The AwesomeSEO method: from what you know to what the engines say.
Six steps that turn a brand’s own context and data into recommendations with the honest case for each, actions a person approves, and monitoring that separates real change from noise. Simple on the surface; every detail one click away.
Six steps, one loop
Monitoring feeds the next round of analysis, so the loop runs weekly for as long as the workspace is open.
Step 1 of 6
Context
Onboarding captures what only you know: who the brand is, what it sells, to whom, where, and against whom.
- One central entity and one source context frame every page and every prompt
- Markets, languages, keyword seeds and competitors come from you, not from a guess
- Nothing about your customers is asked for; the workspace holds no personal data
Detail, on demand
The brand, its domain and platform, the markets and languages it sells in, what it sells in its own words, who it competes with, the keyword seeds it already cares about and the page types it runs. From these the workspace derives its first prompt set, keyword clusters and site scaffold deterministically, so two people entering the same brand get the same plan.
Before the first page, decide who the site is: one central entity and one perspective. Everything measured later is measured against that context, and drift is flagged before it publishes.
documented Define the source context before the first page
Step 2 of 6
Data
Your own sources first, then the market: the more relevant data connected, the better the recommendations.
- First-party: Search Console, GA4, Shopify, Merchant Center, Bing Webmaster
- Market: live top-20 SERPs and keyword data through DataForSEO, our own crawler
- AI answers: seven engines, every tracked prompt five times a week
- Google Search Console
- Google Analytics 4
- Shopify
- DataForSEO
- Bing Webmaster Tools
- Google Merchant Center
- AwesomeSEO crawler
- ChatGPT
- Gemini
- Perplexity
- Claude
- Copilot
- AI Overviews
- AI Mode
Detail, on demand
- Google Search Console
- Queries, positions, clicks and indexing for your own pages, 16 months back.
- Google Analytics 4
- Engagement and revenue per landing page, plus visits referred by AI engines.
- Shopify
- Products, collections, pages and SEO fields, read live from the store.
- DataForSEO
- Live top-20 SERPs, keyword volumes, backlinks and AI answer samples.
- Bing Webmaster Tools
- Bing queries, crawl errors and IndexNow submission state.
- Google Merchant Center
- Product feed status: disapprovals, missing GTINs, price and availability mismatches.
- AwesomeSEO crawler
- Every page as search and AI crawlers read it: HTML, schema, response times, robots.
Each figure in the workspace is stored with where it came from, when it was fetched and how often it refreshes, and the interface shows that line beside the number. A stale figure says so instead of looking current.
Step 3 of 6
Analysis
Competition and niche, keywords and SERPs, the brand, the crawl and the history, with the uncertainty stated.
- The top 20 for every money keyword, compared with your page signal by signal
- Brand monitoring: mentions, reviews, branded search and links in proportion
- Every rate with a 95% interval and its n; a change is a change only when the test says so
Detail, on demand
An answer engine is not a ranking: ask twice and you get two answers. Named in 3 of 5 runs reads as 60% with a 95% interval of 23%–88%, which is why prompts are pooled before anyone acts. The same rate over 175 runs narrows to 53%–67%.
A move from 51% to 63% on 35 runs a week looks like progress and is marked n.s. (p = 0.33). The same difference held over eight weeks is significant (p = 0.01). The product shows both the direction and the test, so a weekly wobble never becomes a decision.
Inside the top 20, pages Google already judged relevant compete on quality. For every money keyword SERP Lab keeps a standing study of those results: who ranks, what they cover and what is left to add. Your page is compared metric by metric, and the aim is the podium on many signals, never an outlier on one.
practitioner Treat the top 20 as the competitive set and watch it for yearspractitioner Be top three on many metrics, never an outlier on one
Referring domains are tracked per 100 branded searches against your niche, with mentions, reviews and share of search beside them. Growth that stays in proportion is the goal; a spike in one line alone is a warning, not a win.
documented Grow branded search alongside links
Step 4 of 6
Recommendations
Probabilistic, with the honest case for each: what we expect, how sure we are, and the evidence tier.
- Expected effect as a range, never a promise; effort and owner suggested
- Every practice carries one of 7 evidence tiers, from confirmed to excluded
- A correlation is called a correlation; a practitioner heuristic is tested on your pages first
Detail, on demand
- Google states it in documentation or official guidance.
- 18 practices
- The signal exists per the 2024 API leak, trial testimony or a Google patent. Weight unknown.
- 4 practices
- Correlates with position in the TopSEOFactors data. Correlation, not causation.
- 11 practices
- A working heuristic from experienced SEOs. We test it on your own data.
- 49 practices
- A hypothesis. Shown with that label, never as advice.
- 5 practices
- Google removed or stopped using it.
- 9 practices
- Breaks spam policies or our ethics. Never automated.
- 10 practices
The finding with its numbers and their source; the practice it follows and that practice’s tier; the factor and its correlation when the tier is measured; the pages it applies to, grouped by template; an expected effect stated as a range with the assumption behind it; and the effort and the person best placed to own it.
Tactics that break Google’s spam policies, consumer law or plain honesty are listed with their reason and never automated. Retired practices are flagged when found, never recommended. See the guardrails.
Step 5 of 6
Actions
You apply recommendations in part or in full, and can question or edit every one before it ships.
- Agents draft; a named person with the admin role approves, with the diff in front of them
- Edit any draft field by field, or send it back with a question
- Everything that publishes is logged with a name and a time, and can be undone
Detail, on demand
Every finding from every module lands in one register. The list stays short enough to finish, each action links back to its evidence, and completion is tracked like any other metric. Agencies export it as a client action sheet.
A draft shows old and new values side by side with the data it used. Approving publishes through the platform’s own API, records who approved and when, and undo writes the previous value back. AI use is declared on every page an agent touched.
confirmed Declare AI use and keep a human in the loop
Step 6 of 6
Monitoring
After a change ships, you see what moved, and whether the move is significant or noise.
- Changes are tested against matched control pages, not against last week
- Decay alerts fire only when a drop is significant; refresh counts only when the page changed
- A weekly digest states the findings as sentences with the numbers in them
Detail, on demand
A page that changed is compared with pages of the same template and traffic band that did not. Seasonality and core updates hit both; the difference between them is the effect, reported with its interval.
A revision counts when at least 10% of the text changed, with facts and years brought current. Changing only the date misleads readers, so the product never does it for you.
documented Refresh for real: money pages monthly, everything within 90 days
Simple on the surface. Exact underneath.
- Engines sampled
- 7
- Weekly, five runs per prompt
- Evidence tiers
- 7
- From confirmed to excluded
- Practices
- 106
- Each with a tier and a source
- Months of ranking data
- 93
- TopSEOFactors, top 200 factors
What the niche gets wrong
Much SEO advice is a practitioner heuristic presented as fact, or a tactic that worked until the next update. The playbook keeps both kinds listed with their tier, so the product can flag them instead of repeating them.
- retiredFirst Input Delay
- retired"Engagement Reliability" and "Content Engagement Duration" as Core Web Vitals
- retiredSitelinks search box markup
- excludedCTR manipulation services
- excludedHidden headings or text only Google sees
- excludedPBNs, paid links that pass PageRank, tiered link schemes
See the method run on a real workspace.
The demo workspace runs every step on a fictional Swedish linen brand, with example data. Nothing to install.