FREE INSURED SHIPPING WORLDWIDE
How SEO Is Done: Four Stages From Diagnosis to Lock In
WebSEOGrow
Done Board 4 columns · 11 cards Message
Done Board · Four Columns

How SEO Is Done: Four Stages From Diagnosis to Lock In

Most sites are not short on SEO activity. They are short on finished work. How SEO is done, in practice, is four columns: diagnose, change, observe, lock in. A card only moves right when the column test is met. That single rule is what turns a year of edits into a site that holds its ground.

4 columns · 11 cards · 1 done test per column

DIAGNOSE query export mobile crawl overlap list CHANGE 42 titles internal links hero weight OBSERVE baseline day 28 read LOCK rule 1 rule 2 cards move right only when the column test is true

What does it mean for SEO work to be done?

Done means the change is live, observed against a baseline, and written into a rule the next person will follow. Until all three are true, the work is still in progress, however many hours went in. A board that tracks activity instead of completion stays full forever, and so does the site.

Start with a real diagnosis: crawl the site, read ninety days of queries, and write the blockers down. If you want the full list of what to check, use our Technical SEO Services as the input to column one.

The board, column by column

Eleven cards, four columns. Read the cards as examples, then read the done test under each column. The test is the part that transfers to your site.

Diagnose3
Query export, 90 days
Search Console, sorted by impressions and clicks.
Mobile crawl
Depth, orphan pages, duplicate titles.
Template overlap list
Which templates fight for the same intent.
Done when
  • Every card names a URL, a number, and an owner
  • The list fits on one page
  • Nothing on it is a guess about the future
Change3
Titles and metas, 42 templates
Shipped on production, old versions logged.
Internal links from top 20 pages
Three links each to the pages that convert.
Hero image weight
Compress, resize, serve the right format.
Done when
  • The change is live, not staged
  • The date and page count are logged
  • Somebody outside the project can verify it
Observe2
Baseline sheet
Same query set as the export, frozen on ship day.
Day 28 read
Compared, not eyeballed.
Done when
  • Two read dates have passed
  • The same queries are compared
  • Noise and season are named out loud
Lock in3
Rule: one intent per URL
Written, linked, and enforced in review.
Rule: no page without an owner
Every URL has a person and a date.
Template checklist in the CMS
So the fix ships without a meeting.
Done when
  • The rule is followed without asking you
  • It survives the next redesign
  • A new hire can apply it in a week

What done looks like on one page

When a piece of work is finished, the page below is the whole story. Six lines, no narrative, no screenshots required.

Locked

Completion record

Page
/products/steel-brackets plus eleven siblings in the same template
Change
Title and meta rewritten to one intent, internal links added from three high traffic pages
Date
Live 14 March, verified on a logged out browser the same day
Baseline
420 impressions, 6 clicks, average position 23 on the frozen query set
Reads
Day 28 and day 56, same queries, same device filter
Rule
One intent per URL. New product pages inherit this template before launch

The rule line is the part most teams skip, and it is the only part that outlives the campaign. Without it, the same page gets re-broken by the next person who has a good idea about titles.

One card through all four columns

A tile supplier noticed that its installation guide was earning impressions and almost no clicks. In diagnose, the card named the page, the impression count, and the reason: the title promised a downloadable PDF that the page did not host. That detail is why the card exists. The problem was not a lack of content, it was a promise the page could not keep.

In change, the title and meta were rewritten, the dead download removed, and the guide linked from two product pages. In observe, the baseline was frozen at nineteen hundred impressions and fourteen clicks, and the day twenty eight read showed twenty three hundred impressions and sixty one clicks on the same query set. In lock in, the team wrote one rule: no title may promise a download the page does not host.

The rewriting was the easy part. What made the work finish was the sentence at the end that stopped the next writer from repeating the mistake, and a number that moved in a direction anyone could check.

Why four columns instead of a task list

A task list measures effort. It is honest about hours and silent about outcomes, so the list refills every Monday and nobody can say what shipped. A board measures position: a card is in a column, and the column has a test.

The order also protects the work. Diagnosis without change is a document nobody uses. Change without observation is a coin flip. Observation without lock in is a lesson you pay for twice, because the next redesign undoes the fix and nobody remembers why it was there.

Most agencies and in house teams are good at one column. Technical teams like diagnose. Content teams like change. Almost nobody owns observe and lock in, which is exactly why so many sites improve for two months and then slide back.

A quick test of any board: count the cards that have sat in the same column for more than thirty days. Those are not work in progress, they are decisions nobody wants to make. Diagnose cards that old usually mean the problem is awkward to raise. Change cards that old mean approval is missing. Observe cards that old mean no calendar entry exists.

The two ways SEO work goes wrong

Failure one: endless editing

Every week brings a new tweak, no card ever leaves change, and there is no baseline to judge any of it. The site looks busy and the query set does not move. The fix is not more discipline, it is a smaller column with a hard exit.

Failure two: fire and forget

Work ships, everyone celebrates, and nobody returns on day twenty eight. Six months later a redesign removes the internal links and the traffic quietly falls. The fix is a calendar entry and a written rule, not another audit.

How we run the board with a client

Step 01

Fill diagnose in week one

Crawl, query export, template overlap, and a blocker list with numbers. This is technical work first, which is why our SEO Audit Checklist team owns the column rather than treating it as a content task.

Step 02

Ship change in batches of five

Small batches make observation possible. One giant release hides the effect of each fix and gives you nothing to repeat.

Step 03

Observe with a frozen query set

We freeze the list on ship day and compare the same rows at day twenty eight and day fifty six. New queries are added to the next cycle, never to the current read.

Step 04

Lock in as a rule

Every completed card ends with a sentence someone else can follow. If the rule needs you to interpret it, it is not finished.

How to start the board this week

Create four columns in whatever tool you already use. Add one card per known problem and delete everything that is not a page level change. Put a done test at the top of each column before you add a second card.

Then pick the oldest card in change and ask the only question that matters: is it live, is it measured, and does a rule protect it. If the answer is no to any of the three, the card goes back, not forward.

The first board is always ugly. That is fine. A board with eleven honest cards beats a packed backlog with no exit, and it is the only version of SEO work a client can actually read.

Want the board filled in for you?

Send the domain. We reply with the first four cards we would put in diagnose, each with the number behind it.

Questions about finishing SEO work

How long should a card stay in observe?

Twenty eight days minimum, fifty six if the query set is small or the season is changing. Reading earlier produces a story, not a result.

What if a change makes things worse?

Revert and write it down. A reverted change with a logged result is still a finished card, and it saves you from repeating the idea next year.

Do small sites need four columns?

Yes, but fewer cards. A ten page site might have two cards in diagnose and one in change. The tests are the same, the volume is not.

Who owns lock in?

Whoever can change the template or the review process. If nobody owns it, the column never empties, which is a useful signal about the organisation.

Is the board the same as a report?

No. A report describes what happened. The board shows what is finished, what is waiting, and what rule now protects the fix.