E1 · Four reminders
20 minPut the template's reminders on your calendar: weekly, monthly, day 30 and day 60 after your next publication.
Done when: your calendar has 4 recurring or dated reminders, each with the task written out.
Lesson 8 · 20 min
SEO pays off in weeks and months, not days. What makes the result is the routine: one look per week, one battery of checks per month and, between days 30 and 60, an answer for each symptom instead of giving up. And one precaution almost nobody takes: Search Console keeps only about 16 months. If you don't archive, you lose your own history.
Start the lesson
Goal: set up the weekly and monthly routine, archive the history and choose the right answer for each symptom from days 30 to 60.
Click each block.
A little each week, always. This routine is what separates signal from scare.
Click each step.
Pick a fixed day and time. Run the export (or the spreadsheet), inspect the key URLs, and write down one decision.
One decision per week is enough: change a title, write a topic, strengthen a page.
The weekly report runs every Tuesday morning and only generates files; a person decides and publishes.
Run the same 20 to 30 questions in the same assistants, in an incognito window, and note whether your site appears, with which address, and who appears in its place.
A paid "share of voice" tool is a sample, not an official score. A local model does not measure what the real AIs answer.
Read the symptom before acting: few indexed, indexed with no impressions, impressions with no clicks, position 11 to 30 stalled, a drop after publishing a lot.
Each one has a response in the table below.
Save the resumo.csv and the non-branded files from every week in a folder with a backup.
Next step: the gsc-historico tool imports that folder into a local database that never deletes anything and generates the up, down, and topic report.
Replace what is between < > with your own. The gsc-historico commands come from the tool's README; the details of the tool are covered in course 7.
| symptom (day 30–60) | likely cause | what to do | |---|---|---| | few pages indexed | crawling, wrong canonical, thin page | URL Inspection, internal links, noindex on the thin ones, request indexing again | | indexed, almost no impressions | keyword with no demand or lack of authority | revalidate with Search Console data; change the topic; backlinks | | impressions with no clicks (CTR < 0.5%) | weak title/description | title with a specific promise and a number; measure for 2 weeks | | position 11–30 stalled | content ok, low authority | external links to the winners; new sections for the queries that showed up | | overall drop after publishing a lot | signal of content at scale | stop, audit with the checklist, merge/remove the weak ones, lower cadence |
Every <Tuesday> <9am>: non-branded export + inspection of the key URLs + 1 decision (30 min) Every day <1>: battery of <25> questions in <4> assistants (1 h) Day <30> and day <60> after publishing: read the symptom and apply Plan B Every <Friday>: check that the week went into the archive folder
# import the weekly export folder (re-importing does not duplicate) python3 gsc_historico.py importar <weekly-gsc-folder> --db dados/gsc.sqlite # report for one site: current week × previous week python3 gsc_historico.py relatorio --db dados/gsc.sqlite --site <seusite.com> --saida saida/<seusite.com>
| date | assistant | question | cited the site? | address | who showed up |
|---|---|---|---|---|---|
| 01/11 | ChatGPT | how to choose clay for beginners | no | - | store X |
| 01/11 | Perplexity | how to choose clay for beginners | yes | /argila-iniciantes/ | - |
Made-up example. At INEMA, the July/2026 round gave 0 out of 48 neutral answers on Claude: zero is also a number worth keeping.
Simulated conversation for practice. Pick the answer you would give; the comment appears right after.
Choose your answer. The comment appears right after, and the conversation goes on based on your choice.
Do it once, and it lasts the whole year.
Put the template's reminders on your calendar: weekly, monthly, day 30 and day 60 after your next publication.
Done when: your calendar has 4 recurring or dated reminders, each with the task written out.
Create the archive folder with a backup and save the most recent week in it (summary and non-branded file, or the spreadsheet).
Done when: the folder exists, has the 1st week saved with the date in the name, and is in a place with a backup copy.
The explanation appears after your answer.
Question 1 of 4
The pages are indexed, have impressions, and are stuck between positions 11 and 30. What is the Plan B response?skill
Question 2 of 4
Why archive the weekly export if the dashboard already shows the numbers?
Question 3 of 4
In the monthly battery, why are the questions fixed?
Question 4 of 4
A general drop right after publishing many pages. What should you do?
Pick an answer. The explanation shows up after you try.
Four steps and the exercises for each one. Click an exercise to see its "done when".
export and inspection (lessons 2 and 3)
2 exercises
Two weeks logged and key URLs inspected.
View the lessontitle, topic list, language, goal (lessons 4 to 7)
2 exercises
One measured change, a validated topic, the language decision, and the goal connected.
View the lessoncalendar and archive
2 exercises
Routine on your calendar and the first week archived.
View the lessonBefore any decision, the starting number. It is what you compare against.
Needs the groundwork. The map shows the lesson 2 exercises; the plan below lists them all.
Needs two weeks of data. The map shows the lesson 5 exercises; the plan below lists them all.
Needs the decisions. From here on it's routine: one decision per week, the battery once a month, and Plan B on day 30 to 60.
Foundation, routine, decision, and Plan B. Only move up a step when the "done when" of the previous one is true.
After measurement comes course 4: getting cited by AIs and measuring that in the official dashboards.
Pick the start date and how much time you have per week. The plan respects the exercises that call for waiting.