Knowing how often AI mentions your business is useful. But what should you do when competitors keep getting recommended instead, or AI rarely cites your website?

"Recommendations" brings together practical suggestions based on the AI answers Klarvu has already collected. You can use them to identify areas worth investigating and keep track of the improvements you're working on.

Find your recommendations

Go to "Improve" → "Recommendations" in the menu.

At the top, you'll see the description "Actionable steps to improve your brand visibility across AI search engines."

Recommendations apply to the brand you've selected in Klarvu.

They're automatically generated from your collected AI answers and the analysis of those answers. As Klarvu analyzes new responses, the recommendations are updated.

This means new suggestions can appear when Klarvu detects something relevant in your results.

Keep your recommendations organized

The page has four tabs to help you manage your work.

Tab	What you'll find
"Suggested"	New recommendations you haven't reviewed yet.
"To-Do"	Recommendations you've decided to work on.
"Done"	Recommendations you've completed or that Klarvu has automatically marked as resolved based on new data.
"Dismissed"	Recommendations you've chosen to hide. You can restore them at any time.

You can move recommendations through your workflow using the available actions:

Under "Suggested", choose "Add to to-do" or "Dismiss".
Under "To-Do", choose "Mark as done" or "Dismiss".
Under "Done" and "Dismissed", use "Restore to suggested" to move it back.

These actions require editing permissions.

You don't need to act on every suggestion. It's fine to dismiss recommendations that aren't relevant to your business.

Understand the table and "Impact"

The recommendations table includes the following columns:

Column	What it means
"Impact"	Klarvu's assessment of the recommendation's potential importance, scored from 0 to 100.
"Recommendation"	A short description of the issue or opportunity Klarvu identified.
"Provider"	The AI platform the recommendation relates to.
"Category"	The area the recommendation concerns.
"Country"	The market the recommendation applies to. This column only appears when you're tracking multiple countries.
"Type"	Whether the recommendation is general, relates to your own website, or concerns other websites.

Under "Done", the "Type" column is replaced with "Since then", which shows what happened after the recommendation was resolved.

How does the Impact score work?

"Impact" helps you decide which recommendations to investigate first.

Level	Score
High	70–100
Medium	40–69
Low	Below 40

Scores can be compared across different recommendation types.

A high score means Klarvu considers the recommendation a higher priority. It doesn't guarantee that taking action will improve your mention rate or ranking.

You can sort recommendations using:

"Impact: High to Low" – highest scores first.
"Impact: Low to High" – lowest scores first.
"Most recent" – newest recommendations first.

Use "Search recommendations..." to find specific suggestions, or "Filter" to narrow the list by "Country", "Provider", or "Category".

Categories and types

Klarvu organizes recommendations into eight categories:

"Reputation", "Entity", "Crawlability", "Prompts", "Content", "Content Partnership", "News & Media", and "Social Media".

Each recommendation also has a type:

"General": A general recommendation that isn't limited to one particular website.
"On-page": Something to investigate or improve on your own website.
"Off-page": Something involving external websites or third-party coverage.

What kinds of recommendations can you get?

Klarvu creates different recommendations depending on what appears in your AI answers.

Here are the types you might encounter:

Recommendation	What it means	What you can do
"0 brand mentions detected on [platform] (in [country])"	Your brand hasn't appeared in any analyzed answers from the platform. At least three answers are required.	Review the questions and investigate which competitors AI mentions instead.
"0 website citations on [platform] …"	None of the platform's answers cite your website. At least three answers are required.	Explore which sources AI uses and whether your own content answers the relevant questions.
"[Platform] describes [brand] negatively on: "[prompt]""	At least two recent answers to the same prompt describe your brand negatively or with mixed sentiment.	Read the answers and check the specific claims about your business.
"[Platform] is steering buyers away from [brand] on: …"	An answer explicitly advises against your brand. One such answer is enough to trigger the recommendation.	Investigate the reasoning and any sources behind it.
"Get [brand] mentioned/featured in this article on [source]"	AI cites a page in at least two of your answers, but the page doesn't mention your brand.	Consider whether your company belongs on the page and whether contacting the publisher makes sense.
New content suggestion	Klarvu identifies a topic where AI cites other sources without mentioning you and suggests a new page.	Review the AI-generated suggestion and check whether your website is missing useful content.
One of your pages isn't being cited	A page on your website doesn't appear as a source in collected answers. This requires a completed crawl under "Site maps".	Investigate the page's accessibility and content.
"N% of your prompts are in [stage]"	At least 70% of your active prompts belong to the same buying journey stage. At least five active prompts are required.	Consider whether you're missing questions from other stages.

Recommendations about appearing in external articles can fall under "News & Media", "Social Media", or "Content Partnership". New content suggestions belong to "Content" and have the type "On-page".

For more on these topics, see Sentiment, "Sources", "Explore a prompt's results", and "Topics and prompts". The Crawlability Checker and Content Checker pages cover the technical tools in more detail.

Open a recommendation to see the details

Click any recommendation in the table to open "Recommendation Details".

Here you'll find:

Title and Impact: What the recommendation concerns and its priority.
Platform, category, and type: Where it came from and what kind of work it involves.
"Reasoning": Why Klarvu created the recommendation.
"Suggestions": Practical ideas for what to investigate or improve.
Supporting evidence: Links to relevant prompts and the AI answers behind the recommendation.

If the recommendation relates to a specific question, you can use "Open this prompt's detail page" to investigate it further.

You'll also see the reminder:

"Recommendations may not always be accurate. Please double-check suggestions."

It's worth keeping that in mind. Understand what AI actually said before deciding whether the recommendation makes sense for your business.

How to work with recommendations

You can use the tabs as a simple workflow, moving from a new suggestion to a completed task.

Start with your highest-priority suggestions

Open "Suggested" and sort by "Impact: High to Low".

Look at the recommendations with the highest scores, but focus on those that are also relevant to your business and responsibilities.

Understand why the recommendation was created

Open "Recommendation Details" and read "Reasoning".

Then explore the actual AI answers. Is the information accurate? Are competitors being recognized for something you also offer? Is an external source describing your product incorrectly?

Add it to your to-do list

If you decide to work on the recommendation, click "Add to to-do".

You'll find it under "To-Do", where you can keep track of the improvements you've chosen to prioritize.

Complete the work and mark it as done

This might involve clarifying content on your website, investigating a source, or contacting a relevant publisher.

Once you've finished, click "Mark as done" so it moves to "Done".

Follow what happens next

Return to "Done" and check whether the relevant metrics change afterward.

Remember that changes in AI answers aren't necessarily caused by the work you've done. Treat them as developments worth investigating rather than proof of impact.

What does "Since then" mean?

Under "Done", you can see how a relevant metric has changed since a recommendation was resolved.

"Since then" compares the 30 days before resolution with up to 30 days afterward.

For example, it might show a change in mention rate:

"Mention Rate: 20 → 35"

During the first seven days after resolution, Klarvu displays "Result in Nd", indicating that more time is needed before a result is shown.

If there isn't enough data for a meaningful comparison, you'll see "Too little data to judge".

Importantly, "Since then" shows a change over time, not a causal claim.

An improvement may be related to the action you took, but AI answers can change for other reasons too.

Recommendations can resolve automatically

You don't always need to mark a recommendation as done yourself.

If newly analyzed answers show that the original issue no longer appears in the data, Klarvu can automatically move the recommendation to "Done".

If the issue appears again later, the recommendation can return with a "Returned" label.

This helps you keep track of problems that disappear and reappear without having to create a new task each time.

Frequently asked questions

Why am I not getting new recommendations?

Recommendations are based on your collected and analyzed AI answers. If Klarvu hasn't identified a relevant issue or opportunity yet, there may be nothing new to suggest.

Some recommendations also require a minimum number of answers before they can be generated.

What does "Returned" mean?

It means a recommendation was previously resolved, but Klarvu has detected the same issue again.

Open the recommendation and review the latest answers to understand what changed.

Why does it say "Too little data to judge"?

It means Klarvu doesn't yet have enough data to assess what happened after the recommendation was resolved.

It doesn't necessarily mean the work had no effect.

What if I don't want to work on a recommendation?

Choose "Dismiss" to hide it.

The recommendation isn't permanently deleted. You can find it under "Dismissed" and use "Restore to suggested" if you want to revisit it.

Why is my "Done" tab empty?

If you don't have any resolved recommendations yet, you'll see "Wins land here — items resolve automatically when the data improves, or when you mark a to-do done."

Recommendations are a starting point, not a definitive set of answers. Use them to identify relevant issues, understand the underlying AI responses, and prioritize the work that makes sense for your business.