The Problems to fix page

Problems to fix is the audit half of the SEO module in one list. Its description says what's on it: "Everything Rankbank found wrong when it read your site — broken links, pages nothing points to, missing titles. Worst first, with the reason it matters."

Problems and findings are different things

Rankbank keeps two lists, and the distinction is the one thing to get straight before reading either.

A problem is something structurally wrong that Rankbank saw while reading your site. A link points at a dead page. A page has no title. Nothing on your site links to a page. These are facts about your HTML, they need no search data to detect, and they appear the moment a read finishes.

A finding is an opportunity, usually priced in clicks per month, worked out from your search data and your pages together. "This page is seen 4,000 times a month and wins almost no clicks" is a finding. Findings live on the Overview under Do these three next, in the growth plan, and in the other Search growth pages. See findings: how to read a card.

Some things appear in both forms, deliberately. A broken link is a problem on this page and also rolls up into a Links to dead pages finding, because the same fact deserves both a task list entry and a place in the priority queue.

The summary tiles

Five tiles across the top, each with a one-line description.

Tile	What it counts
Still to fix	"Open problems on your site."
Urgent	"Costing you the most — start here."
High	"Worth fixing this week."
Fixed	"Confirmed gone on the latest read."
Came back	"Fixed once, broken again."

Came back is the one worth pausing on. A problem lands there when Rankbank confirmed it gone on one read and then found it again on a later one. That usually means a template change reverted, a deploy overwrote a fix, or a CMS setting reset itself. A problem that keeps coming back is a process problem, not a page problem, and it's worth fixing at the template level rather than the page level.

Tiles only render when they have something to show, so a clean site shows fewer than five.

The filter row

Under the tiles is a row labelled Show me only: with five buttons: Urgent, High, Medium, Low, and Everything. They filter by priority band. Everything clears the filter and returns you to the unfiltered page.

Clicking a row in the Worth fixing first panel on the Overview brings you here filtered to that one problem type, which is usually a faster way in than filtering by hand.

Since the last read

A comparison card appears once there's been more than one read and something has actually moved. It's titled Since the last read with the description "What changed between the previous audit and this one." Inside are up to four figures: New, Still there, Fixed and Came back. Each one only appears when it's non-zero, so the card never pads itself with empty numbers.

Read it after a fixing session, then read your site again and check whether Fixed matches what you did. If it doesn't, the change didn't land the way you thought.

The problems themselves

The list is headed The problems, described as "Worst first — the ones likely costing you the most are at the top."

Each row shows:

• The plain-language title, like "A page has no main heading"
• A priority chip, but only for urgent and high, and an Ignored chip when the problem is ignored
• One sentence saying what it means and why it matters
• The URL of the page it's about
• "Affects {n} pages." when it covers more than one
• A bold Fix: line naming the action, like "Add one clear main heading (H1) to the page."

All 23 titles, meanings and fix actions are documented in every problem type, explained.

If nothing matches, the empty state reads Nothing here with the description "No problems match this filter — try Everything, or run a fresh audit."

What you can do with a problem

Each row has its own controls on the right.

I fixed this marks it resolved. Use it when you've made the change on your site. The next read is what actually confirms it, so a problem you marked fixed can come back if the change didn't take.

Ignore opens a small field labelled Why ignore it? (optional) with the placeholder "e.g. that page is meant to be private", and a confirming Ignore this problem button. The reason is optional and it's for you, not for us. Six months later "that page is meant to be private" is the difference between a considered decision and a mystery.

Reopen puts a fixed problem back on the list. Stop ignoring does the same for an ignored one.

Ignore is the right tool more often than people think. A staging page nothing links to, a legal page with no incoming links by design, a thin page that exists to redirect: these are all problems by the rules and fine in reality. Ignoring them keeps the list honest, and a list you trust is a list you'll actually work.

Export CSV

The Export CSV button at the top of the page downloads every problem, not just the ten shown, with these columns:

issueType, category, severity, priority, page, relatedUrl, evidence, status, comparisonState, crawlRun

issueType is the machine key, like broken_internal_link. comparisonState is where the problem sits relative to the previous read: new, persisting, resolved or regressed. crawlRun identifies which read produced the row.

This is the right export for handing work to a developer, splitting a large site's problems across a team, or tracking a backlog in a spreadsheet. It's also the only way to see all of a large list at once.

Where to go next

Problems you've decided to act on don't automatically become tasks. Your fix list holds the changes you accepted from findings and from suggested links, and it has its own check-back loop where Rankbank re-reads the page to confirm a link really appeared.

For the structural problems here, the loop is simpler: fix it on your site, mark I fixed this, and read the site again. How Rankbank reads your site covers what a read does and when to run one.