Technical & Auditing

New in V3

AI recommendations that tell you what to fix, and how to fix it

Pulls your site audit findings, Search Console performance, AI visibility gaps and technical health into one ranked list of opportunities, each with the pages affected, a difficulty rating, an estimated time and step by step guidance for actually doing it.

Start your 14-day free trial See pricing

100% data privacy
99.99% uptime
Nothing to configure
Export anytime, no limits

Trusted by leading companies

The problem

Knowing what is wrong is the easy half, most tools hand you a list of problems and leave you to work out which are worth an afternoon, which ones need a developer and what to actually do about any of them.

The tool

Everything worth doing, in the order worth doing it

Recommendations reads your latest site audit, your Search Console and GA4 data and your AI visibility results, then writes a single ranked board of opportunities with the evidence behind each one, so the question stops being what is wrong with the site and becomes what are we doing this week.

One ranked list, not five separate reports

Technical issues from your audit, low CTR pages from Search Console, striking-distance keywords and gaps in your AI visibility all arrive on the same board, ranked against each other rather than sitting in tools that never talk to one another.

Enough context to make the call

Every recommendation carries a severity, a difficulty, an estimated time and the number of pages it touches, so you can tell in a couple of seconds whether it is a fifteen minute job or something you need to book developer time for.

Guidance, not just a diagnosis

Open any card and you get a plain description of the problem, why it matters, the steps to fix it and every affected URL, which is the difference between a list of complaints and a list of work.

AI recommendations & opportunities

11 showing of 11 · 1 critical · 5 quick wins

Five sets of data, weighed up together

A recommendation is only as good as what it was calculated from, so rather than running a checklist against your homepage, the board is built from the data you have already connected.

Your site audit

Every detector the crawl runs feeds in, so broken links, missing canonicals, 404s, invalid schema and long titles all arrive with their issue code and the pages affected.

Google Search Console

Pages with strong impressions and weak click-through, keywords sitting in positions 11 to 20, and queries you are being served for but not clicked on.

GA4

Organic sessions and conversion behaviour, so a recommendation on a page that actually earns you money gets weighted differently to one on a page nobody lands on.

AI visibility

Prompts where competitors are named and you are not, unlinked brand mentions, and the technical readiness checks that decide whether AI crawlers can reach you at all.

Your own performance history

Because your Search Console data is warehoused rather than capped at sixteen months, the comparison behind a recommendation can reach back further than Google's own reports will let it.

Every card prints the source it came from underneath it, so you can always see whether a recommendation came out of the crawl, out of Search Console or out of your AI visibility prompts.

Six fields, and every one of them changes what you do next

The reason most audit output never gets implemented is that it tells you what is wrong without telling you how big it is, how hard it is or how long it takes, so every card here answers all three before you have opened it.

A single recommendation card in SEO Stack for fixing broken internal links, showing category, SEO score impact, affected pages, difficulty, estimated time and the data source — One card from a real board, with the issue code and affected count printed underneath it.

Category and severity

Technical, On Page, Content, Off-Page SEO, AI (GEO), Keyword Opportunity or Website Setup, each rated critical, high, medium or low.

Impact

Either the points it will add to your SEO health score, or the outcome it moves, i.e. improve AI citations or push keywords into the top 10.

Affected pages, URLs or keywords

The exact count, with the full list one click away and every URL copyable rather than a sample of three.

Difficulty

Easy, medium or hard, which is usually the field that decides whether you do it yourself this afternoon or raise a ticket.

Estimated time

From fifteen minutes for an llms.txt file up to one to two weeks for an outreach campaign, so a week's work can be planned rather than guessed at.

Data source

The scan, the issue code, the number affected and a sample URL, printed on the card so nothing on the board is unexplained.

Inside the tool

From the whole board down to one issue code

Every opportunity on one screen

A summary strip with opportunities, critical issues, quick wins, your SEO health score and the estimated traffic gain
How many are new and how many have been fixed since your last check
Categories down the left with a live count against each one
Quick wins listed separately, so the low effort work is always visible
Export the whole board when it needs to live in a spreadsheet or a ticket system

The AI Recommendations and opportunities board in SEO Stack with the summary strip, categories and the first opportunities

The guidance

Written for the issue in front of you

Anybody can generate a list of SEO problems, the harder part is explaining what to do about a specific one on a specific site, so every recommendation opens into guidance that describes the issue, says why it is worth your time and gives you the steps.

Fix guide

Add canonical tags

Status: P1

Issue description

Missing canonicals leave duplicate URL handling ambiguous for Google. SEO Stack cannot change your live host automatically, so follow the steps below in Site Health, then re-run a full audit to clear this card.

Why this matters

Improve crawlability and indexation quality
Reduce Site Health issue count and raise SEO score
Prevent ranking loss from technical debt
Verify fixes with the next full audit

How to fix it

1Add a self-referencing rel=canonical to every affected template
2Point parameter and pagination URLs at the version you want indexed
3Re-run a full audit and check the card clears

Audit issue snapshot

47 pages

Affected

https://www.assertive-media.co.uk/blog/

Sample URL

Priority

MISSING_CANONICAL

Issue code

Affected URLs

https://www.assertive-media.co.uk/blog/
https://www.assertive-media.co.uk/blog/a-new-way-to-explore-the-web-with-ai-mode-in-chrome/
https://www.assertive-media.co.uk/blog/b2b-seo-what-is-business-to-business-seo/
https://www.assertive-media.co.uk/blog/can-ai-agents-run-seo-autonomously/

Every affected URL is listed and copyable, not just the first few.

Illustrative. Switch recommendation to see how the guidance changes with the issue.

The issue description and why this matters sections of a fix guide in SEO Stack

The audit issue snapshot in SEO Stack showing 47 pages affected, a sample URL, a P1 priority and the MISSING_CANONICAL issue code — The snapshot ties the recommendation back to the exact detector that raised it, so it can be verified rather than taken on trust.

What it checks

From broken links to hreflang, and everything between

The recommendations are only as broad as the checks behind them, so the crawl looks at the technical, structural and content signals that actually decide whether a page performs, then writes up whatever it finds.

Links and anchors

Broken internal links, with the page each one sits on
HTTP 404s from your own navigation and from stale sitemap entries
Redirect chains and links pointing at a redirect rather than the destination
Anchor text that carries no meaning, and orphan pages nothing links to
Internal linking opportunities into pages sitting in positions 11 to 20

Canonicals and indexability

Missing canonical tags and canonicals pointing at a different URL to the one crawled
Noindex directives on pages that are linked from your navigation
Robots.txt rules blocking pages you want indexed
Pages crawled but never indexed, checked against Google URL Inspection

Hreflang and international

Missing or malformed hreflang annotations
Return tags that do not point back, so the pairing is never honoured
Language and region codes that are not valid
x-default handling on the pages that need it

Schema and markup

Pages with no JSON-LD structured data at all
Invalid markup and required fields left empty
Product, Article and FAQ opportunities on pages that qualify for them
Markup that will not earn a rich result in its current state

Titles, metas and content

Titles beyond the length Google will display, and titles too short to say anything
Missing and duplicated meta descriptions
Pages with strong impressions and weak click-through rate
Images with no alt text

AI and off-page

llms.txt detection, and generation if you do not have one
AI crawler access for GPTBot, ClaudeBot, PerplexityBot and the rest
Prompts where competitors are named and your brand is not
Unlinked brand mentions worth chasing

Estimated traffic gain

A number to argue for the time with

Getting technical work prioritised usually means putting a figure against it, so the board estimates the monthly upside of the open opportunities using your own Search Console clicks rather than a third party volume estimate.

16
Opportunities: 2
Critical issues: 8
Quick wins: +56K
Estimated traffic gain

The summary strip on the recommendations board in SEO Stack showing 16 opportunities, 2 critical issues, 8 quick wins, an SEO health score of 81 and an estimated traffic gain of 56K — The strip that sits above the board, taken from a real account.

Worth being clear about

An estimate is an estimate. The figure is modelled from the clicks and impressions those pages already earn and what they would earn at a better position or a better click-through rate, so it tells you which opportunities are bigger than the others, it is not a promise of a number. Naturally the accuracy depends a great deal on how much history the property has, and it really does vary from site to site, so we would treat it as a way of ranking work rather than as a forecast you take to a board meeting.

Quick wins and actions

There is always something you can do in fifteen minutes

Quick wins are pulled out into their own list, so when you have half an hour rather than half a week there is a set of low effort recommendations waiting, each with the number of pages it touches against it. Typically these are the ones that clear the fastest and move the health score first, which makes them a reasonable place to start on a site you have only just taken on.

Generate the file for you
Where the fix is a file rather than a change, i.e. an llms.txt or a starter sitemap.xml, the card generates it there and then instead of describing what one should contain.
Open the fix guide
The full write-up, the affected URLs and the steps, for anything that needs a change on your side rather than a download.
Go straight to Site Health
For issues raised by the crawl, the card takes you to the audit view where you can see the pages, the source and the line the issue came from.
Ignore, and stay ignored
Anything that does not apply gets dismissed and stays dismissed, so the board does not re-suggest the same thing after every audit.

The quick wins list in SEO Stack with eight low effort recommendations and the number of pages each one affects

The changes, status, severity and sort by filters on the recommendations board in SEO Stack — Filter by what has changed, by status or by severity, then sort by impact or by effort.

How SEOs use it

What Recommendations is actually good for

Plan the week in ten minutes

Sort by impact, take the top three, check the estimated times against what you have got, and the plan is written. Most of the arguing about what to do next comes from not having effort and impact in the same view.

Hand work over cleanly

A developer gets the issue, the reason it matters, the steps and every affected URL in one place, so the handover is a link rather than a meeting and a spreadsheet.

Report progress to a client

The board shows how many opportunities are new and how many have been fixed since the last check, which is a far easier conversation than trying to describe technical work that nobody outside the team can see.

Work one area at a time

Filter to Technical while a developer has capacity, then to On Page while a copywriter does, so two people can work off the same board without tripping over each other.

Bring AI visibility into the same plan

Recommendations about llms.txt, crawler access and prompts where competitors are named sit alongside the technical work rather than in a separate tool nobody opens.

Ask the AI assistant to go further

The board tells you what to do, and if you want the reasoning, a comparison or a version written for a client, the AI assistant is reading the same warehoused data underneath it.

What you get out of it

The outcome, not the feature list

Start the week with a ranked list instead of a blank page

Give a developer the issue, the pages and the steps in one handover

Show what changed since the last check, and what has already been fixed

Works well with

Connect a property. Watch the data start stacking.

Warehousing begins the moment you connect your Search Console property, there is no Google Cloud project to set up, no BigQuery and no schema to design. You simply connect your property and SEO Stack starts storing every row of your data from that day onwards.

Start your 14-day free trial See pricing

Credit card required
Cancel within the trial and you won't be billed
No contracts

AI recommendations that tell you what to fix, and how to fix it

Everything worth doing, in the order worth doing it

Fix broken internal links

Add canonical tags

Fix HTTP 404 Not Found pages

Create llms.txt file

No XML sitemap detected

Add internal links to striking-distance keywords

Fix invalid JSON-LD markup

Optimize meta descriptions

Shorten overly long title tags

Add Product schema markup

Unlinked brand mentions

Five sets of data, weighed up together

Your site audit

Google Search Console

GA4

AI visibility

Your own performance history

Six fields, and every one of them changes what you do next

Category and severity

Impact

Affected pages, URLs or keywords

Difficulty

Estimated time

Data source

From the whole board down to one issue code

Every opportunity on one screen

Ranked, described and evidenced

Narrow it down to what you are working on

The detail behind a single recommendation

Written for the issue in front of you

Add canonical tags

From broken links to hreflang, and everything between

Links and anchors

Canonicals and indexability

Hreflang and international

Schema and markup

Titles, metas and content

AI and off-page

A number to argue for the time with

There is always something you can do in fifteen minutes

What Recommendations is actually good for

Plan the week in ten minutes

Hand work over cleanly

Report progress to a client

Work one area at a time

Bring AI visibility into the same plan

Ask the AI assistant to go further

The outcome, not the feature list

More in Technical & Auditing

SEO Site Audit

SEO Presence

Connect a property. Watch the data start stacking.