Recommendations

The Recommendations page is a priority queue of content to publish where competitors appear in AI answers and you're missing. Each recommendation groups related prompts, says why it matters and what to publish, and lists the prompts that page should answer. It's available on every plan — you don't need Pro to use it.

What you see

1. Four headline numbers

• Open — how many recommendations are open, with how many of them are critical, high and medium.
• Prompts affected — how many prompts those open recommendations cover, and across how many topic clusters.
• Mention rate in these clusters — how often your brand is named in the reviewed answers on those prompts, and how many answers that is.
• Shipped · 30d — recommendations marked done in the last 30 days. Each teammate's mark counts, so one that two people marked done counts twice. It counts the work you've recorded; it doesn't measure what changed afterwards.

Under the tiles, a bar shows the answers each open recommendation covers, one segment per recommendation, coloured by severity. Hover over a segment for its name. These numbers always describe the whole open queue: the queue's own filters don't change them.

2. The queue

Tabs across the top hold your recommendations by status: Open, Snoozed, Done and Dismissed, each with a count. Beside them you can filter by Severity and Intent, and choose the Sort: Priority (the default), Severity, Prompts or Lowest mention rate.

Each row shows:

• # — the recommendation's place in priority order within its tab: P1 is the highest. Filtering or re-sorting doesn't change these numbers.
• Recommendation — its name (the intent and category of its prompts, plus any shared tags), its severity, the kind of page to publish and how many prompts it covers.
• You're named — how often your brand is named in its answers, on a short bar.
• Answers — how many reviewed answers it rests on.
• Winning instead — the competitors that appear across the most of its prompts.
• Mark done — on the Open tab. On the other tabs the row shows its status instead, with an Undo button.

The kind of page depends on the prompts' intent:

3. An opened recommendation

Click a row to open it. You'll see:

• Why — who is winning these answers, and how often you're named in them.
• What to publish — the page to create and up to three example prompts it should answer, starting with the ones where you're named least.
• Snooze 7 days and Dismiss (on the Open tab).
• Target prompts — every prompt in the recommendation, least-named first, with how many of its reviewed answers name you and the competitors seen. Each prompt links to the Scans page.

4. Acting on several at once

Tick the box at the start of open rows to select them. A bar appears at the bottom of the screen with Mark done, Snooze 7d, Dismiss and Clear.

5. The Actions view

The switch beside the page title moves between Recommendations and Actions. Actions lists the work your team has taken on, who has it, and what changed after it was done. Actions are started with Start action on the Citations and Competitors pages.

How recommendations are built

SimplyRank reviews a sample of your most recent answers from the last 30 days, from visibility prompts and sentiment probes. The coverage note beside the page title gives the sample's dates and limits, and says if the limit was reached.

A prompt goes into the queue when:

• it has at least two reviewed answers,
• your brand is named in fewer than one in five of them, and
• a tracked competitor appears in its answers.

Prompts are then grouped by intent, category and shared tags. A group needs at least three prompts to become a recommendation. Only active prompts are grouped, so paused and archived prompts are left out.

Priority is a score from 0 to 1. It multiplies:

• the share of the group's prompts where you're named in fewer than one in five answers,
• the share of its prompts where its most frequent competitor appears,
• the number of prompts, on a log scale (ten or more count in full), and
• a weight for intent: transactional highest, then commercial, navigational and informational.

A score of at least a half is critical, at least a fifth is high, and anything lower is medium. Priority uses no search volume or revenue data.

What good action looks like

For each recommendation, the goal is one concrete next step. Examples:

You don't have to act on every recommendation. Pick 1–3 a week. The compounding effect over a quarter is more valuable than trying to fix everything at once.

Done, snoozed and dismissed

• Mark done when you've shipped the work. It moves to the Done tab and counts towards Shipped · 30d.
• Snooze hides a recommendation for seven days. Once the snooze has ended, it's back in Open the next time the page loads.
• Dismiss when a recommendation doesn't fit your business. It moves to the Dismissed tab.

Undo removes your own mark. A recommendation goes back to Open once nobody's mark is left on it. Everyone in the workspace works from the same queue, so marking a recommendation changes the list for the whole team.

A status belongs to the group (its intent, category and tags), so a recommendation you've marked done stays done when the prompts inside it change.

How often recommendations refresh

The queue is rebuilt from your latest answers after each scan finishes, so it moves at the pace of your scans: daily on plans that scan daily, weekly on weekly plans. A prompt you pause or archive drops out when the queue is next rebuilt, after the next scan or within the hour.

Export

Export downloads the recommendations in the tab you're on that match your filters, as CSV. CSV export starts on the Starter plan. See Exports and API.

Plan availability

The Recommendations page is available on all plans, including the trial. The depth of recommendations scales with the data available — more prompts and more scan history produce richer suggestions.

What's next

• The visibility playbook
• Gap analysis
• Interpreting your results