Quick Links

Introduction

How to set up a Negative Match Suggestions Strategy

How to edit the Negative Match Suggestions Strategy

How does Perpetua identify terms to negative match?

Introduction

One of the most effective ways to drive efficiency is to reduce wasted spend on irrelevant keywords that may not be bringing you any traffic. To accomplish this, advertisers may choose to negative match these keywords.

Identifying strong candidates for negative matching can take a lot of labour-intensive research where you are staring at spreadsheets for hours. Perpetua's Negative Match Suggestions is a feature where our AI will find keywords that will actually impact goal performance when negative matched, without the risk of wiping out good terms.

Our platform will present the terms you should consider negative PHRASE matching in the form of recommendations. If you click Recommendations on the left side menu you will see a Strategy called "Negative Match Suggestions". You can then pick the goals you want to be notified about by following the steps below to set up and edit a Strategy.

Once the ad engine identifies search terms that are seeing wasteful spend, it will suggest terms it thinks that you should negative match. There will be 3 places where you will get alerted.

The first place is under Recommendations tab. You can approve the suggestion by clicking the checkmark, or dismiss by clicking X. If you click approve, the term will be negative matched.

The second place you will see it is on the Sponsored Products page. Hover over a goal, and click the 3 dots to reveal Recommendations. Once you click Recommendations, a popover will appear, with each suggestion listed (if any).

Finally, if you click into a specific Sponsored Products goal, suggestions will populate under the Recommendations widget.

To apply negative matches manually on Perpetua, follow the steps here.
For Perpetua's advice on how to treat Negative Keywords, click here.

How to set up a Negative Match Suggestions Strategy

Negative Match Suggestions will be ON by default for all goals. If you want to have this on for a select group of goals, follow these steps.

  1. Go to the Recommendations tab on the left, then Strategies at the top.

  2. Under Negative Match Suggestions, click on Add Strategy.

  3. A pop up will appear. Enter in your Strategy Name, and click Select Goals to pick the goals you want to include.

  4. Click Create Strategy, and you are done! 🎉 Once the ad engine identifies search terms it thinks that you should negative match, you will get notifications. The recommendations will not be applied automatically, so you will have to approve ✅ or dismiss ❌ the suggestion for the negative phrase match to take effect. Once approved, the search term will be added as a negative phrase match target.


How to edit the Negative Match Suggestions Strategy

  1. Go to the Recommendations tab on the left, then Strategies at the top.

  2. Under Negative Match Suggestions, click the 3 dots on the strategy to edit or delete it. Proceed to step 3 if you are choosing to Edit.

  3. A pop up will appear where you can edit the strategy name and click Edit to change the goals included. After you click Edit, you can select or de-select goals.

  4. Once you have made your update, click Set Selected to confirm. Click Update Strategy to save your changes, and you are done! 🎉 Once the ad engine identifies search terms it thinks that you should negative match, you will get notifications. The recommendations will not be applied automatically, so you will have to approve ✅ or dismiss ❌ the suggestion for the negative phrase match to take effect. Once approved, the search term will be added as a negative phrase match target.

    🚨 Important Note: Please read the notification carefully before approval. It is important to proceed with caution because you may run the risk of wiping out relevant terms if the suggestion is not a good fit for your goal. The AI is smart, but sometimes a human is smarter! 😉 For example, if our system suggests that you negative match the word "shirts", but you sell t-shirts, it might be more sensible to reject the suggestion. Our algorithm makes recommendations based on statistical data from clicks and conversion throughout the history of your goal. It may not account for the fact that a keyword is still related to your product (and therefore relevant), despite its lack of conversions. If I accept the recommendation to negative phrase match the word t-shirt, I may block too many search terms which would result in my ad reaching fewer customers!


How does Perpetua identify terms to negative match?

Our ad engine looks at words that [in last 12 months] have had 0-1 conversion(s) in any search term they have been a part of. When surfacing potential negative matches, our systems only considers Sponsored Product goals that are 30+ days old.


What model do we use, and what data do we take into consideration?

Along with other statistical controls, a clicks threshold is determined based on the performance of the goal and a spend threshold is determined based on the target ACoS and spend-per-conversion for the goal. Once a term has generated too many clicks, and spent too much, with less than 1 conversion, a recommendation will appear on our app.

The clicks threshold is the minimum number of clicks that must be achieved by the word in order for the algorithm to recommend negative matching it from the goal. This threshold differs for every goal.

These thresholds are further used to limit the words recommended to be negative matched.

What else happens?

We convert all words to their root word. For example, words such as cared/caring/cares/care all become care. We remove stop words like is, to, are, etc.

Overall, even with the statistical controls in place, the model becomes more robust with more historical data. Recommendations will become better, and smarter, as a goal matures and has been active for a longer period of time.

For more details, check out our tech blog and blog post. Here we explain how Natural Language Processing, Tokenization and Statistical Modelling is used to isolate irrelevant terms that are seeing wasteful spend.


Article last updated February 2022. If you find this information to be out of date, please contact hello@perpetua.io.

Did this answer your question?