Direct Channel Decline Diagnosis (2026-08-07)

This is a site-ops research note for the Direct acquisition channel on monsterlabsimulator.wiki. It uses only the Direct comparison object supplied on the action card. It does not invent campaign UTMs, bookmark cohorts, or page-level Direct landings that were not provided.

Snapshot metrics (provided Direct window)

Metric	Current	Previous	Absolute change	Relative change
Sessions	31	122	-91	-74.59%
Key events	0	0	0	0%
Engagement rate	0.1613	0.0328	+0.1285	+391.94%
Bounce rate	0.8387	0.9672	-0.1285	-13.29%

Raw comparison object (unchanged from the action card):

{"sessions":{"current":31,"previous":122,"change":-91,"changePct":-74.59,"change_pct":-74.59},"keyEvents":{"current":0,"previous":0,"change":0,"changePct":0,"change_pct":0},"engagementRate":{"current":0.1613,"previous":0.0328,"change":0.1285,"changePct":391.9355,"change_pct":391.9355},"bounceRate":{"current":0.8387,"previous":0.9672,"change":-0.1285,"changePct":-13.2859,"change_pct":-13.2859}}

Why Direct matters to the site-wide story

From the site-wide GA4 diagnosis:

Slice	Absolute session Δ	Share of site-wide −145
Direct (this card)	-91	≈ 62.8%
Organic Search (sibling card)	-53	≈ 36.6%
Residual after both	-1	≈ 0.7%

Direct is the largest absolute hole among the provided channel slices. SEO-only work cannot close a −91 Direct gap.

Does Direct loss prove conversion damage?

Question	Answer from provided numbers
Did Direct sessions fall hard?	Yes (−74.59%)
Did key events fall?	Not measurable — 0 → 0
Did remaining users engage worse?	No — engagement rate rose (0.0328 → 0.1613)
Did bounce get worse?	No — bounce fell (0.9672 → 0.8387)

One-line root framing

Direct volume collapsed; the remaining Direct sessions look higher quality; conversion impact is unknown because key events are zero in both windows.

That pattern fits “fewer untracked / habitual / dark-social / bookmark entries still reach the site,” or “low-intent Direct noise left,” better than “the site suddenly became unusable for Direct users.”

What kind of problem is this?

Hypothesis	Supported?	Why
Direct UX totally broken for everyone	Unlikely from these numbers	Engagement up, bounce down among remaining sessions
Core conversion path destroyed on Direct	Not proven	keyEvents 0 / 0 — either no events configured or none fired in either window
Volume of Direct entry points shrank	Yes	Sessions −91
Brand / habit / untagged campaign decay	Plausible, not proven	No UTM or source detail attached
SEO ranking is the Direct cause	No	Direct is not Organic Search

Core pages: what we can and cannot say

No Direct landing-page export was attached. Therefore this note does not rank /tools/breeding, /monsters, /fulus, or home by Direct sessions.

Content still prioritizes protecting the pages that other ops cards already treat as core for the site:

Core surface	Role	Content trust rule
Breeding Calculator	Recipes list/tool	Prefer verified rows
All Monsters list	All-monsters list intent	Verified 20 only
All Fulus catalog	Catalog depth + phantom blocklist	Verified 20 only
Monster Database	Interactive DB UI	Eng: verified-default still pending
Empower guide	Mid-game rarity	Not recipes list
Pet Care	Care / 养成	Not all-monsters SERP

Until a Direct landing export exists, do not claim which of these caused the −91.

Action map

Priority	Action	Owner
P0	Confirm GA4 key events are defined and firing	Analytics / eng
P1	Pull Direct landing page + new vs returning for the same window	Analytics
P1	Audit recent changes that could drop untagged traffic (domain, redirects, app links, social dark links without UTM)	Ops / eng
P2	Keep catalog trust high so returning Direct users who do land still see verified 20	Content: All Monsters, Verified policy
P2	Do not “fix Direct” by stuffing SEO keywords into care pages	Content duty split on All Monsters

Re-measure checklist

Same window length; export Direct sessions again.
Require a landing-page table before assigning page blame.
If keyEvents remain 0 site-wide, fix instrumentation before talking about conversion recovery.
Success is session recovery toward the previous 122, not bounce-rate cosmetics alone (bounce already improved).