1. What user satisfaction actually is
Google, Bing and every LLM ranking system converge on the same working definition: did the user complete the task they came to do, on this page, in one visit. Everything else — dwell time, scroll depth, click-back, follow-up query — is a proxy for that one question.
The failure mode for programmatic sites is subtle: pages rank because they match the query on the surface, users land, immediately realise the page is a shell, and bounce. The ranking system counts the bounce and quietly demotes the whole template.
2. The five proxy metrics
Dwell time — 25%
Time from click-in to click-out. Compared against category expected, not global average.
Interaction rate — 25%
Clicks, form starts, copies, expansions, video plays. Measures active engagement, not passive scrolling.
Task completion — 20%
Did the user do the thing the page promised — filled the form, copied the code, downloaded the file, added to cart.
Scroll depth — 15%
How far down the page users read. Lower weight because it's easily gamed by long intros.
Return rate — 15%
Users coming back to the same URL later. Bookmark-worthy content earns this signal over time.
3. Low-satisfaction workflow (4 weeks)
- 1
Week 1 — Diagnose
Analytics flags URLs with a composite score below 40. Drill into which of the five metrics is the outlier — dwell alone means intent mismatch, interaction alone means missing CTA.
- 2
Week 2 — Intervene
Fix at the template layer where possible. Rewrite the intro to match query intent, move the primary action above the fold, cut boilerplate that inflates dwell without helping.
- 3
Week 3 — Republish + tag
Push the fixed template, tag the affected URLs for satisfaction re-measurement. Do not touch pages the fix does not affect.
- 4
Week 4 — Re-measure
Compare the new composite score against the pre-intervention baseline. If it hasn't moved, the intervention was wrong — do not chain a second fix on top.
4. What we don't do
- We do not use satisfaction scores as pre-publish gates. They need real user data first — that means post-publish, not template-time.
- We do not compare satisfaction across query classes. A definitional page's dwell target is different from a buying guide's.
- We do not treat pogo-sticking as satisfaction data on its own — small samples are noisy.
- We do not sell 'satisfaction guarantees' on rankings. Satisfaction feeds ranking; ranking is not a satisfaction guarantee.
5. How WBP measures it
WBP measures template-level satisfaction weekly and surfaces the worst-performing templates on your dashboard. A template with a pogo-stick rate above cohort average gets an amber flag; two consecutive weeks red and the template is auto-paused until the underlying page contract is fixed.
Combined with the Bulk Publisher's page contract system, this means bad templates cannot silently poison your whole site — they fail fast, on a small slice, and get pulled before they scale.