Skip to content
Get Ranked On TopIllustrative case studyContent Engine plan

A fictional business. The figures are a modelled scenario, not a client result.

Wazen Zakat Desk · Jeddah, Saudi Arabia

How a Jeddah zakat and VAT desk became the source AI answers quote when the rules change

Cited in 46% of tracked AI answers, up from 7%

12 months · Middle East & Africa · Content Engine plan · Zakat, VAT & e-invoicing advisory · 25 advisers

  • Dated guidance hub
  • Visible changelogs
  • Rule-change playbook
  • AI answer tracking

Simplified tax invoice

Study no. WZ-2025-012

Issued by
Wazen Zakat Desk
Period
12 months · 2025
Dated guides published
86
Tracked AI prompts
120
وازن
AI answers citing Wazen7% → 46%

Each module stands for one guide. Once every guide is dated and current, the code scans.

01The challenge

The rule changed. The answers didn't.

When rules changed, AI answers kept quoting old thresholds from forums. The firm's up-to-date guidance sat in undated PDFs that nobody cited.

Wazen's 25 advisers knew every change the day it was announced. Their clients didn't, because clients now ask an AI assistant first, and the assistants answered from what was easiest to quote: years-old forum threads with confident numbers. The firm's own guidance lived in PDFs with no date, no changelog and three different figures for the same threshold. Nothing on it told a machine, or a person, that it was the current version.

Modelled AI answer · month 1

Which businesses must join the next e-invoicing integration wave?

Businesses with taxable revenue above SAR 3M must integrate in the next wave.

Source: a majlis-traders.example thread from 2022

Out of date since 12 May 2025: the threshold is now SAR 1M.

Meanwhile, on the firm's own site

PDF guides
214
with no date anywhere
172
different figures for one threshold
3

02Diagnosis

An invoice that would never pass validation.

Undated PDFs

172/214

214 guidance PDFs, 172 of them with no date in the file, the URL or the text. Assistants can't tell a 2019 note from this month's update, so they skip both.

Finding 1 of 4

Stale answers

61%

Across 120 tracked prompts in Arabic and English, 61% of modelled AI answers quoted a superseded threshold or deadline, usually from forum threads written before the latest change.

Finding 2 of 4

Update lag

31 days

A median of 31 days passed between a rule change and Wazen's guidance reflecting it. Old PDFs were never retired; new ones were simply added next to them.

Finding 3 of 4

Contradictions

3

The firm's own site gave three different figures for the same e-invoicing threshold on three pages. A careful reader, human or machine, had no way to know which one was current.

Finding 4 of 4

Who the modelled AI answers cited · 120 tracked prompts

Month 1

  • Forum threads
  • Outdated news & PDFs
  • Official sources
  • Wazen

03The strategy

Four modules of a code that has to scan.

Built on our Content Engine plan: content that answer engines can find, trust and quote, kept current by a process rather than by heroics.

  1. 01

    Dated guidance hub

    214 PDFs consolidated into 86 HTML guides in Arabic and English, one topic per page, each with a named adviser, a visible "last reviewed" date and Article schema carrying datePublished and dateModified. Old PDF URLs redirect to the guide that replaced them.

  2. 02

    Visible changelogs

    Every guide ends with a dated changelog: what changed, the old figure, the new one and the source. Freshness you can see, written for people first, and the clearest signal an answer engine gets that this page is the current one.

  3. 03

    Rule-change playbook

    A one-page runbook for announcement day: which guides a rule touches (mapped in advance), who updates each, a two-adviser review, then publish, bump dateModified, ping the sitemap and update the internal links. Target: within two working days.

  4. 04

    AI answer tracking

    120 real client questions, in Arabic and English, run every week through the main AI assistants. Each answer is logged by source and by whether its figure is current, so the team sees which guides are cited and which still lose to an old forum post.

04Freshness QR

Pick a rule change. Watch which modules break, then publish the fix.

Each module of the code is a page on the site. When a rule changes, the pages that quote it go stale and the code stops scanning. Updating them, with a new dateModified and a visible changelog, is what moves the modelled AI citation from an old forum thread to Wazen.

Rule change

Won't scan: 3 pages out of date

E-invoicing wave, Before the update. Won't scan: 3 pages out of date

Announced 12 May 2025

WasNext e-invoicing integration wave: taxable revenue above SAR 3M

NowNext e-invoicing integration wave: taxable revenue above SAR 1M

Pages that quote this rule

3 pages
  • wazen.example/en/e-invoicing/integration-waves

    dateModified

    Out of date
  • wazen.example/ar/e-invoicing/integration-waves

    dateModified

    Out of date
  • wazen.example/en/e-invoicing/phase-2-checklist

    dateModified Not dated

    Out of date

Changelog shown on the page

No changelog. The page never said what changed, or when.

Modelled AI answer

Which businesses must join the next e-invoicing integration wave?

Businesses with taxable revenue above SAR 3M must integrate in the next wave.

Cites majlis-traders.example · a thread from 2022

Wazen cited in 6% of 34 tracked prompts on this topic

Rule changes are simplified for illustration. This is a modelled scenario, not tax advice.

05Timeline

12 months, module by module.

  • One square = one guide published
  • A rule change landed
  1. Months 1–2

    Audit & prompt panel

    PDF inventory, a crawl of the site, and 120 client questions in Arabic and English set up as a weekly AI answer panel. Baseline: Wazen cited in 7% of answers.

  2. Months 2–4

    PDF to guides

    214 PDFs merged into 86 HTML guides, with the highest-traffic topics first. Each guide gets an owner, a review date and redirects from the PDFs it replaces.

  3. Months 4–5

    Freshness system

    dateModified and datePublished in Article schema, visible changelogs, a rule-to-page map and the two-day playbook, rehearsed on a dry run.

  4. Months 5–11

    Rule-change runs

    Four real-world style changes handled with the playbook, each guide updated within one to three days. Citation share climbs after each one.

  5. Months 10–12

    Track & reinforce

    Weekly panel reports, internal links pointing to the current guides, and outreach so the forum threads that ranked link to the updated pages.

06Results

The code scans. The answers quote Wazen.

AI answers citing Wazen (120 tracked prompts)

7% → 46%

Organic consultation requests

+64%

58 → 95 requests a month

Median time to update after a rule change

31 → 2 days

31 days → 2 days, median

Undated PDFs replaced by dated guides

214 → 86

214 PDFs in, 86 dated guides out

Modelled scenario for a fictional business — not a client result. Rankings, AI citations and traffic are never guaranteed; see our Terms.

Modelled AI answers citing Wazen, by monthRule change
0%10%20%30%40%50%46%M1M2M3M4M5M6M7M8M9M10M11M12
Sources cited, month 1 vs month 12

Month 1

Month 12

  • Forum threads47% → 14%
  • Outdated news & PDFs22% → 8%
  • Official sources24% → 32%
  • Wazen7% → 46%

Is your guidance the version the answers quote?

Our Content Engine plan turns expertise into dated, citable pages and keeps them current when the rules move.

← All case studies