River
Y CombinatorBacked by Y Combinator
FREE TEMPLATE

Documentation Effectiveness Report Template

Two documents and three sheets that credit a ticket drop to a page only once forum mentions of that topic rule out a channel switch.

Free download  ·  No account needed

Report Narrative

[Company], [n] zero-result query clusters this quarter

Checked against [n] weeks of ticket and forum data before any of it counts as deflection

The number a dashboard would hand you

Clusters with a page shipped against them
Company-wide weekly tickets, before & after
Confirmed only once a shipped topic's weekly tickets fall 30% or more with no matching rise in forum mentions of the same topic.

What the confirmed cohort is worth

Every dollar below traces to the topics that passed this check, never to every page that happened to ship.

A quarter of documentation teams track no metrics at all, and most of the rest default to page views or a raw before-and-after ticket count, according to GitBook's 2026 State of Docs survey of over a thousand practitioners. Zero-result search queries are a sharper starting signal: Zendesk's own self-service metrics guide tracks searches that return nothing as a distinct number from searches nobody clicked, because a missing page and an unfindable one need different fixes. Neither number alone says whether a shipped page actually worked.

This pack's Search Failure Register ranks those zero-result clusters by volume, and the Support Deflection Analysis joins every page shipped against one to its own ticket-tagged topic before crediting any ticket saved. On the worked example, Larkspire Cloud, an invented SaaS company, 46 clusters were logged this quarter and pages went out against 22 of them. Credited without a second check, all 22 count: company tickets on those topics fell from 220 to 94 a week. Run against the confirmation rule in the Method Note, only 14 of the 22 clear it.

Those 14 confirmed topics fell from 174 to 52 tickets a week, a 70.1 percent drop, and account for 122 of the company's 172-ticket weekly decline, 70.9 percent. That is worth roughly $53,924 for the 13-week quarter at Larkspire's own $34-per-ticket support cost. The other 8 shipped pages closed a real search gap without moving enough ticket volume to confirm, not a wrong one. Deciding where a newly written page belongs is the information architecture and navigation pack; keeping one that's already live from going stale is the internal knowledge base pack.

46 zero-result clusters this quarter, 22 pages shipped against them, and only 14 confirmed by a check the other 8 don't clear

The Search Failure Register, the Support Deflection Analysis and the Page Performance sheet.

Search Failure Register

Larkspire Cloud, an invented SaaS company. 46 zero-result query clusters this quarter, ranked by weekly search volume. Top rows shown.

Query clusterSearches/wkStatus
Migrating from the v2 to the v3 API61Shipped
Bulk CSV import validation errors54Shipped
Exporting audit logs to a SIEM47Shipped
Connecting a private VPC peering link42Open
Rotating API keys without downtime39Shipped

22 of 46 clusters got a page this quarter, 47.8%. The other 24 stay open, one held for a platform-team decision, the rest queued behind bigger clusters.

Support Deflection Analysis

Every shipped page's topic, checked against a 30% ticket-volume drop and no more than a 10% rise in forum mentions of the same topic.

TopicTickets/wkForum/wkConfirmed
Migrating v2 to v3 API26→96→7Yes
Bulk CSV import errors21→65→5Yes
Overage calculation, pricing page7→62→9No
Viewer vs. commenter roles6→52→2No

14 of 22 shipped topics confirmed, 63.6%. Tickets on those 14 fell from 174 to 52 a week; the pricing-overage row shows why the forum check matters, mentions rose from 2 to 9 the same week tickets barely moved.

Page Performance

Standard page-level signals, kept as context. Ranked last of the three sheets on purpose, see the Method Note for why.

PageViews/moAnswered bandDeflection
Migrating v2 to v3 API guide64031%Confirmed
SIEM audit log export guide38824%Confirmed
Overage calculation explainer24035%Unconfirmed

A high answered-band share on a short reference page and a low one on a confusing narrative page can look identical here. Confirmed status comes from the other two sheets, never from this one alone.

What's in the pack

01

Search Failure Register

Every zero-result query cluster ranked by search volume, its status shipped or open, and which page it now resolves to once shipped.

02

Support Deflection Analysis

Every shipped page's ticket-tagged topic checked against the 30 percent volume drop and the forum-mention ceiling, confirmed or not, row by row.

03

Page Performance

Standard page-level signals, monthly views and the answered-band share, kept as supporting context rather than the number a headcount case runs on.

04

Report Narrative

The manager-facing number: what shipped, what's confirmed, what the confirmed cohort is worth, and how much of the backlog is left at this pace.

05

Method Note

The exact confirmation rule, stated in full, so a skeptical reader can re-run the check against your own numbers instead of taking the total on faith.

06

A space rule every prompt reads first

A ticket-volume drop only counts as deflection once forum mentions of the same topic are checked and haven't risen to match it.

How to use it

  1. 1

    Open it in River, or download it

    Open the pack and the agent reads your analytics, search log and ticket export before ranking a single cluster, or download the blank sheets and docs.

  2. 2

    Send your logs

    Your page analytics export, your on-site search query log with result counts, and a support ticket export tagged by topic if you already have one.

  3. 3

    Rank gaps, then join tickets

    Zero-result queries rank by volume into the Search Failure Register, then every shipped page's topic is checked against a ticket drop and a forum-mention ceiling together.

  4. 4

    Get the confirmed number

    The Report Narrative states what shipped, what's confirmed, what the confirmed cohort is worth in dollars, and how much of the backlog is left at this pace.

Frequently asked questions

Is this template free?

Yes. Download the whole pack as Word documents and CSV sheets, no signup and no credit card. Edit with AI is a separate, optional path for anyone who wants the agent to join your own logs and check them against the confirmation rule. Nothing happens until you send it something to join.

What counts as a confirmed ticket deflection?

A shipped page's topic only counts once its tagged ticket volume falls 30 percent or more against the 30 days before shipping, with forum mentions of that same topic rising no more than 10 percent over the same window. On the worked example, that check passed for 14 of the 22 pages shipped this quarter, not all 22.

What if we don't have a community forum or any usage data yet?

Build the Search Failure Register and the Page Performance sheet from your analytics and search logs alone, and leave the Support Deflection Analysis unbuilt rather than guessing at an outcome. Add the ticket and forum exports the moment you have a real window of data to check against, not before.

Isn't a ticket-volume drop after a page ships already good enough evidence?

Not on its own. A drop can mean the page worked, or it can mean the searcher gave up and asked in a community forum, a Slack channel, or a colleague instead of filing a ticket at all. The forum-mention check is what tells those two apart before either one gets credited.

Does this check whether an existing page's content is accurate, not just whether it worked?

No, that is a separate question with a separate fix. This pack checks a shipped page's ticket outcome against the confirmation rule; checking an already-published page's claims against your product's current behavior is the documentation audit.

What format are the downloaded files?

Two Word documents and three CSV sheets, zipped. The documents open in Word, Pages and Google Docs; the sheets open in Excel, Numbers and Sheets. Add a PDF query string to the download if you want to circulate the narrative rather than fill it in yourself.

Does this work the same way for a tutorial or an internal runbook, not just a reference page?

Yes. The join only needs a ticket tagged to a topic and a second signal for that same topic, so it applies to a tutorial the same way it applies to a reference page or an internal wiki entry. What differs is the source of that second signal: a public community forum for customer-facing docs, an internal channel for an internal wiki.

Find out which of this quarter's shipped pages actually earned their ticket drop

Download the blank Search Failure Register, Support Deflection Analysis and Page Performance sheet as Word and CSV, or have River check your own logs against the confirmation rule.

Edit with AI