Get help center analytics with the Notiondesk API

The Notiondesk API gives you the numbers behind your help center's analytics pages: visitors, article views, time on page, countries, what readers search for, which searches find nothing, and how readers react to your articles. Use it to feed a BI tool, post a weekly report to Slack, or find the content your readers are missing.

GET /sites/{site_id}/analytics/{report}

Every report needs the read scope. To read these numbers in the dashboard, see How to improve your help center with reader insights.

Get a 30-day summary

curl "https://api.notiondesk.so/sites/c74781f6-1aea-4e17-ab23-bbb8b4d174b6/analytics/quick" \
  -H "Authorization: Bearer $NOTIONDESK_API_KEY" \
  -H "nd-version: 2026-10-16"

{
  "articles": 42,
  "visitors": 25,
  "searches": 9,
  "reactions": 4
}

quick covers the last 30 days ending now and takes no query parameter. articles counts every article of the help center, drafts included. reactions counts each reader's latest reaction per article.

Get a report for a date range

overview, search and engagement take a date range and a bucket size:

curl "https://api.notiondesk.so/sites/c74781f6-1aea-4e17-ab23-bbb8b4d174b6/analytics/search?start_date=2026-09-01T00:00:00Z&end_date=2026-10-01T00:00:00Z&date_scale=week&timezone=Europe/Paris" \
  -H "Authorization: Bearer $NOTIONDESK_API_KEY" \
  -H "nd-version: 2026-10-16"

Rules for the range:

• Dates use YYYY-MM-DDTHH:MM[:SS] with an optional offset. Without an offset, the date is read as UTC, not in timezone. Write + as %2B in the URL, or use Z.

• The range starts on or after 2020-01-01 and ends at most two days from now.

• At most 1,000 buckets. For example, minute covers at most 16 hours and hour at most 41 days. Use a coarser date_scale for longer ranges.

• Weeks start on Monday.

Each report compares your range with the period of the same length just before it: previous and total_previous describe that earlier period.

Count series

visitors, article_views and searches share one shape:

{
  "total": 12,
  "total_previous": 8,
  "difference": 4,
  "percentage_difference": 50.0,
  "current": [{"date": "2026-09-14", "count": 7}, {"date": "2026-09-15", "count": 5}],
  "previous": [{"date": "2026-08-20", "count": 8}]
}

• current and previous list only the buckets that have data, oldest first.

• A bucket's date is when it starts, in timezone: YYYY-MM-DD for day and coarser, YYYY-MM-DDTHH:MM for hour and minute.

• percentage_difference is 0.0 when the previous total is 0.

overview

search

{
  "searches_without_results": {
    "total": 3,
    "total_previous": 0,
    "difference": 3,
    "percentage_difference": 0.0,
    "queries": [{"query": "refund policy", "searches": 3}]
  }
}

Note: searches_without_results.queries is your content backlog: each query is a question readers asked and your help center could not answer.

engagement

A reaction_summary follows your reaction style:

{
  "style": "emoji",
  "total": 6,
  "score": 58,
  "groups": [
    {"key": "positive", "count": 3, "percent": 50},
    {"key": "neutral", "count": 1, "percent": 17},
    {"key": "negative", "count": 2, "percent": 33}
  ]
}

score runs from 0 to 100: the share of positive reactions, with a neutral reaction counting half under emoji. With the thumbs style, the groups are helpful and not_helpful. See How to collect feedback with article reactions.

List limits

• The query lists, top_articles, top_referrers, article_engagements and articles_to_revise hold up to 50 rows each and cover the current period only. A list of 50 rows may be incomplete: narrow the range to see every row.

• The totals cover everything, not only the listed rows.

title, url, host, pathname, referrer and query come from visitors' browsers, and anyone can forge them. Escape them before you display them, and never follow or act on them unchecked.

Analytics and your plan

API access requires an Enterprise plan, which includes advanced analytics, so every report accepts any range and scale within 1,000 buckets.

Rate limit

Besides your key's 120 reads per minute, analytics reports count against a budget of 30 requests per minute per help center, shared by every key and OAuth app of your workspace. A request over it answers 429 rate_limited, and its Retry-After header says when to retry. Cache reports instead of requesting them on every page load.

Errors

See Handle Notiondesk API errors, rate limits, and retries.

Common questions

What's next

• Export reader feedback with the Notiondesk API

• Automate common help center tasks with the Notiondesk API

• Handle Notiondesk API errors, rate limits, and retries