AISEOCourse.academy Module 8

Module 08 · Lesson 3 of 4

Building Repeatable Workflows

About 11 minutesPrerequisite: Lesson 8.2
After this lesson you canwrite a working SOP for each ritual of the system, hand any of them to a colleague or client without a meeting, and scale the method across sites without diluting it.

A method that lives in one person's head is a liability with a holiday schedule. A workflow is the method written down so precisely that a competent stranger can execute it and get your results. Writing them is an afternoon. It is also the difference between running AI SEO and being someone who once ran AI SEO.

The three rituals, as documents

Your whole system compresses to three standard operating procedures:

RitualCadenceThe SOP must specify
The runWeekly, fixed slotExact conditions, the query set location, every column's vocabulary, the diff step, time budget
The roll-upMonthly, one hourThe counts to compute, the three interpretation sentences, the caveat sentence, where the report goes
The reviewQuarterly, half a dayThreshold check from lesson 7.3, opportunity re-ranking, update-or-merge decisions, automation spot-check

Write each SOP in the second person, in imperative steps, with the time budget stated. "Run the set" is not a step. "Open the saved session, run queries 1-30 in order, record surface, mention type, context label per the vocabulary sheet" is a step. If you cannot write the step that precisely, you have found the part of your own process that secretly depends on improvisation, and that is exactly where your numbers wobble.

DefinitionAn SOP is a written procedure precise enough that a competent stranger reproduces your result. The stranger test is the standard: hand it over, say nothing, watch what they stumble on, fix that paragraph.

Scaling without diluting

Agencies and in-house teams with multiple sites face the temptation to scale by cutting the method short: fewer queries, no roll-up, dashboards only. The honest way to scale is to scale the rituals, not trim them. Per site: a 20-30 query set (lesson 3.1's floor and ceiling), the weekly run, the monthly roll-up batched across sites in one sitting, and one shared opportunity queue reviewed together. What does not scale: bespoke interpretation per site. Budget real thinking time for the sites that matter commercially; give the long tail the rituals at maintenance depth. A site either gets the real method or an honest decision that it gets less; what it never gets is a dashboard pretending to be the method.

Handover is the test

Once the SOPs exist, the handover test is free: give the run SOP to a colleague for two weeks. Their rows will diverge from yours somewhere: a label used differently, a surface interpreted differently. Every divergence is a precision failure in the document, not a failure of the colleague. Fix the document until the divergence stops. That converged document is now an asset: it trains new team members, survives your holiday, and is half of what a client is buying if you sell this service.

Worked example: the run SOP, first page (illustrative)
  1. Friday 09:30, one-hour block. Open the dedicated browser profile (link).
  2. Run queries 1-30 from the set sheet, in order, new chat per prompt. Record per row: date, run number, surface, mention type, context label (vocabulary sheet tab 2), cited domains, competitors present, classic position, answer-changed yes or no.
  3. Run the diff (link). Read flagged answers yourself.
  4. Anything crossing a lesson 7.3 threshold: one investigation note, maximum one, in the notes column.
  5. Stop at the hour. Unfinished rows are finished Monday; the log records the gap honestly.
Workbench 8.3
  1. Write the run SOP this week, in imperative steps, with the stranger test in mind.
  2. Write the roll-up and review SOPs at outline depth now; finish them before the next occurrence of each.
  3. If anyone works with you: schedule the two-week handover test of the run SOP.
  4. Add the SOP links to your log's front sheet, so the documents and the data never separate.
Self-check
What is the stranger test?
Hand the SOP to a competent person with no verbal briefing. Every stumble marks an under-specified step. The document is done when their output matches yours.
How do you scale the method across many sites honestly?
Scale the rituals (each site keeps a real query set, run and roll-up) and budget deep interpretation only for commercially important sites. Never substitute a dashboard for the method anywhere.
Your colleague's logged rows differ from yours. Whose fault is it?
The document's. Divergence means a step was not precise enough; fix the SOP rather than blaming the reader, and re-run the test.
Key principleWrite the system down until a stranger can run it: a method that survives handover is an asset; one that does not is a liability.

Sources used in this lesson
Google Search Central: AI features and your website

Previous 8.2 Automation Without Nonsense