If all historical values for the last 7 days are 0 OR if the reporting day’s price is 0.

  • Product Related Events

Investigates if there was a product-related event that could have affected the eCommerce funnel performance. It explores recent product-related news and announcements on Google News to identify possible explainers of the shift in sales and revenue.

Triggered: If the “Product Performance Differences” hypothesis has been retained.

Retained: When product news exists AND might have affected sales performance.

Rejected: When product news doesn’t exist OR exists but it’s irrelevant.

  • New vs. Returning Users

Checks if there was a different pattern between New vs Returning users that could have affected the eCommerce funnel performance. Explores the eCommerce funnel performance and compares it across New and Returning users to identify possible explainers of the shift in sales and revenue.

Triggered:Always, for every new incident.

Retained: When “New” or “Returning” users reach at least a revenue contribution score of 60%.

Rejected:When neither of “New” or “Returning” users reach at least a revenue contribution score of 60%.

  • Company Related Events

Checks if there was a general company-related event that could have affected the eCommerce funnel performance. This hypothesis explores recent company-related news and announcements on Google News to identify possible explainers of the shift in sales and revenue.

Triggered: Always, for every new incident.

Retained:When company news exists AND might have affected sales performance.

Rejected: When company news doesn’t exist OR exists but it’s irrelevant.

  • Special Dates

Checks Public Holidays and client-specific eCommerce calendars if available to evaluate if the reporting date is a "special date" in the US and might have impacted the sales and revenue figures.

Triggered: Always, for every new incident.

Retained:When the reporting day is a significant eCommerce calendar date (i.e. “special”) AND might have affected sales performance.

Rejected: When the reporting day is not a significant eCommerce calendar date (i.e. “special”) OR it is but it’s irrelevant.

  • Competitors News

Checks if there was an event/announcement related to competitors that could have affected the eCommerce funnel performance. It explores recent news/announcements related to competitors on Google News to identify possible explainers of the shift in sales and revenue.

The competitors used in this hypothesis are automatically selected by Tywin, by finding and including relevant companies to yours. However, if you want to add different/specific, and/or more competitors please contact Baresquare to provide your request, at [email protected].

Triggered: Always, for every new incident.

Retained: When news/announcements related to competitors exist AND might have affected sales performance.

Rejected: When news/announcements related to competitors do not exist OR exist but are irrelevant.

  • UX Issues - Website Errors

Checks if users encountered an elevated number of errors. It analyzes the number of website errors recorded, from Microsoft Clarity, and evaluates their impact on the reported sales funnel performance.

Specifically, the following error types are monitored:

- Excessive scrolling: A user session is flagged as 'excessive scrolling' if that session has higher amounts of vertical scrolling than the expected average. Excessive scrolling could indicate inaccessible or frustrating UX, poor content relevance, discovery, or nonstandard site interactions.

- Rage clicks: A page view or user session is flagged with 'rage click' if the user clicks multiple times in a clustered area in rapid succession. Each of these clustered clicks is marked as a rage click. Rage clicks can be strong indicators of user frustration, where users expect to be interactive elements. These clicks can also indicate key events in a session where a user might be uniquely frustrated.

- Dead clicks: Similar to rage clicks, 'dead clicks' suggest a poor user experience. Click interactions are marked as dead clicks when a user clicks on an element but gets no feedback in a reasonable amount of time. This click means that the visual status doesn't change, and there's no navigation away from the page. A dead click can signify broken elements, high latency requests, or misleading UX.

- Quick backs: A 'quick back' is a click that leads users away from the current page to another website. The user doesn't find the navigation useful and returns to the original page or website under a certain threshold of time. Clarity detects Quick backs when a user moves from a new to previous page in shorter time than defined threshold (the dwell time for that page).

Triggered: If the actual Revenue value is below the baseline AND the top contributing funnel step is Product views, Carts, Checkouts or Orders.

Retained: When the percentage of sessions with errors is above the 30% threshold set.

Rejected: When the percentage of sessions with errors is below the 30% threshold set.

  • Promo Calendars Checks

Reviews the company's promotional calendars, if available, to identify any unexpected campaign launches or pauses.

Triggered: If the company’s promotional calendars have been shared with Baresquare AND the “Campaign performance differences” hypothesis has been retained.

Retained: When the campaign(s) identified was/were launched or paused unexpectedly.

Rejected: When the campaign(s) identified was/were launched or paused expectedly.

  • Product Availability Issues

Checks if there was a change in product availability that could have affected the eCommerce funnel performance. Focusing on the top contributing products, identifies product availability changes that can explain the shift in sales and revenue.

Triggered: If the “Product Performance Differences” hypothesis has been retained AND the actual Revenue value is below the baseline AND the top contributing funnel step is Carts, Checkouts or Orders.

Retained: When product availability issues for the products detected in the “Products Performance Differences” exist.

Rejected: When product availability issues for the products detected in the “Products Performance Differences” hypothesis don’t exist.

  • Broken PDP Links

Examines if there are broken links on product pages that impact the UX and conversion rates. Specifically, Tywin examines the source code of the product page(s) and finds any broken links that may have created roadblocks in the conversion process.

Triggered: When the “Product performance differences” hypothesis has been retained AND the actual Revenue value is below the baseline AND the top contributing funnel step is Carts AND Product views change is positive.

Retained: When broken pdp links exist and may have impacted the UX or conversion rates.

Rejected: When broken pdp links don't exist OR broken pdp links exist but don't impact UX or conversion rate.

Did this answer your question?