Operator claim
Words, logos or numbers found on an operator-controlled page. Useful for monitoring, but not treated as independently true.
Methodology 2.0 · effective 26 July 2026
How the monitor works
The system detects public-page changes every day. It does not turn every detected word into a fact, every fact into a score, or every score into a recommendation.
Core distinction
Observation is not verification.
If a page displays “licensed,” “fast withdrawal” or a local payment name, the monitor records that as an operator claim. It does not copy the language into editorial prose as a proven fact.
Each material evidence item carries a source, first-seen and last-seen time, confidence state and review status. Hands-on testing is dated separately from automated monitoring.
Evidence ladder
Confidence must be earned.
The score engine reduces the influence of categories supported only by claims or thin observations. Missing evidence cannot receive full credit.
Observed
A dated public-page snapshot with source URL, response status and content hash. It proves what was visible at that time.
Corroborated
The same material fact is supported by an independent record or a separate reliable source.
Verified
Matched to an authoritative source or documented through a lawful, reproducible editorial test.
Scoring framework
Six categories, with safety first.
The weighted result is evidence-adjusted and can receive risk penalties. A high marketing-quality score cannot overcome a failed identity or licence gate.
25%
Identity & licence
Legal entity, ownership, domain lineage and regulator-source verification.
25%
Payments & withdrawals
Published limits, fees, KYC timing and evidence from a lawful test journey.
20%
Bonus & core terms
Wagering rules, exclusions, maximum cash-out and conflicting language.
10%
Bangladesh relevance
Language, currency and market claims—recorded as claims until corroborated.
10%
Support experience
Channel availability, answer quality and whether replies match written terms.
10%
Risk controls
Age gates, limits, time-outs, self-exclusion and independent help routes.
Published resultweighted category score × evidence confidence − risk penalties
Search attention methodology
Relative attention, not market size.
The Bangladesh Google Search Attention Index measures relative Google Web Search interest in Bangladesh (geo=BD) from 26 July 2025 through 25 July 2026. It is not absolute search volume, revenue, users, safety or a recommendation.
Fixed inputs
One query manifest for every refresh.
Each tracked brand uses a documented, fixed query. Common anchor queries connect comparison batches so that the normalized results can be placed on one relative scale. The strongest observed brand in the baseline is indexed to 100; the remaining values are relative to it.
Known limits
Zero and ambiguity need context.
A low-volume value of 0 means the sampled interest was below Google Trends' reporting threshold, not that nobody searched. Queries that may refer to another product, phrase or spelling receive lower confidence and are not presented as precise brand demand.
Refresh control
Monthly data, manually reviewed.
The attention index is refreshed monthly, or rebaselined when a material trend shift makes the old comparison misleading. Publication uses manually reviewed official CSV exports while the official Trends API remains in alpha. Daily page monitoring does not silently alter the attention index.
Crawler boundaries
What the automation may do.
- Fetch curated HTTPS public pages at a limited rate.
- Respect robots.txt and stop on access restrictions.
- Record redirects, structured facts and content hashes.
- Store private snapshots for a traceable comparison.
Prohibited actions
What it may never do.
- Bypass a CAPTCHA, login, block or age gate.
- Create an account, submit identity documents or install an APK.
- Deposit, place a wager or test a withdrawal automatically.
- Scrape Google search-result pages or invent source evidence.
Three clocks
Dates that mean different things.
The latest successful automated public-page check.
The most recent human-approved change to a conclusion or key fact.
The date of the most recent lawful editorial product test, if one exists.
Daily checking is not daily publishing.
Public pages are monitored daily. Minor factual changes are batched, page-noise changes are ignored, and sitemap dates move only after a meaningful editorial update. Search attention is refreshed monthly unless a material trend requires a new baseline.
Material changes wait for review.
Critical and major changes—including ownership, licence, redirects, payments, withdrawals and core terms—enter a manual verification queue before they can affect the public record or a score.
Read the exact thresholds ↗