---
title: "Turn Your SEO Audit Into a Ranked Fix List With Claude"
description: "Use Claude through groas to rank SEO audit findings by search traffic and conversions, explain the priorities, and apply approved quick fixes."
url: "https://groas.com/post/your-seo-audit-explained-by-claude-in-plain-english"
image: "https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/a2d80f0b-ae0b-4abb-ae32-0e4355ef0851.png"
published: "2026-11-19T20:23:58.047Z"
modified: "2026-11-19T20:23:58.114Z"
---

November 19, 2026 · 11 min read

# Turn Your SEO Audit Into a Ranked Fix List With Claude

[](https://groas.com/author/alexander-perelman)

[Alexander Perelman](https://groas.com/author/alexander-perelman)Personally Tested

Head Of Product @ groas

[LinkedIn](https://www.linkedin.com/in/alexander-433793253/)

![A towering heap of red-flagged audit printout spills off a desk, topped by one yellow sticky note naming the single fix that actually matters.](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/a2d80f0b-ae0b-4abb-ae32-0e4355ef0851.png)

In this article

1. [Before starting: connect the data and check the tracking](#before-starting-connect-the-data-and-check-the-tracking)
2. [Why the audit needs more than crawler data](#why-the-audit-needs-more-than-crawler-data)
3. [Step 1: get ten priorities in plain English](#step-1-get-ten-priorities-in-plain-english)
4. [Step 2: put clicks and conversions beside the findings](#step-2-put-clicks-and-conversions-beside-the-findings)
5. [Step 3: separate quick fixes from engineering projects](#step-3-separate-quick-fixes-from-engineering-projects)
6. [Step 4: approve and apply the quick fixes](#step-4-approve-and-apply-the-quick-fixes)
7. [Step 5: turn engineering projects into tickets](#step-5-turn-engineering-projects-into-tickets)
8. [Keep judgment calls out of the automatic queue](#keep-judgment-calls-out-of-the-automatic-queue)
9. [Repeat monthly, but ask only for what changed](#repeat-monthly-but-ask-only-for-what-changed)
10. [Verify the fixes, then keep conversions in the ranking](#verify-the-fixes-then-keep-conversions-in-the-ranking)

A 200-row SEO audit export is where good intentions go to die. Put actual search traffic beside those rows, and forty missing meta descriptions can stop looking more important than one broken canonical on a page that brings in customers.

That is the job here. Use Claude through groas to turn an audit export into **a ranked fix list, not another report**: ten priorities explained in plain English, quick fixes ready for approval, and engineering tickets that say why the work matters. Apply the approved site changes with a logged reason and a one-step undo.

Row count is not priority. Neither is whatever the crawler painted red.

## Before starting: connect the data and check the tracking

Have these ready:

- A [groas account](https://groas.com/). It costs $19 a month for the whole team, with unlimited sites and accounts.
- [Claude with groas added as a connector](https://groas.com/claude), with access to the relevant website, SEO audit findings, Search Console data, and GA4 data.
- Your latest SEO audit export from Screaming Frog, Semrush Site Audit, Ahrefs, or your platform. If it is more than 30 days old, rerun it first.
- The site’s platform and any boundaries: what can change, what needs engineering approval, and what is out of scope.

**Confirm the tracking before asking Claude to rank anything.** Check that Search Console covers the right domain variant and that GA4 counts the conversions the business actually cares about. Broken inputs produce a beautifully formatted wrong answer.

Give Claude the platform and boundaries before the audit. Otherwise, it can propose a Next.js refactor for a Drupal store and sound completely sure of itself.

## Why the audit needs more than crawler data

Paste a raw export into a plain chat and Claude can explain the labels. That does not mean it has inspected the pages or checked their performance. Ask it to fill those gaps, and an authoritative answer can become a list of problems the site never had.

Through the [groas connection](https://groas.com/mcp), the workflow brings three datasets together:

- **Audit findings:** what the crawler flagged and which URLs it affects.
- **Search Console clicks and impressions:** which flagged pages receive search traffic.
- **GA4 conversions:** which of those pages contribute to business results.

The third dataset is the one most audit workflows skip. A page with modest clicks and a 4% conversion rate may deserve attention before a busy blog post that converts nobody. The crawler’s severity column cannot make that call on its own.

I use _traffic at stake_ to mean the existing traffic attached to affected pages, not a forecast of clicks a fix will recover. **Clicks establish exposure; conversions and the actual fault inform priority.** A missing description and an indexing blocker do not become equivalent because their pages get the same traffic.

Bring the real data. Do not ask the model to invent the missing columns.

## Step 1: get ten priorities in plain English

Ask Claude to consider the full audit before choosing what to summarize. Starting with ten arbitrary rows only produces a better-written arbitrary list.

```text
Through groas, read our latest SEO audit alongside Search Console
clicks and impressions and GA4 conversions for the affected pages.

Consider all findings, then identify the ten priorities affecting
the most search traffic. Flag crawling or indexing blockers first.

Explain each priority in one sentence, without jargon. Name the fix
in plain words and the role that usually owns it: marketing,
content, or engineering.

Flag missing data rather than guessing. Do not change the site.
```

**Expected result:** ten short explanations a founder can understand, each with a proposed fix and an owner. The summary should explain the problem, not merely translate _canonical issue_ into slightly longer SEO jargon.

**Common mistake:** asking for a summary of all 200 rows. Claude will oblige, and the spreadsheet becomes a wall of prose. The point is to reduce the decision load, not change its file format.

Keep the full audit available for the ranking check. Keep the summary short.

## Step 2: put clicks and conversions beside the findings

Now make the priorities inspectable. A confident sentence is not enough; the affected URLs and their numbers need to sit beside it.

```text
For each proposed priority, show the affected pages, Search Console
clicks and impressions in the last 90 days, and GA4 conversions
for the same period.

Sort by the clicks attached to those pages. Flag crawling and
indexing blockers separately, and explain where conversions
change the recommended order.

Check the remaining audit findings for anything that should
displace an item in the top ten. Explain any change.

Treat clicks as traffic exposure, not predicted losses or gains.
Flag missing data rather than filling it in.
```

**Expected result:** a ranking table with URLs, clicks, impressions, conversions, and a short explanation of exceptions to the click-based order.

The reorder may feel wrong. Forty missing descriptions can fall below a single canonical pointing at a dead URL on a busy hub. A lower-traffic page with conversions can move ahead of a blog post that attracts visitors but no customers.

**Common mistake:** treating _clicks at risk_ as a measured loss. Those clicks show which pages deserve scrutiny. They do not prove that the flagged issue threatens every visit, or that fixing it will produce a gain.

Use the groas connection rather than a spreadsheet trimmed to fit a chat window. **Rank from the actual numbers, then inspect the reason for the rank.** Severity badges are not a substitute for either.

## Step 3: separate quick fixes from engineering projects

A ranked list still needs a work plan. Split it by who can act, and keep blockers from disappearing beneath easy copy edits.

```text
Split the ranked findings into three buckets:

1. Quick fixes a non-developer can ship this week: titles,
   meta descriptions, internal links, and image alt text.
2. Projects needing engineering: templates, site speed,
   redirect rules, and schema across many pages.
3. Leave it: findings unlikely to move traffic. Give a reason
   for each decision to leave something alone.

Put crawling and indexing blockers ahead of cosmetic quick fixes.
For engineering projects, identify the responsible template or
layout where the data supports it. Mark unknown causes as unknown.
Do not change the site.
```

**Expected result:** three work queues, with the reason for each assignment and any unresolved diagnosis made explicit.

Two judgments matter here. First, **fix the template, not every page it produces**. A shared layout can spread one problem across hundreds of URLs; fixing each URL individually is a particularly expensive way to avoid fixing the cause.

Second, _leave it_ needs a written reason. “Later, or never” is a decision, not an oversight. An audit that marks nothing ignorable has not finished the prioritization work.

**Common mistake:** chasing quick wins while a blocker remains live. A `noindex` on a revenue page, a robots disallow, or a canonical pointing at a 404 deserves attention before a title rewrite on that page. Do not let ease of execution become the ranking system again.

## Step 4: approve and apply the quick fixes

This is where an explanation becomes work shipped. Through groas, Claude can apply site changes such as titles, meta descriptions, and internal links within the guardrails set for it. Changes have logged reasons and a one-step undo.

Start with a preview, not permission to tidy the whole site.

```text
For each quick fix in bucket 1, draft the exact change.
Show the URL, the current version, the proposed version,
and the reason in a before-and-after table.

Keep changes limited to the approved field or element.
Do not replace page bodies or remove unrelated content.
Show before-and-after word counts for any body-content change.

Wait for my approval. Then apply only the changes I approve
through groas, and log the reason for each.

After applying them, verify what the live page serves and
report any difference from the approved version.
```

**Expected result:** an approval table first. After approval, the specified changes appear on the live pages, with reasons in the change log and an undo available.

**Common mistake:** accepting “updated” as verification. Open the pages. Check that the title, description, or link matches the approved change and that unrelated content remains intact. A successful tool response is not the same thing as a correct page.

**Approval comes before execution; live-page verification comes after it.** Keep both checks.

A read-only connector stops short of this step. Semrush’s Claude integration can retrieve completed Site Audit data and help explain and prioritize it, but it cannot touch the site. The output remains an assignment list. With groas, the approved quick fixes can move beyond the list.

## Step 5: turn engineering projects into tickets

Bucket 2 dies in a shared drive unless each item becomes an assignment. Ask for the reason, affected scope, and verification method, not a copied crawler label.

```text
For each item in bucket 2, write a ticket with four lines:

1. The problem in plain English.
2. The affected pages, their clicks in the last 90 days,
   and their GA4 conversions for the same period.
3. The responsible template or layout, or what still needs
   investigation if the cause is unknown.
4. How to verify the fix worked.

Address each ticket to the owning role: engineering, content,
or SEO. Preserve the priority order and flag blockers.
```

**Expected result:** a ranked set of tickets that an owner can assess without reopening the entire audit.

Compare “canonical issue on `/pricing`” with an illustrative ticket that says: “`/pricing` points its canonical at a deleted URL, receives 3,100 clicks a month, and shares the affected template with 40 product pages.” The second explains why the work belongs in the queue. The first mostly explains why someone needs another meeting.

**Common mistake:** sending every ticket as equally urgent, or sending them one at a time without the ranked context. That recreates the original audit in a project-management tool.

**Ranking is the deliverable. The export is raw material.** Hand over the ordered set, with owners and reasons attached.

_The report nobody triages: when everything is critical, the useful instruction fits on a sticky note._

## Keep judgment calls out of the automatic queue

Claude can misjudge severity and sound certain while doing it. Flags such as _multiple H1s_ or _titles over 60 characters_ can reflect deliberate choices. An export does not explain the intent behind them.

Meta descriptions are the classic trap. The crawler flags them; Google rewrites many of them in search results anyway. That does not make every description useless. It makes a blanket “fix all missing descriptions first” instruction a poor substitute for prioritization.

**Before approving a fix, open the page and confirm the problem exists.** Check the recommendation against the site, not just against another sentence in the audit.

Keep developers responsible for templates and speed work. Do not commission a Core Web Vitals sprint simply because a score looks uncomfortable. Ahrefs’ [Patrick Stox describes Core Web Vitals as a small ranking factor](https://ahrefs.com/blog/core-web-vitals) and says he generally will not prioritize them unless a site is extremely slow.

I want a specific reason for that engineering work: which template needs changing, which URLs it affects, and which of those URLs get clicks. “Make the score greener” is not a useful ticket.

## Repeat monthly, but ask only for what changed

Sites drift. A template changes, thin pages appear, or a rewrite quietly drops content from a hub. Monthly is often enough; use a weekly routine when publishing at volume.

```text
Through groas, rerun the SEO audit and compare it with last month's.

Return three things only:
1. Which fixes held.
2. What is new and affects pages with search clicks.
3. Which findings moved up the priority ranking, and why.

Include Search Console traffic and GA4 conversions for the
relevant pages. Flag blockers separately.
Do not apply changes without approval.
```

**Expected result:** a short _delta report_, not another explanation of every finding the crawler has ever encountered.

**Common mistake:** commissioning a fresh wall of text each month. That is how the workflow becomes the [60-page audit that misses the door](https://groas.com/post/your-ai-technical-seo-audit-is-60-pages).

Keep the loop small:

1. Rerun the audit.
2. Rank the changes by traffic exposure, blockers, and conversions.
3. Approve and ship the quick fixes; assign the projects.
4. Verify the changes and review performance.

**Ask what changed, not what can fill a report.**

## Verify the fixes, then keep conversions in the ranking

Verify execution immediately: inspect the live changes, confirm the intended fix, and check for unintended edits. After a month, pull the ranking table again using the same 90-day window length as step 2. Review clicks, impressions, and conversions alongside the work shipped.

Keep the questions narrow:

- Does the hub page now serve the intended canonical?
- Have its impressions recovered?
- Have the rewritten titles been followed by movement from position 11 toward position 8?
- Do the same audit findings return after the next crawl?

**A shorter audit helps only if the important problems stop returning.** A delta report saying _nothing new_ is more useful than a rising score. If the same template keeps producing new issues, fix the upstream process rather than celebrating another round of cleanup.

The first thing to change once the loop works: make **GA4 conversions a permanent column in every ranking table and ticket**. Do not let it disappear when the report gets shortened for a meeting.

“High-impact SEO fix” asks the room to trust a label. “3,100 clicks and 90 conversions a month on this page” gives it a business reason to care. Numbers that touch revenue survive the room. Everything else is a row in a spreadsheet.

## Frequently Asked Questions

### Why does an SEO audit need Search Console and GA4 data alongside it?

Crawler findings alone cannot tell you which problems matter. Search Console clicks and impressions show which flagged pages receive search traffic, and GA4 conversions show which of those pages contribute to business results. Forty missing meta descriptions can matter less than one broken canonical on a page that brings in customers.

### How do I turn a 200-row SEO audit into a short priority list?

Ask Claude, through groas, to read the full audit alongside Search Console and GA4 data, consider all findings, then identify the ten priorities affecting the most search traffic, flagging crawling or indexing blockers first. Each priority comes back as a one-sentence explanation in plain words, with a proposed fix and an owner: marketing, content, or engineering.

### Does a click count attached to a flagged page mean I am losing those clicks?

No. Traffic at stake means the existing traffic attached to the affected pages, not a forecast of clicks a fix will recover. Clicks establish exposure; conversions and the actual fault inform priority. A missing description and an indexing blocker are not equivalent just because their pages get the same traffic.

### How should I split ranked SEO findings into a work plan?

Split them into three buckets: quick fixes a non-developer can ship this week, such as titles, meta descriptions, internal links, and image alt text; projects needing engineering, such as templates, site speed, redirect rules, and schema across many pages; and a leave-it bucket with a written reason for each decision. Put crawling and indexing blockers ahead of cosmetic quick fixes.

### Can Claude apply the approved SEO fixes to my live site?

Yes, through groas. Claude first drafts each quick fix in a before-and-after table showing the URL, current version, proposed version, and reason. After your approval, it applies only the approved changes through groas, logs the reason for each, and offers a one-step undo. Then verify what the live page actually serves.

### What should an engineering ticket for an SEO fix contain?

Four lines: the problem in plain English; the affected pages with their Search Console clicks and GA4 conversions for the last 90 days; the responsible template or layout, or what still needs investigation if the cause is unknown; and how to verify the fix worked. Each ticket goes to the owning role: engineering, content, or SEO, preserving the priority order.

### Should I fix every issue my SEO crawler flags?

No. Open the page and confirm the problem exists before approving a fix, because flags such as multiple H1s or titles over 60 characters can reflect deliberate choices. Meta descriptions are a classic trap: Google rewrites many of them in search results anyway, so a blanket instruction to fix all missing descriptions first is a poor substitute for prioritization.

### How often should I rerun an SEO audit and what should I ask for?

Monthly is often enough; use a weekly routine when publishing at volume. Rerun the audit and compare it with last month's, then ask for three things only: which fixes held, what is new and affects pages with search clicks, and which findings moved up the priority ranking and why. The result is a short delta report, not another explanation of every finding.

## Run Your Marketing On groas

Plug ChatGPT, Claude or any AI agent into groas and let it do the work of a whole marketing team: $19 a month flat with every feature included, and autopilot whenever you’re ready.

[Start For $19](https://groas.com/pricing)

[Or have groas run it for you](https://groas.com/done-for-you)

## Structured data

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://graph.groas.com/entity/groas","name":"groas","url":"https://groas.com/","description":"groas is the marketing infrastructure for your own AI: it connects your AI to all your marketing and gives it proprietary marketing data collected 24/7. Plug Claude, ChatGPT or any AI agent into groas and run all your marketing from it, for a tiny fraction of the usual cost, with the groas app there whenever you want to see more. It is the layer ChatGPT, Claude and any AI assistant run a business’s online marketing on. Through groas an assistant can do anything you could do natively in Google Ads, Bing Ads, Meta Ads and ChatGPT ads, on your website and in AI search, and, where the owner switches it on, can run connected campaigns by itself, so AI agents do the work a whole marketing team used to. groas also gives those agents large-scale proprietary marketing data it collects around the clock, on any schedule the owner chooses: every ad competitors run on Google and Meta, their landing pages and full funnels with the offers and prices in them, Reddit customer research, AI search answers from ChatGPT, Google AI Overviews, Gemini, Claude and Perplexity, and SEO data from Semrush and Ahrefs, backlinks included ($19 a month, flat, with everything included, for unlimited teammates on one shared workspace and unlimited ad accounts and websites, delivered as an MCP server). Separately, its done-for-you service has senior people and proprietary quantitative systems run paid ads, SEO and AI search for businesses, and white label for agencies, from $5,000 a month.","logo":{"@type":"ImageObject","url":"https://groas.com/icon-512.png","width":512,"height":512},"sameAs":["https://www.linkedin.com/company/groas/"]},{"@type":"WebSite","@id":"https://groas.com/#website","name":"groas","url":"https://groas.com/","inLanguage":"en-US","publisher":{"@id":"https://graph.groas.com/entity/groas"}},{"@type":"WebPage","@id":"https://groas.com/post/your-seo-audit-explained-by-claude-in-plain-english#webpage","url":"https://groas.com/post/your-seo-audit-explained-by-claude-in-plain-english","name":"Turn Your SEO Audit Into a Ranked Fix List With Claude","description":"Use Claude through groas to rank SEO audit findings by search traffic and conversions, explain the priorities, and apply approved quick fixes.","inLanguage":"en-US","isPartOf":{"@id":"https://groas.com/#website"},"about":{"@id":"https://graph.groas.com/entity/groas"},"primaryImageOfPage":{"@type":"ImageObject","url":"https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/a2d80f0b-ae0b-4abb-ae32-0e4355ef0851.png"},"breadcrumb":{"@id":"https://groas.com/post/your-seo-audit-explained-by-claude-in-plain-english#breadcrumb"},"dateModified":"2026-11-19T20:23:58.114Z","mainEntity":{"@id":"https://groas.com/post/your-seo-audit-explained-by-claude-in-plain-english#article"}},{"@type":"BreadcrumbList","@id":"https://groas.com/post/your-seo-audit-explained-by-claude-in-plain-english#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://groas.com/"},{"@type":"ListItem","position":2,"name":"Blog","item":"https://groas.com/blog"},{"@type":"ListItem","position":3,"name":"Turn Your SEO Audit Into a Ranked Fix List With Claude","item":"https://groas.com/post/your-seo-audit-explained-by-claude-in-plain-english"}]},{"@type":"BlogPosting","@id":"https://groas.com/post/your-seo-audit-explained-by-claude-in-plain-english#article","url":"https://groas.com/post/your-seo-audit-explained-by-claude-in-plain-english","mainEntityOfPage":{"@id":"https://groas.com/post/your-seo-audit-explained-by-claude-in-plain-english#webpage"},"isPartOf":{"@id":"https://groas.com/blog#blog"},"headline":"Turn Your SEO Audit Into a Ranked Fix List With Claude","description":"Use Claude through groas to rank SEO audit findings by search traffic and conversions, explain the priorities, and apply approved quick fixes.","datePublished":"2026-11-19T20:23:58.047Z","dateModified":"2026-11-19T20:23:58.114Z","image":{"@type":"ImageObject","url":"https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/a2d80f0b-ae0b-4abb-ae32-0e4355ef0851.png"},"author":{"@id":"https://groas.com/author/alexander-perelman#person"},"publisher":{"@id":"https://graph.groas.com/entity/groas"},"keywords":"audit, claude, seo audit, seo, fix, pages, clicks, list","wordCount":2252,"inLanguage":"en-US"},{"@type":"Person","@id":"https://groas.com/author/alexander-perelman#person","name":"Alexander Perelman","jobTitle":"Head Of Product @ groas","description":"My background includes Goldman Sachs and Stanford Computer Science.","image":"https://groas.com/media/blog/4f17cac3a81acc224cc1b2cfab8f94bbc295a3089d13e8421125771ef8d4064f.jpg","worksFor":{"@id":"https://graph.groas.com/entity/groas"},"sameAs":["https://www.linkedin.com/in/alexander-433793253/"]},{"@type":"FAQPage","@id":"https://groas.com/post/your-seo-audit-explained-by-claude-in-plain-english#faq","mainEntity":[{"@type":"Question","name":"Why does an SEO audit need Search Console and GA4 data alongside it?","acceptedAnswer":{"@type":"Answer","text":"Crawler findings alone cannot tell you which problems matter. Search Console clicks and impressions show which flagged pages receive search traffic, and GA4 conversions show which of those pages contribute to business results. Forty missing meta descriptions can matter less than one broken canonical on a page that brings in customers."}},{"@type":"Question","name":"How do I turn a 200-row SEO audit into a short priority list?","acceptedAnswer":{"@type":"Answer","text":"Ask Claude, through groas, to read the full audit alongside Search Console and GA4 data, consider all findings, then identify the ten priorities affecting the most search traffic, flagging crawling or indexing blockers first. Each priority comes back as a one-sentence explanation in plain words, with a proposed fix and an owner: marketing, content, or engineering."}},{"@type":"Question","name":"Does a click count attached to a flagged page mean I am losing those clicks?","acceptedAnswer":{"@type":"Answer","text":"No. Traffic at stake means the existing traffic attached to the affected pages, not a forecast of clicks a fix will recover. Clicks establish exposure; conversions and the actual fault inform priority. A missing description and an indexing blocker are not equivalent just because their pages get the same traffic."}},{"@type":"Question","name":"How should I split ranked SEO findings into a work plan?","acceptedAnswer":{"@type":"Answer","text":"Split them into three buckets: quick fixes a non-developer can ship this week, such as titles, meta descriptions, internal links, and image alt text; projects needing engineering, such as templates, site speed, redirect rules, and schema across many pages; and a leave-it bucket with a written reason for each decision. Put crawling and indexing blockers ahead of cosmetic quick fixes."}},{"@type":"Question","name":"Can Claude apply the approved SEO fixes to my live site?","acceptedAnswer":{"@type":"Answer","text":"Yes, through groas. Claude first drafts each quick fix in a before-and-after table showing the URL, current version, proposed version, and reason. After your approval, it applies only the approved changes through groas, logs the reason for each, and offers a one-step undo. Then verify what the live page actually serves."}},{"@type":"Question","name":"What should an engineering ticket for an SEO fix contain?","acceptedAnswer":{"@type":"Answer","text":"Four lines: the problem in plain English; the affected pages with their Search Console clicks and GA4 conversions for the last 90 days; the responsible template or layout, or what still needs investigation if the cause is unknown; and how to verify the fix worked. Each ticket goes to the owning role: engineering, content, or SEO, preserving the priority order."}},{"@type":"Question","name":"Should I fix every issue my SEO crawler flags?","acceptedAnswer":{"@type":"Answer","text":"No. Open the page and confirm the problem exists before approving a fix, because flags such as multiple H1s or titles over 60 characters can reflect deliberate choices. Meta descriptions are a classic trap: Google rewrites many of them in search results anyway, so a blanket instruction to fix all missing descriptions first is a poor substitute for prioritization."}},{"@type":"Question","name":"How often should I rerun an SEO audit and what should I ask for?","acceptedAnswer":{"@type":"Answer","text":"Monthly is often enough; use a weekly routine when publishing at volume. Rerun the audit and compare it with last month's, then ask for three things only: which fixes held, what is new and affects pages with search clicks, and which findings moved up the priority ranking and why. The result is a short delta report, not another explanation of every finding."}}]}]}
```
