> ## Documentation Index
> Fetch the complete documentation index at: https://docs.vortexiq.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Browse and compare archived Vortex Mind report runs
> Every Vortex Mind report run auto-archives with full evidence and findings. Browse past runs, compare two runs over time, and export for stakeholders.
Every time Vortex Mind runs a diagnostic, the full report - cover, evidence, findings, recommendations, and structured output JSON - auto-archives into Vortex Memory. You do not trigger this; the writes happen silently as each run completes. Over weeks the archive becomes a longitudinal record of your diagnostic history. Over months it captures recurring patterns. Over a year it captures seasonal cycles. This page covers what each archived run contains, how to browse and search past runs, how to compare two runs over time, and how to export.
## What each archived run contains
Each entry in the archive is a complete, reproducible snapshot of the run at the moment it completed:
* The merchant-facing report (cover, diagnostic body, recommendations)
* The structured output JSON with findings, evidence, and recommendations
* Trigger metadata (scheduled, event, conversational, or on-demand) with timestamps
* The connectors and profile that were active at run time
* The exact data values read from each connector
* The recipe template used (default or custom)
* The list of citations the run's findings used as evidence
A run archived a year ago shows exactly what the report said at the time, with the same evidence and the same numbers. Nothing is retroactively changed.
## Which reports archive automatically
All seven Vortex Mind report types archive on every run, plus the four Actions briefings:
| Report | Default cadence |
| ---------------------------------------------------------- | ---------------------- |
| Payment Performance Intelligence | Weekly |
| Decline Recovery Intelligence | Weekly |
| Daily Revenue Leakage | Daily |
| Customer Recovery Opportunity | Weekly |
| Checkout Conversion Failure | Weekly |
| Google Ads Revenue Intelligence | Weekly |
| Paid Traffic Waste | Monthly |
| Daily Morning Briefing, Weekly, Monthly, and QBR briefings | Per configured cadence |
Each run's archive entry is tagged with its report type, profile, recipe template, trigger type, and the severity tier of the most-severe finding in that run.
## How to browse past runs
Navigate to Vortex Memory in the platform sidebar and open the **Reports** tab. The default view shows all report types, all profiles you have access to, sorted by run date with the newest first, covering the last 30 days.
Use the filter controls to narrow the list:
* **Report type** - filter to one or more recipe types (for example, Payment Performance only).
* **Date range** - presets: last 7 days, last 30 days, last 90 days, last 12 months. Or set a custom range.
* **Profile** - if your workspace uses multiple profiles to slice data, filter to a specific profile.
* **Severity** - show only runs with findings at or above a chosen severity tier (critical, high, medium, low, info).
A **calendar view** toggle switches from the list to a month-grid layout where each day with runs shows a colour-coded mark. Calendar view makes it easy to spot cadence breaks (a missing scheduled run) and clusters of high-severity days.
Each row in the list shows the report type icon, run date, profile, recipe template, severity badge of the most-severe finding, finding count, and trigger type. Click any row to open the full archived run.
### Deep-linking from a current run
Every Vortex Mind report has a **View past runs** link in its cover section. Clicking it opens the archive filtered to the same report type and profile, covering the last 90 days - the fastest path from a current run to its historical context.
## How to search across runs
The search box at the top of the Reports tab searches every layer of every archived run: finding IDs, finding titles, finding bodies, evidence content, recommendations, cited entities (SKU IDs, BIN ranges, decline codes, customer IDs, campaign IDs), and run metadata.
The most powerful use is finding every run that ever flagged a specific entity:
| Search goal | Example query |
| ---------------------------------------- | ----------------------------- |
| Every run that flagged a specific SKU | `SKU-4421` |
| Every run with a specific finding | `finding:PAY-AUTH-001` |
| Every run citing a BIN range | `"414720-414729"` |
| Every run mentioning a decline code | `"do not honor"` |
| Every run with a specific recommendation | `"renegotiate with acquirer"` |
Boolean operators work the same way as in the Knowledge Base search: `AND`, `OR`, `NOT`, `"exact phrase"`, `term*`. You can also use search shortcuts directly in the query string:
| Shortcut | Effect |
| ---------------------------- | --------------------------------------- |
| `report:payment-performance` | Filter to Payment Performance runs |
| `severity:critical` | Runs with at least one critical finding |
| `finding:PAY-AUTH-001` | Runs containing the specific finding ID |
| `trigger:event` | Event-triggered runs only |
| `before:01 Apr 26` | Runs before a date |
| `after:01 Apr 26` | Runs after a date |
| `profile:marketing` | Runs scoped to a specific profile |
Clicking a search result opens the archived run scrolled to the matched passage, with the matched terms highlighted.
## How to compare two report runs
Use the archive to build a business health timeline. Pin the run that first flagged a significant issue, compare it to runs from before and after a fix, and you have a clear record of what changed and when. Over time this becomes an audit-ready history of the decisions your business made in response to what the platform surfaced.
The compare surface shows two runs of the same report side-by-side with deltas highlighted. New findings, persistent findings, and cleared findings are visually separated. Numerical changes (auth rate moving from 87.4% to 89.1%) are shown as deltas with directional colour cues.
Navigate to Vortex Memory and open the **Reports** tab. Use the report-type filter to narrow the list to the recipe you want to compare (for example, Payment Performance Intelligence).
Check the checkbox on the first run, then check the checkbox on the second run. The order does not need to be chronological - you can flip A and B in the compare view.
Alternatively, open one archived run and click **Compare with another run** in the run header. A date-picker dialog opens defaulting to the previous run of the same recipe.
Click the **Compare selected** button that appears at the top of the list once two runs are checked. The compare view loads with both runs.
The compare view is structured into three sections:
* **Cover comparison** - headline KPIs side-by-side with numeric deltas, and executive summary text with line-level diffs (added text in green, removed text in red).
* **Findings comparison** - findings sorted into three buckets: **New** (in run B, not A), **Persistent** (in both runs), and **Cleared** (in A, not B). Within each bucket, findings are ordered by severity. Persistent findings show severity changes between runs.
* **Recommendations comparison** - new recommendations, recommendations that persist, and recommendations from the earlier run that no longer appear.
Click **Save comparison** to store it as a named view in the Reports sidebar. Saved comparisons recompute deltas fresh when you reopen them - useful when one run is set to "most recent of this report type."
Click **Share** to generate a URL that resolves the same comparison for any workspace member with access to both runs. To share externally, use **Export** and choose PDF - the export renders both runs side-by-side with deltas highlighted.
The compare surface only works on runs of the same report type. For cross-recipe synthesis ("what does this week's Payment Performance tell me alongside last week's Customer Recovery?"), ask Ask Viq directly - it can synthesise across multiple runs and cite each one in its answer.
## How the archive powers longitudinal pattern detection
The archive is not just a record you browse manually. The platform reads it continuously. For every new report run, the pattern engine compares the new findings against prior runs of the same recipe over the past 30, 60, and 90 days.
When the same finding (or a structurally similar one) recurs enough times to cross the recurrence threshold, the platform treats it as a **structural pattern** rather than a transient blip. That pattern then appears as a card on the Vortex Memory dashboard ("Patterns we noticed"), as a recurrence note in the next relevant report, and as context in Ask Viq answers that touch the pattern's domain.
A pattern clears automatically when the underlying findings stop recurring across two consecutive scheduled runs of the same recipe. Cleared patterns move to a historical view; the full lifecycle - detected, surfaced, snoozed or muted, cleared - is preserved in the audit trail.
Without the archive, the platform cannot tell the difference between noise that happens once and a structural condition recurring weekly. The archive is the substrate that makes this distinction possible.
## How to export a run
Open any archived run and click the **Export** action in the run header. Three formats are available:
A polished, print-ready document with branding, charts at print quality, and recommendations. Best for board packs, investor updates, and compliance trails.
A `.md` file with prose, headings, tables, and references. Best for pasting into a wiki (Confluence, Notion) or a docs system.
The structured output JSON the run produced - schema-stable across runs of the same recipe. Best for custom analysis scripts, BI tools, and external archives.
For a board-ready PDF from a Quarterly Business Review run: enable cover, diagnostic body, and recommendations; enable workspace branding; disable the structured output block. The result is the QBR document you can include directly in a board pack.
To export multiple runs at once, select them in the Reports list and click **Export selected**. The bulk export produces a zip with one file per run.
## How long runs are kept
The default retention is **forever**. The longitudinal value of the archive compounds with depth; most workspaces keep all runs because a year or more of structured findings is what makes pattern detection meaningful.
If your workspace operates under compliance requirements, you can configure a report-retention policy from the Vortex Memory settings page (common windows: 7 years, 5 years, 3 years, 2 years). When the window closes, the run moves to archived state - visible with the "include archived" toggle - rather than being deleted. Explicit deletion by a workspace admin is permanent and irreversible.