How Zaagan works
Every Zaagan engagement answers these three questions, again and again. This page follows one illustrative decision through the whole system—from the first signal, through expert review, to a measured result and the lesson it leaves behind.
Before the first question
A domain, Search Console, existing analytics and a short conversation about the business are enough to begin. Each extra source unlocks new questions—and anything the evidence can’t yet support is reported as exactly that.
Coverage explorer
Questions Zaagan can answer
5 of 9
Has an important page broken, redirected or dropped out of the index?
AnswerableWhich valuable topics are gaining or losing search visibility?
AnswerableWhere are you seen in search, but not chosen?
AnswerableWhich landing pages attract visitors but rarely lead to an enquiry?
AnswerableWhich pages convert well but are hard to find?
AnswerableWhere exactly do buyers abandon a quote, booking or checkout?
Needs Zaagan event tag
Is search growth turning into qualified pipeline?
Needs CRM or commerce data
Which pages bring valuable customers, not just form fills?
Needs CRM or commerce data
What do buyers ask that the website doesn’t answer?
Needs AI Lead Agent
Measurement depth
Evidence reaches Conversions
“Not enough evidence” is never reported as “no issue”. When a question can’t be answered reliably, you’re told—rather than handed a clean bill of health.
01–02 · Inside one decision
Every recommendation is assembled from the same distinct parts. Keeping them apart is what makes the advice inspectable: you can see what the data says, what is inferred from it and what is proposed as a result.
01What should change next?
01Evidence
Search, analytics and crawl data become compact evidence records. Each keeps its source, metric definition, comparison window and data quality, so every later claim can point back to exactly what supports it.
Evidence records
EV-871 – EV-878
Last 28 days vs previous 28 days
02Observation
Detectors look for material, commercially relevant patterns—not every wobble in a chart. What they produce is a plain factual statement with a confidence level. Interpretation and advice wait their turn.
Observation
OB-124
Detector · Search CTR deterioration
Organic click-through fell from 3.9% to 3.2% while average position stayed broadly stable.
Detector outcome
03Opportunity
Several detectors can point at the same underlying problem. Grouping them into a single opportunity, weighted by commercial significance, is what stops the service from becoming another alert feed.
Opportunity
OP-19
Recover the search appeal of the commercial solar PV page.
Built from 4 observations across 3 sources
02Why that change?
04Interpretation
AI helps draft the reasoning from a bounded evidence packet and must cite evidence for every material claim. Alternative explanations stay on the table until the evidence rules them out.
Interpretation · hypothesis
IN-07
Leading hypothesis
The new title may be less compelling for high-intent commercial searches.
Alternative explanations
05Challenge & review
A second pass challenges causal overreach, thin samples and effort out of proportion to the evidence. Then an expert approves, edits, defers or rejects. Nothing reaches a client without that sign-off.
Recommendation · review trail
RC-19
AI candidate · draft
Build a new landing page for rooftop solar on warehouses.
Critic pass · challenged
Expert reviewEdited & approved
Revise the page title and opening proposition around commercial buyer intent. Keep the existing page.
Expected mechanism: A clearer commercial proposition should convert more of the existing visibility into qualified visits.
Scored separately—never collapsed into one misleading number.
03 · Did it work?
The measurement plan is agreed before anything ships, so success can’t be redefined after the fact. The result is then allowed to mature at the pace of the client’s buying cycle.
Change record
CH-31
Shipped by the client’s web team
Recorded as it actually shipped—which isn’t always exactly what was recommended.
Measurement plan · agreed before launch
The primary metric follows the mechanism—the thing the change can most directly move—not the most impressive number available.
Evaluation timeline
Baseline
28 days before
Change ships
Day 0
Warming
7 days
Week 1Warming up
Too early to judge. Nothing is reported as a result.
Evaluation
6 weeks
Week 6Positive · acquisition
Click-through and organic sessions moved as expected. Position held.
Commercial follow-through
One sales cycle
Month 4Positive · strengthening
Qualified enquiries rose. Still associated, not attributed.
Revenue still matters most. It is reported as commercial context—not claimed as Zaagan’s doing.
The change
Title change
Very close
Search click-through
Primary
Close
Organic sessions
Secondary
Further
Quote submissions
Secondary
Further again
Qualified opportunities
Context
Many other factors
Won revenue
Context
Observed: a claim we would make
“Qualified enquiries rose during the evaluation window.”
Associated: a claim we would make
“The change was associated with the rise in qualified enquiries.”
Attributed: a claim we would not make
“The change caused the rise in qualified enquiries.”
Needs a controlled test or equivalent evidence.
Every outcome is recorded as a finding—including the unflattering ones.
Retained learning
LE-07
On this client’s service pages, a clearer commercial proposition in search results has been associated with stronger organic acquisition, without any loss of position.
Kept after the raw data expires. It becomes context for the next decision.
Back to the start of the loopThen it runs again
The system watches every day so people don’t have to. Machine activity is compressed into a short expert review queue, and you hear about what deserves a decision—not every metric that moved.
Continuously
The system
Syncs search, analytics and crawl data, runs detectors and advances measurement windows.
Weekly
Expert review
Prioritises, approves recommendations and reviews matured results.
Monthly
Your update
What changed, what was learned, what’s measuring and what’s next.
Quarterly
Strategy
Direction, themes and the shape of the programme.
Evidence healthy, changes still measuring, nothing new deserving action. Activity is never manufactured to fill a report.
A key page returning a 404 goes straight to expert review, then to you. Not an automated false alarm.
Updates cover what changed, what was learned and what to do next—including the limits of the evidence.
Practical questions
A website domain, access to Search Console and existing analytics where available, plus a focused conversation about customers, valuable outcomes and current constraints. Zaagan can begin without a CRM, warehouse or bespoke tracking project.
No. Zaagan Intelligence can provide the evidence, prioritisation and recommendations for your internal team to implement. If you need delivery support, a managed service can take responsibility for agreed technical, conversion and content improvements.
It depends on the change and the buying cycle. An ecommerce checkout improvement may produce a useful signal quickly; a specialist B2B service page may need several months. The evaluation window is defined before implementation so short-term noise is not mistaken for a result.
No. Some work is corrective and only needs technical verification, some is measurable before and after, and some can support a controlled experiment. Zaagan uses the strongest practical evaluation method without pretending every change can be cleanly attributed.
AI helps draft interpretations and recommendations from a bounded set of evidence, and a second pass challenges them. It never decides whether something happened in the data—that is established by deterministic analysis—and no AI-drafted recommendation reaches you without expert review.
Start with evidence
The Assessment is where this system begins for your business. It establishes the evidence base, the site model and your first prioritised opportunities—and if you continue with Zaagan, nothing restarts.
