What’s new.

When we built AI Visibility, daily was the obvious cadence — that's how every rank tracker on Earth works. After a year of watching customer SoV dashboards, we realized daily misses too much. Competitor launches, model retrains, news cycles — they all move recommendations on a faster cadence than 24 hours. So we built hourly.

How it works: mark up to 500 prompts per project as priority. Those queries now run every hour against every tracked engine (ChatGPT, Perplexity, Gemini, Claude). Everything else stays on the daily cadence. Available on Pro plans today; Enterprise gets unlimited hourly out of the box.

Most schema validators check whether your structured data is valid. Ours now also checks whether it's the kind LLMs actually cite — Article + Organization on every URL, accurate publishDate, clean canonical chain. The audit report has a new "AI eligibility" tab that ranks every URL by how cite-able it is.

• Keyword groups: saving a cluster with 200+ keywords no longer hits the 10-second timeout.
• Sitemap parser: now correctly handles `sitemapindex` files with relative URLs.
• CSV exports: numeric columns now use the locale you've set, not en-US by default.
• Mobile dashboard: the project switcher now respects safe-area insets on iOS.
• Slack integration: a race condition that occasionally double-posted alerts is gone.

We watched a half-dozen of our larger customers build the same warehouse pipeline: CSV export → SFTP → DBT model → Looker dashboard. Every team rebuilt it the same way. So we built it once, properly, and ship it as an integration.

Connect your BigQuery (or Looker Studio) workspace from Integrations. Pick what gets synced (everything, or just specific projects). Cadence is hourly for new data, daily for full rewrites. Schema is documented at /help/integrations/bigquery — copy-paste-ready into your dbt project.

We added the following triggers, configurable per project:

• AI SoV drop: alert when share-of-voice falls more than X points over Y days on a prompt.
• Lost backlink (high DR): alert when a backlink from a source above DR 60 disappears.
• Rank threshold crossing: alert when a tracked keyword crosses into or out of the top 10.
• New competitor SERP: alert when a domain not in your competitor list appears in the top 3 for a tracked keyword.

Previously the Rank Tracker history view showed a single trend line per keyword, averaging desktop and mobile positions. We retired that. The new view defaults to two separate lines — mobile and desktop — overlaid. You can toggle to combined, but most teams that switched are leaving it on split-view permanently.

• Audit retries: failed audits now retry once automatically with a 5-minute backoff before reporting failure.
• Backlinks export: anchor text containing commas no longer breaks CSV parsing downstream.
• Content Brief: brief generation occasionally returned an empty H2 outline when the seed keyword had under 10 results in the index — fixed.
• Workspace deletion: the confirm-delete flow now requires typing the workspace name (was just a click — too easy to misclick).

Enterprise customers asked for this for the last six months and we kept saying "soon." It's now soon. Public beta means the integrations work end-to-end but the management UI is still being polished. Setup needs a short conversation with our team to provision the IdP correctly.

Available on Enterprise plans only. If you're on Pro and want to test it, reach out — we're onboarding a handful of design partners ahead of GA in Q3.

Teams with multiple writers had to pick one when setting up custom voice, which made briefs read like that one person rather than the team. The new extractor accepts up to 5 distinct samples (one per author) and synthesizes the shared signals — sentence length, register, contraction frequency, hedging — into a unified house voice.

Mostly a query-shape problem — we were N+1ing across project metadata for every workspace member. Now batched. Boring under the hood. Noticeable in the UI.

We've been quietly running a private read API for two enterprise customers since last year. It's now in public beta, available to any Pro or Enterprise customer. Request access from your account dashboard. Stable endpoints, JSON responses, cursor-based pagination, and the same data freshness as the dashboard.

What's not here yet: write endpoints, webhooks, OAuth. Those are on the GA roadmap. Today the API is for pulling your data into your own systems — dashboards, custom alerts, internal tools.

• Hreflang validation: the audit now correctly handles `x-default` tags in mixed hreflang clusters.
• Keyword Explorer: filter combinations with 4+ active filters no longer hit the 5-second query timeout.
• AI Visibility: prompts containing curly quotes (`‘ ’`) now normalize before storage — caused a small dedup gap.
• Email digests: time zone of the digest send is now picked from your workspace setting, not your last login.