Machine Learning Automated Quote Suppression FAQs

1) What changes are being made by Skyscanner to minimise price inaccuracies?

We are introducing a Machine Learning model to predict inaccurate quotes before they are presented to our users. This feature will only be applied to partners who have price accuracy event counts over 1000 and price inaccuracies greater than 10% in the past week.

2. How will filtering with the Machine Learning model work?

Our model works by predicting the likelihood of inaccuracy based on a variety of factors and using historical data from the past week. When users search in the week following, content will be processed using the model’s learnings to determine level of accuracy and a decision will then be made on whether a partner’s price is suppressed from the partner selection screen. The list of partners which inaccuracies of 10% or more will automatically refresh each week. Partners should work to improve price accuracy to below this threshold to avoid suppression. 

3. Is my price accuracy affected by Skyscanner caching?

We have a max quote age setting of 10 mins on our booking panel, which means that the oldest price we can show is 10 mins old at the point of redirecting. If you think we should cache (some of) your content even shorter, please reach out to our commercial managers. 

4. How can partners pro-actively prevent having quotes suppressed?

We are now asking for price inaccuracy no higher than 5% (of all your redirects) with a [-4% to +1% or -25 EUR to +7 EUR (whichever comes first)] price change threshold. Our commercial managers will reach out if your price inaccuracy is identified problematic. You will be asked to reduce your inaccuracies below 5%, with help from our price accuracy report. However, if your price accuracy is not improving and detected higher than 10% in a week, machine-learning suppression will be applied on your content (in that given market if you are a global OTA) in the following week. 

Our team appreciates any improvements partners can make, and we will take appropriate steps not to suppress content unless we detect serious issues. Please reach our to your commercial manager if you would like more data from the portal report.

5. How long will my quotes be suppressed for under Machine-learning?

We will automatically renew a list of problematic partners (where inaccuracies are higher than 10% in the past week as mentioned above) every week. If you can improve your price inaccuracy below 10% then you will be removed from the list automatically, otherwise we will keep you on the list in the following week. 

6. What impact can you foresee on me?

Given the high rate of inaccuracies (>10%), we estimate a loss in redirects by up to 5%.

7. WilI Skyscanner still be calling my API when doing suppression?

Yes, we will still be calling your API and then decide whether or not we should suppress the quotes. Your live price is also one of the model’s considerations to make the decision.

8. Where can I find out which quotes have been suppressed?

From the Missing Quote Finder in our partner portal, you can find the filter result with the reason "Quote Suppression Filter".