Using AI Data Stream

Reports & Skills

Run ready-made analysis skills or build your own to get consistent, repeatable reports from your data.

Skills are reusable analysis recipes with expert methodology built in. Running a skill produces a report - structured, formatted output you can share with your team or schedule to arrive by email.

Skills and Reports

Two words to know:

  • A skill defines an analysis: what data to look at, what methodology to follow, and how to present the results. Skills come from the built-in catalog or you can create your own.
  • A report is what you get when a skill runs - a finished, formatted analysis of your property’s data.

Unlike free-form conversations, a skill runs the same way every time. You select it, optionally fill in fields (like a URL or date range), and get consistent output.

Skills are useful for:

  • Regular check-ins (weekly traffic reviews)
  • Consistent analysis formats
  • Sharing with stakeholders
  • Onboarding team members to standard analyses

Running a Report

  1. Click Skills in the sidebar
  2. Click Run Report on a skill
  3. Fill in any fields
  4. Click Generate Report

Team owners can add more skills with Add skill → Browse the skills catalog on the same page.

The AI will analyze your data following the skill’s methodology and return formatted results.

Scheduling Reports

On paid plans, any skill can run automatically on a recurring schedule, with the finished report emailed to your team.

Creating a Schedule

  1. Click Skills in the sidebar, click Scheduled Reports, then New schedule (or click Schedule this report on a skill’s page, which opens the form directly)
  2. Choose the skill and give the schedule a name
  3. Pick a frequency:
    • Weekly - runs on a chosen day of the week
    • Monthly - runs on a chosen day of the month (shorter months use their last day)
    • Every N days - runs at a fixed interval
  4. Set the time and timezone
  5. Fill in the skill’s fields - these values are reused on every run
  6. Choose which team members receive the email (only people with access to the property; if nobody is selected, just you)

What Happens on Each Run

  • The report runs exactly like a manual run and appears in your reports history, shared with the team
  • Recipients get an email in their own language with a link to the finished report
  • Scheduled runs show a Scheduled badge in the recent reports list, and skill cards display their cadence

Managing Schedules

The Scheduled Reports page lists every schedule with its next and last run, a run history, and an on/off toggle. Editing a schedule takes effect from the next run.

If a schedule fails - for example the skill was deleted, a required connection was disconnected, or no AI provider is active - the creator gets an email explaining why. After three consecutive failures the schedule pauses itself; fix the issue and switch it back on to resume.

Plan Limits

Scheduling is included on Pro and Business, with a cap on active schedules that depends on your plan. The Free plan doesn’t include scheduling - every skill can still be run manually at any time.

Organizing Your Skills

As you add skills, you can organize them to find them quickly.

Ordering Skills

Use the Order by dropdown on the Skills page to choose how skills are displayed:

Last Used

  • Skills are sorted by when they were last run by anyone on your team
  • Most recently used skills appear first
  • Great for quick access to frequent analyses

Manual

  • Create your own ordering by dragging skills
  • Drag the grip handle (⋮⋮) to reorder
  • Your custom order is saved to your account and persists across sessions
  • Each team member can have their own custom order

Managing Skills

Team owners manage skills on the same page they run them from. The ⋮ menu on a skill card has Edit, Delete and, when the catalog has a newer version, Update to latest.

Skills set to inactive are hidden from Available Skills. Team owners still see them under Inactive skills at the bottom of the page, where Edit lets them set a skill back to active.

Deleting a skill also removes its schedules. Its past reports stay in Recent Reports, marked Skill deleted, so you can still open them. To remove them as well, tick Also delete its past reports in the delete dialog. That deletes them for everyone on the team.

The Skills Catalog

The catalog includes more than 30 ready-made skills organized by category. Team owners open it with Add skill → Browse the skills catalog on the Skills page. Add one to your property and it’s ready to run or schedule.

Each skill lists the connections it needs (for example Google Ads or Search Console). Skills that need a connection you haven’t set up yet are marked in the catalog, and they can only run while their required connections are active. A connection marked (optional) is used for a cross-check when your property has it; the skill still runs without it.

Catalog updates

Adding a skill makes a copy on your property, so you can edit it freely. When we improve a catalog skill, copies you haven’t edited show a Catalog updated badge on the Skills page, team owners get a notification, and Update to latest (or Update all in the banner) replaces the copy’s prompt, fields, connections and icon with the new version. Your skill name, description, status and schedules stay as they are. A copy you’ve edited is never updated automatically; add a fresh copy from the catalog if you want the new version alongside your own.

Audits

In-depth teardowns that score what’s working and rank what to fix.

  • Google Ads Account Audit - Full account teardown: wasted spend, conversion tracking sanity, budget pacing, and quality score, scored out of 100
  • Search Campaign Deep Audit - Keyword-level teardown of Search campaigns: quality score components, match-type mix, search-term waste, and ad testing gaps
  • SEO Audit - Search Console teardown of queries, pages, and indexing, enriched with Core Web Vitals and backlink data when connected
  • Core Web Vitals Audit - LCP, INP, and CLS against Google’s thresholds: real-user field data first, with fixes ranked by estimated savings
  • AI Search Visibility - Where the site is cited in Google AI Overviews and ChatGPT answers today, which pages earn it, and how competitors compare: an honest baseline before anyone pays for GEO or AEO (DataForSEO)
  • Landing Page Audit - Single-page teardown: traffic, engagement, search queries, and page experience for one URL
  • GTM Container Audit - Tag Manager container teardown: duplicate GA4 tags, orphaned tags, dead triggers, Consent Mode wiring, and draft drift, scored out of 100
  • LinkedIn Ads Account Audit - Campaign structure, spend efficiency, creative testing gaps, and professional-audience waste, scored out of 100
  • Meta Ads Account Audit - Campaign efficiency, creative fatigue, placement waste, and the Facebook/Instagram/Audience Network split, scored out of 100
  • Meta Tracking Health Check - Diagnose the gap between Meta pixel conversions and GA4 reality: modeled conversions, view-through inflation, blended lead actions, and UTM hygiene

Monitors

Rhythm checks designed to run on a schedule and flag problems early.

  • Weekly Traffic Monitor - Weekly GA4 rhythm check: traffic and conversions week-over-week, with tracking-break alerts
  • Monthly Performance Review - Last full month vs the month before: traffic, channels, top pages, and search, explained with annotations
  • Quick Health Check - Two-minute status: is anything broken or sliding? Traffic, engagement, and search, 7 days vs 7 days
  • SEO Performance Monitor - Search Console rhythm check: clicks, positions, rising and falling queries, and quick wins, 28 days vs the prior 28
  • Google Ads Weekly Monitor - Weekly guardrail: spend spikes, conversion-tracking breaks, and dead campaigns, week over week
  • GTM Change Monitor - Weekly Tag Manager check: what changed in the container since last run, plus standing health checks and unpublished-draft alerts
  • AI Crawler Visibility - Which AI systems are reading the site, what they read, wasted crawls on dead URLs, bandwidth cost, and whether AI assistants send visitors back (Cloudflare)
  • Sitemap Content Changes - Which pages appeared in and disappeared from the sitemap, and how much search traffic the removed pages were earning (Sitemap)
  • Brand Mentions Digest - Where the brand was mentioned across the web in the period (press, blogs and forums, with republished copies merged), the trend behind it, and the visits it sent (DataForSEO)

Analysis

Focused deep dives that take input fields.

  • Custom Date Comparison - Compare any two periods across traffic, channels, pages, and search, with annotation context
  • Content Section Analysis - How a content section performs: top and bottom pages, trend, and the queries that drive it
  • Traffic Source Deep Dive - One source under the microscope: volume, quality, landing pages, and UTM hygiene
  • Keyword Topic Analysis - Every query matching a topic: performance, position opportunities, and the pages that rank for them
  • Backlinks vs Search Performance - Joins your real backlinks to Search Console pages: which pages to earn links for, which have links but underperform, and whether lost links cost traffic (DataForSEO and Search Console)
  • Conversion Tracking Debugger - Trace a conversion event end-to-end: the GTM tag that should fire it, its trigger conditions, and whether GA4 actually records it
  • LinkedIn B2B Audience Insights - Who actually sees, clicks, and converts on your LinkedIn ads: job functions, seniorities, industries, and company sizes, with spend-vs-results mismatches
  • LinkedIn Lead Attribution Check - Reconcile LinkedIn-reported conversions and leads against GA4 reality: native lead forms vs website conversions, view-through inflation, and UTM hygiene
  • Meta Lead Attribution Check - Decompose the blended Meta “lead” number into native forms vs pixel leads, verify pixel leads against GA4, and follow them into the CRM for cost per verified lead
  • Meta vs Google Ads Channel Mix - Spend, efficiency, and verified results across Meta and Google Ads with GA4 as the arbiter: where the next unit of budget should go
  • Edge vs Analytics Gap - How much traffic GA4 never sees: Cloudflare edge page views against GA4 sessions, split into AI crawlers, other bots, and untracked humans

CRM

For properties connected to HubSpot or Pipedrive.

  • Ideal Customer Profile Analysis (HubSpot) - Data-driven ICP from your won deals and customer base: firmographics, personas, deal patterns, and an anti-ICP
  • Pipeline Momentum Analysis - Is the pipeline moving? Velocity, stage distribution, stalled deals, and win patterns

Ecommerce

For sites with ecommerce tracking enabled.

  • Ecommerce Weekly Review - Revenue, orders, conversion, and product movers, week over week, with alert thresholds
  • Purchase Funnel Health Check - View → cart → checkout → purchase drop-offs against benchmarks, split by device
  • Promotion Performance Analysis - Did the promotion actually work? Revenue lift vs baseline, AOV effects, and what it pulled forward

Creating Custom Skills

Team owners can create their own skills tailored to your specific needs.

Generate with AI

The easiest way to create a skill is to let AI do it for you:

  1. Click Skills in the sidebar, then Add skill → Blank Skill
  2. Click Generate with AI
  3. Describe the report you want in plain English
  4. AI generates the skill name, description, input fields, and prompt
  5. Review and adjust if needed
  6. Save it to your property

Example: Type “Track our top performing blog posts by traffic and engagement, compared to last month” and AI will create a complete skill with the right fields and analysis prompt.

Manual Skill Creation

You can also create skills manually:

  1. Click Skills in the sidebar, then Add skill → Blank Skill
  2. Enter a name and description
  3. Define input fields (optional):
    • Text - For short inputs like topics, keywords, or names
    • Textarea - For longer multi-line inputs like descriptions or context (up to 10,000 characters)
    • URL - For specific pages to analyze
    • Date range - For time period selection
  4. Write the prompt using {{field_name}} placeholders
  5. Choose an icon and color
  6. Save it to your property

Prompt Syntax

Use placeholders for input fields:

Analyze the landing page at {{page_url}}.

Include:
- Traffic for {{date_range}}
- Top traffic sources
- Engagement metrics
- Recommendations for improvement

Example Custom Skill

Name: Competitor Comparison Note

Fields:

  • competitor_event (text)
  • our_url (url)
  • date_range (date range)

Prompt:

Our competitor announced: {{competitor_event}}

Analyze our page at {{our_url}} for {{date_range}}:

1. Current performance metrics
2. Search visibility for related keywords
3. Opportunities to capitalize on competitor news
4. Recommended actions

Sharing Reports

Team Access

Skills are available to all team members with write access to the property.

Sharing Results

After running a report, the report page lets you:

  • Copy the results to paste into documents
  • Share the report with your team (the report’s owner only)
  • Print / Save as PDF (on the Business plan, printed reports leave out the “Generated with AIdata” footer)
  • Continue in Chat to ask follow-up questions

Tips for Using Skills

When to Use Skills vs Chat

Use skills for:

  • Regular, repeatable analyses
  • Standardized team outputs
  • Stakeholder communication
  • Onboarding and training

Use chat for:

  • Ad-hoc questions
  • Exploratory analysis
  • Follow-up investigation
  • Quick checks

Building Your Skill Set

Start with catalog skills, then create custom skills for:

  • Your specific KPIs
  • Stakeholder-specific views
  • Recurring analysis needs
  • Client deliverables (agencies)

Combining with Annotations

Reports automatically check annotations. Keep annotations updated to ensure reports have context for any anomalies.

Troubleshooting

Report takes too long

  • Try a shorter date range
  • Use a faster AI model
  • Break the analysis into smaller, focused skills

Missing data in report

  • Check that required connections are active
  • Verify the date range has data
  • Check field values are correct (valid URLs, etc.)

Report output is inconsistent

  • Use a more capable model for complex analyses
  • Refine the skill’s prompt
  • Add more specific instructions

Next Steps