The cards on the scrapbook point here when a short observation needs more room. These notes add source context to the main-site guides.
Published by Ryan Lenk · Last reviewed August 14, 2026 · A material change receives a dated correction note. A withdrawn note stays listed with its date and reason.
Source note. X parameter file published August 13, 2026
Scrolling past you is a scored penalty.
X published production-default coefficients for 26 scoring heads in its open ranking repository. The file assigns NotDwelledWeight a value of -0.02. That term applies when a viewer moves past a post without dwelling on it.
The score adds weight x predicted probability for each action. A large coefficient can belong to a rare action. Comparing coefficients alone cannot show that one copied link is worth 40 likes because each action enters the calculation with its own predicted probability.
Ryan's visual summary of the published coefficients and the probability-weighted calculation.Bangermeter applies the published weights to visible post signals and collects no account data.
What you can use
The practical job is to stop the scroll before asking for a reply, share, or click. Ryan built Bangermeter to show the same weights against a post. The tool estimates probabilities from visible signals and does not predict final reach.
Evidence record. Clarity capture: June 10 to July 28, 2026
114 citations in one Clarity view.
Microsoft Clarity's My cited pages list shows https://kraftybandit.com/pages/how-to-care-for-a-handmade-quilt with 114 citations. The reporting view covers June 10 through July 28, 2026.
Source capture supplied by Ryan. The page-specific citation count is visible in the My cited pages panel.
What the record shows
The view also shows 157 citations across queries. AI referral traffic was below 1%, defined by Clarity as AI-referred sessions divided by total sessions for the selected period. Visible grounding queries include how to safely wash a handmade personalized large quilt, handmade cotton quilt care instructions, and handmade quilt long-term preservation care.
The screenshot names its data source as Microsoft Copilot and partners. That is Clarity's vendor-defined coverage, not all AI systems.
One owned store and one reporting view. This is not proof that the page caused traffic, revenue, ranking, prominence, or a general rule. It is a dated citation record with a named page and visible query context.
A title tag, page heading, and visible offer do separate jobs. A visitor still reads them as one opening statement. When they point at different things, the page asks the visitor to solve a problem before the page has earned their attention.
Start with the first screen. Say what the business sells, who it is for, and the next useful action. Then check whether the title and description support the same answer. The page can still have a slogan. The plain line needs to arrive first.
Structured data helps search engines understand visible content. It can identify a person, organization, page, or question and answer when the same information is already present on the page.
Put the real work on the page first. Then add matching schema. Keep the two in sync when the offer, author, dates, or answers change.
A small card earns a separate URL when it answers a narrow question with enough care that a reader can use it. The card can set up the point. This page can provide the reasoning and source. The full guide stays on the main site if that is where the durable explanation belongs.
That division keeps the scrapbook from becoming a duplicate. It also gives each link a real job instead of using a page as a hallway.
If someone outside the business cannot repeat the offer after the first paragraph, the page needs a clearer sentence. A slogan can make a good second line. It cannot carry the entire explanation alone.
Try the plain version before changing the markup. The page may already have all the technical pieces it needs. The missing part may be a sentence that names the product, customer, or service directly.
A number needs its starting point, time window, and condition. Without those details, a clean result can create a false impression. The reader should be able to find the source and decide what the result means for their own site.
That is useful for humans and answer engines. It also makes a future update easier, because the earlier claim has a clear receipt attached to it.
