The Content Audit
Evidence: giggulf.ae Search Console + GA4 (16 months, UAE-filtered) and an in-house AI-visibility run. The diagnosis of GIG's whole content estate, and the plan to brief the new producer. Figures are labelled UAE or global throughout; the 3 recommended blogs ship separately, upload-ready.
Part 0 — What we audited, and how
Scope — what we looked at
| Surface | Size | Depth of audit |
|---|---|---|
| Blog | 131 posts | Full — every post dispositioned |
| Product / funnel pages | 25 | Performance + ranking |
| Document library (PDFs) | 527 | Counted; remediation queued |
| Arabic estate | 81+ | Parity check |
| Whole site | 1,765 URLs | Mapped + classified by type |
| Competitors | 4 | Sukoon, Salama, Orient, Lemonade — volume + format |
| AI answer engines | 3 | ChatGPT, Claude, Google — 20 UAE buyer prompts |
| Customer reviews | 1,000 | Theme + sentiment mining — voice of customer |
| On-page video | — | Embedded video presence + the produced reels |
| SEO & technical | — | Rankings, CTR, cannibalisation, schema, index hygiene |
| GEO / AI-readability | — | AI citation share + Open Knowledge Format readiness |
How — the method
Every audit is reproducible — the model, the prompt and the script are linked below, not described.
| What | Tool / model | Prompt / query | Script |
|---|---|---|---|
| Search performance | GSC + GA4 export, UAE-filtered | gviz page/query pulls, 16 mo | method |
| Estate + competitor map | Firecrawl (firecrawl_map) | sitemap of giggulf.ae + 4 competitors | method |
| AI visibility | GPT-4o-mini · Claude Haiku 4.5 · Google (Serper) | 20 UAE buyer prompts · system: "insurance shopping assistant for the UAE — name specific insurers" | run.py · report.py |
| 1,000 reviews | qwen3:14b — local, Ollama, $0/call | fixed theme + sentiment taxonomy (9 themes) | triage_reviews.py |
| Per-page disposition | rule-based on GSC metrics | keep / refresh / rewrite / prune thresholds | register · Airtable |
Honest scope. Done: estate map, GSC + GA4 pull and analysis, AI-visibility baseline, competitor scan, the database. Queued (more time or access): full 131-post text scrape + per-post /20 scoring, the 527-doc library, multi-market GSC, backlinks (needs an Ahrefs API tier). Each carries a one-line "how to finish" in the register.
Part 1 — The Audit · what we found
The picture — in 10 charts
UAE Search Console, 16 months. The first four are the story: GIG is shown far more and clicked less.
The content estate, by type
A review of all of GIG's content, not the blog alone. Each type plays a different role in the funnel; the strategy differs by type. The blog is one type among eight. (Clicks are 16-month: global totals for cross-type comparison; UAE-specific figures appear in the sections below.)
| Type | Vol | Role | Performance (global) | Strategic action |
|---|---|---|---|---|
| Product / funnel | 25 | Conversion | Car 18.7K, travel 13.2K, health 10.8K clicks — rank 8–23 | #1 lever — lift rankings, schema + video, internal-link from blog winners |
| Blog | 131 | Top-funnel demand | 47K clicks; 68% from 5 posts; CTR collapsing | Refresh winners, prune the dead (the backlog, Part 3) |
| Document library | 527 | Trust / compliance | Largest surface; unversioned, search-invisible | Inventory, version, index, surface key docs + schema |
| Arabic estate | 81+ | In-market reach | Product pages rank (car 4.3K); Arabic blog is dark | Parity on product + utility winners, not the blog |
| Non-blog articles | ~10 | Demand capture | Road-trip-Oman alone 10.4K — beats the whole Travel blog | Bring into the system; replicate the format |
| Tools / calculators | ~0 | Conversion + GEO | Near-absent | Build — the biggest differentiator (the Lemonade model) |
| Video | reels | Engagement / trust | Produced (Arjun reels), not embedded on pages | Place on product + top pages; VideoObject schema |
| Reviews / UGC | 1,000 | Trust | eKomi badge only | Surface on product pages; mine for FAQ/topics |
The blog is one type among eight
The biggest commercial lever is the product pages (ranking 8–23 on the most valuable queries); the biggest untapped one is tools/calculators; the biggest risk is the unaudited 527-doc library. The blog's job is to feed the product pages, not to exist for its own sake.
Blog — what the data says (UAE)
| Finding | So what |
|---|---|
| Motor/Utility wins; Travel doesn't. 897 clicks/post vs 86 vs 5 (Wellbeing) (global). | Double down on motor + government-service how-tos. Stop over-feeding Travel. |
| 5 utility how-tos carry the blog — accident report, traffic fines, Orange Card, black points, Salik. | Crown jewels — refresh and protect first. |
| Best content isn't in /blog/. "Road trip UAE to Oman" — 10,379 clicks (global). | Audit the whole estate; replicate what wins. |
SEO output
The headline signals. The full page-level technical crawl — the ~50 pages of per-URL data a standalone SEO audit produces — lives in the audit register and the Airtable base; this is the summary.
| Signal | State |
|---|---|
| Average position (UAE) | 7.8 — page 1, stable |
| Click-through rate (UAE) | Collapsing 6.3% → 1.8% (the CTR war) |
| Cannibalisation | 5+ duplicate "traffic-fines" URLs competing — consolidate to one canonical |
| Index hygiene | Dozens of /c/portal/update_language? URLs indexed — noindex |
| Schema | Template lacks Article/FAQPage JSON-LD; FAQs carry broken placeholder answers |
| Product pages | The money pages rank 8–23 — the biggest ranking upside |
Growth, the war, and seasonality (UAE, YoY)
Seasonality: clicks peak in spring (Mar–May); impressions build through autumn/winter. Blog was biggest spring 2025. The Sep 2025 25-post Travel blitz did not lift clicks — volume for its own sake failed. Ship to the calendar → seasonal calendar.
Customer voice — 1,000 reviews
Content implication: customers care most about renewal, speed and the digital/app experience; the negative cluster sits in digital/app (67) and claims (32). Those are the blog + FAQ topics to own and the friction to fix. Full review analysis →
What works, what's tired
| Pattern that works | Evidence (UAE clicks) |
|---|---|
| Practical / regulatory how-tos | Accident report 5,343 · traffic fines 4,168 · black points 2,815 · Orange Card 2,242 |
| Newsjack tied to an event | Al Barsha fire / home insurance — ~1,100 off one event |
| Think-pieces on what's changing | Chinese cars cost more to insure · EV insurance — rank and hold |
| Destination & seasonal | Oman road-trip 10,379 · long tail of destinations, each small |
Tired: off-line-of-business Wellbeing posts (yoga, self-care) and duplicate URLs — near-zero UAE clicks. Destination posts look dead globally but rank page-1 in the UAE (Thailand pos 10, not 46) — refresh, don't retire.
On-page video
GIG produces video — the Arjun testimonial and explainer reels — but it is not embedded on the content or product pages. Competitors are text-only. An open lever: place a testimonial/explainer on the top product pages and high-traffic how-tos for dwell-time, engagement and VideoObject schema (a SERP + AI-answer signal). The supply already exists; the gap is on-page placement.
Competitive positioning
| vs | Their content | GIG |
|---|---|---|
| Sukoon | ~25 blogs, but 80+ news + 70+ awards — heavy authority content. | Out-blogs them 135:25; trails on authority. Close it with original UAE data. |
| Salama / Orient | Product-led, no blog. | Far ahead on content already. |
| Lemonade (global) | Structured "Explained" guide-hubs + programmatic guides. | The model to borrow; GIG's vehicle-model pages are an embryo of it. |
GEO output — AI answers
What we did: ran 20 UAE insurance-buyer prompts across ChatGPT, Claude and Google and recorded, per answer, whether GIG (incl. AXA legacy) and each competitor were cited — scored by funnel phase. How: run.py + the prompt set · re-runs weekly · full report →. Result: 83% overall, 92% on category — but carried by AXA legacy equity (~25% of GWP — protect, don't scrub).
| Funnel phase | GIG cited | Read |
|---|---|---|
| Trust / brand | 100% | Owns its own name |
| Competitive (vs X) | 100% | Named in head-to-heads |
| Category ("best X insurance") | 92% | Strong — but on AXA equity, not "GIG" |
| Attribute (features) | 83% | Surfaces on feature queries |
| Utility how-tos | 33% | The weak spot — the clean GEO win via schema/OKF |
The GEO build: structure GIG's content into the Open Knowledge Format (OKF) — the AI-agent-readable layer — so GIG surfaces on-message in AI answers. The UAE early-mover window is open now. A worked market prototype for the "best travel insurance" query already exists → InsuranceCompared prototype.
Part 2 — The Gap · current vs ideal
Gap analysis
Where GIG falls short of a best-in-class content estate — the gap each priority closes.
| Dimension | Now | Ideal | The gap | Sev. |
|---|---|---|---|---|
| Product-page ranking | pos 8–23 on the money queries | page 1, schema-rich, video | Conversion pages under-ranked | High |
| Tools / calculators | None | Premium estimators, comparators (Lemonade) | A whole high-converting content type missing | High |
| Click-through | CTR 6.3→1.8%, falling | Titles/snippets/schema recovering clicks | Losing the click to AI Overviews | High |
| Authority / E-E-A-T | No research, no expert bylines | Original UAE data, named authors | Trust layer absent (vs Sukoon) | Med |
| GEO / AI-readability | 83% via AXA equity; no OKF | OKF deployed; utility how-tos cited | Not AI-agent-readable — window open now | Med |
| Document library | 527 PDFs, unaudited | Versioned, indexed, schema'd | Trust/compliance + crawl waste | Med |
| On-page video | Reels produced, not embedded | Video on product + top pages | Engagement signal unused | Med |
| Blog quality | No byline/date/sources/tables; broken FAQs | The 10-point /20 standard | Below the credibility bar | Med |
| Arabic parity | Product ranks; blog dark | Parity on the winners | Half the in-market audience under-served | Low |
Part 3 — The Order · what to do, in priority
The plan, by content type
| Type | Priority | The move |
|---|---|---|
| Product / funnel | 1 — highest commercial | Title/schema/internal-link the car, travel, health pages off page 2 → page 1. Embed video. |
| Blog | 2 — fastest measurable | Work the disposition backlog: refresh the CTR-leakers, prune the dead 8. |
| Tools / calculators | 3 — biggest differentiator | Build the first calculator (premium estimator) — out-converts & out-cites 20 blogs. |
| Document library | 4 — risk + trust | Inventory the 527 PDFs, version + date, surface and schema the key ones. |
| Arabic + video + reviews | 5 — ongoing | Arabic parity on winners; embed video on pages; surface reviews on product pages. |
Blog backlog — the page-by-page verdict
Every post has a disposition (UAE data): 4 keep · 6 refresh-CTR · 9 rewrite · 43 thin · 8 prune · 24 review. The full page-by-page list, each linked to the live page, is the blog backlog →. Top of the queue:
| Top of the queue (UAE) | Clicks | Pos | Verdict |
|---|---|---|---|
| Dubai traffic fines & black points — 552K impr · 0.75% CTR | 4,168 | 7.7 | Refresh — CTR |
| A complete guide to Salik — 256K impr · 0.28% CTR | 717 | 5.1 | Refresh — CTR |
| Top health insurance companies UAE — 62K impr · 0.27% CTR | 167 | 10.7 | Refresh — CTR |
| Vehicle-model cluster (Lexus/Nissan/Toyota ×10) | low | 13–30 | Rewrite — rank |
| 36 zero/low-click Wellbeing + duplicate URLs | <5 | — | Prune / merge |
Full page-by-page list → the backlog. Underlying data → the audit register.
First 30 days
- Consolidate the duplicate URLs. 5+ traffic-fines variants — 301 to one canonical.
- Rewrite titles + snippets on the CTR-leakers — recover clicks already earning impressions.
- Fix the broken FAQs + add schema (Article + FAQPage) on the top 20.
- Internal-link winners into the product pages to lift the money pages ranking 8–23.
- Prune / merge the 36 dead posts.
The standard & cadence
Every asset scores 0–2 on ten elements (Content Richness /20): trust · data · sources · authors · facts · tables · quotes · pictures · calculators · research — what Google E-E-A-T and the AI engines reward. Cadence ~8–10 assets/month: refresh the backlog first, one authority piece + one tool per quarter. Gate: GillyBlog (voice) → Belinda (brand) → Candace (facts & quotes).
Next steps
- Confirm the backlog is the new producer's day-one queue.
- The 3 upload-ready rewrites are built (traffic-fines, Salik, top-health) — each lifted from 4/20 to 15/20, paste-ready for Sucheta.
- AI-photo policy for non-people hero scenes (yes/no).