UNICEF RFPS-NYH-2026-503962 · Proof of expertise
Digital intelligence that keeps 176 sites legible to search and AI
Five engagements mapped to UNICEF's scope: analytics governance, tag management, Search Console at scale, and visibility in AI-generated answers.
01 / 15What UNICEF is asking for
The AI-readiness roadmap exists. This contract is about running it every week.
In 2026 UNICEF assessed how AI systems read and cite its content across more than 500,000 pages on its Drupal estate. The result was a set of audit reports, a prioritized engineering backlog and a success-metrics framework.
This LTAS asks a vendor to operate that system: keep the measurement clean across every office, watch search and AI visibility, route technical issues to the platforms team, and teach staff to self-serve.
Sources: RFPS-NYH-2026-503962 Annex B (Terms of Reference); Material, "Strengthening Global Authority in the Age of AI Answers," July 2026.
02 / 15Scope to proof
Each service area in the scope maps to work we have shipped for a client we can name
| UNICEF scope area | OHCHR | PerkinElmer | IRONMAN | AMA Insurance | AMA |
|---|---|---|---|---|---|
| Tag management, consent mode and change control | |||||
| GA4 data quality, governance and analysis | |||||
| Search Console at scale (indexing, sitemaps, URL Inspection) | |||||
| Technical SEO findings routed to owners as a backlog | |||||
| SEO platform configuration (Conductor, Semrush) | |||||
| AI-search visibility measurement, KPIs and baselines | |||||
| Recurring reporting from aggregate data | |||||
| Measurement for new site launches and migrations | |||||
| Staff enablement and self-serve tooling |
Filled dot: delivered in that engagement. Server-side GTM and the unified data warehouse are covered on slide 13.
03 / 15Case 1, part 1 · OHCHR · UN Human Rights Office · Analytics and tag governance
Finding the tracking that looked configured but was not
The situation
ohchr.org publishes in all six official UN languages. Years of tags and modules had been added without follow-through, so the dashboards looked complete while the numbers underneath were wrong.
What it proves for UNICEF
What we did
- Audited GTM end to end: a consent banner that had never fired on production, a tag allowlist switched off, declared events that never reached the dataLayer.
- Fixed GA4 data quality: donations counted twice, donation value and ID not captured, test environments writing to the live property, share tracking restored.
- Fixed the sitemap in Search Console and flagged unused paused tags for removal.
- Set up a monthly GA4 and Search Console report, generated automatically from aggregate data only.
- Advised OHCHR's web unit on a proposed UN Secretariat-wide GTM rollout, including the change control and disconnect procedure UN OICT cybersecurity would need.
Sources: OHCHR GTM Audit (May 2026); OHCHR Analytics Audit (Oct 2026); OHCHR Metrics, September 2026.
04 / 15Case 1, part 2 · OHCHR · Drupal on AWS, six UN languages · Technical SEO and AI visibility
A trusted UN authority that AI under-cites, and the ranked fixes to close the gap
The situation
OHCHR is a primary source on human rights, yet AI answers were not naming or citing it in proportion to that authority. The office wanted to know why, and what its Drupal site could change.
What it proves for UNICEF
What we did
- Audited the Drupal codebase directly across 12 areas: configuration, custom modules, templates, hreflang, canonicals, structured data, sitemaps and redirects.
- Cross-checked code findings against an automated crawl and Semrush, keeping what the code confirmed and correcting two automated misreads.
- Measured how ChatGPT, Gemini and Google's AI answers name and cite OHCHR.
- Ranked fixes easiest first: sitemap and robots.txt, homepage canonical, Organization schema marking OHCHR as a UN body, and hreflang for all six UN languages.
- Followed with content-strategy recommendations: a keep, update, consolidate, redirect or archive policy for old pages, and freshness dates handled by content type.
Sources: OHCHR Technical SEO and AI Visibility Audit (July 2026); OHCHR content strategy notes (August 2026).
05 / 15Case 2 · PerkinElmer · Life sciences · Drupal on Acquia
Four data sources joined on one URL, so every fix has evidence and an owner
The situation
perkinelmer.com held a large library of application notes and brochures with no clear view of which pages were losing visitors or leads. The team needed a ranked backlog, not a generic audit.
What it proves for UNICEF
What we did
- Crawled all 1,867 library pages and every link on them.
- Joined the crawl with the Drupal export, six months of GA4 and Search Console URL Inspection API results for every URL.
- Hand-checked 39 pages in a browser against the dataset; all 39 matched.
- Delivered a 388-row fix list with evidence, action and suggested owner on each row, ranked P1 to P3.
- Recommended template changes over page edits: one heading fix covers 1,848 pages, and structured data extends from the existing chromatography hub pattern.
Source: PerkinElmer library and application-note inventory, 28 October 2026 (28,981 Google clicks analyzed, 16 April to 15 October).
06 / 15Case 3 · IRONMAN · Global sports brand · Drupal on Acquia, multi-domain
Measuring how AI engines answer for a global brand, then changing what they read
The situation
IRONMAN serves a global athlete audience across ironman.com and two sister domains. Search demand is moving into AI answers, and the team needed to see that shift and act on it.
What it proves for UNICEF
What we did
- Planned and governed AI-visibility tracking in Conductor across Google AI Overviews, AI Mode, ChatGPT, Perplexity and Microsoft Copilot.
- Designed the tracking model: 7 audience personas, 25 topics and 1,750 weekly prompts, sized to the credit allowance.
- Used GA4 and Semrush to map about 99,700 yearly organic entries before an information-architecture change.
- Rebuilt evergreen articles for answer engines with answer capsules, FAQs and structured data; 45 are live, with results on the next slide.
- Packaged the method as a content operating system and reusable team tools so editors can self-serve.
Sources: IRONMAN Conductor AI-tracking plan; Training IA migration analysis (GA4, Semrush, Conductor); refresh register, 2026.
07 / 15Case 3 results · IRONMAN evergreen refresh · 45 pages live
A refreshed library that compounds, in search and in AI answers
Weekly organic sessions: 716 to 10,116 in five weeks
Week starting. Aug 9 was a soft week before the steepest climb.
AI-referred sessions: 3 to 99 a week
All from ChatGPT so far. Small volume, same inflection as organic search.
Source: Axelerant performance report for The IRONMAN Group. Conductor, GA4 and Semrush, page group "Refreshed Content," 28 June to 5 September 2026.
08 / 15Case 4 · AMA Insurance · Financial services · WordPress to Drupal 11
Baselining search and AI visibility before the platform changes
The situation
AMA Insurance is moving amainsure.com from WordPress to Drupal 11. The risk was losing rankings and AI citations that nobody had measured before launch.
What it proves for UNICEF
What we did
- Captured the pre-migration baseline: 5,527 ranking keywords, 4,232 of them showing Google AI Overviews.
- Flagged that the site was not yet tracked in Conductor and defined the AI-citation baseline to capture before content freeze.
- Crawled 606 URLs with JavaScript rendering, then built the URL parity inventory and redirect map.
- Documented schema parity per template and a fix-don't-carry-forward list, including 22 live 404s and 48 pages without an H1.
Source: AMA Insurance pre-migration SEO and discoverability baseline, 7 October 2026 (Semrush, Screaming Frog).
09 / 15Case 5 · American Medical Association · Healthcare · Two ethics sites plus ama-assn.org
A trusted medical authority that AI cites often but surfaces rarely
The situation
The AMA publishes the Code of Medical Ethics and the Journal of Ethics. How well do search and AI read them?
What we did
- Audited both sites on 80 technical checks, ordered by points lost.
- Code of Medical Ethics AI gaps: AI-friendliness 7 of 31, no structured data, 236 JavaScript-reliant URLs.
- Benchmarked ama-assn.org by AI platform and by topic.
AI visibility by platform, ama-assn.org (score of 100)
Overall score 25 of 100, while AI tools cite the site 140,000 times.
Source: Axelerant SEO and AI Visibility deck for the AMA, August 2026 (Semrush Site Audit, Screaming Frog, Semrush AI Visibility, 25 August 2026).
10 / 15Proposed approach
One operating rhythm across 176 sites, not 176 separate projects
Govern
Inventory every GA4 property, GTM container and Search Console property. Individual named accounts, UNICEF-owned assets, documented change control with rollback.
Measure
One event and naming standard for every office. Automated data-quality checks that flag anomalies by site. Consent mode verified per site.
See
GA4 via BigQuery, Search Console, the SEO platform and AI-visibility data in one model, surfaced in Looker Studio globally and per office.
Enable
Up to five recorded training hours a quarter, self-serve guides, and findings routed to the platforms team with evidence and an owner.
11 / 15Security and governance
Built to work inside UN security review
- Individual named accounts with multi-factor authentication. No shared or vendor-owned logins.
- UNICEF keeps ownership of every account, container, property and dataset, and can revoke access without us.
- Every GTM publish documents purpose, sites affected, test results, approver and rollback.
- Custom HTML only with written justification, security review and UNICEF approval.
- Reporting from aggregate data. UNICEF data is never used for model training or product development without written approval.
From the OHCHR engagement
When the UN Secretariat proposed one GTM container across UN sites, we helped OHCHR's web unit assess the risk of a single compromised container and set out the change-control and disconnect steps UN OICT cybersecurity would need before any rollout.
Mapped to Annex B5, UNICEF Security Requirements for Class II Systems.
12 / 15The two builds in the scope
Server-side GTM and the data warehouse are builds, and we run them as builds
Server-side GTM on UNICEF's Azure
- Audit and document every tag, the data it collects and how consent is handled.
- Measure the data-quality gain on unicef.org traffic, not industry benchmarks.
- Written go/no-go with sizing, cost, security and ownership.
- If go: parallel collection to prove parity, a planned cutover with rollback, then handover and training.
Unified warehouse with a plain-language layer
- Cloudflare logs, the GA4 BigQuery export and social analytics in one consistent model.
- Recurring reports generated automatically, with anomalies surfaced rather than hunted for.
- Staff ask questions in plain language, answered from governed data only.
- The PerkinElmer inventory used the same join-on-URL discipline at smaller scale.
Each build gets a named lead, its own plan and acceptance criteria under a separate contract for services.
13 / 15Why Axelerant
The team that knows the platform, and fixes what it finds
Drupal and Acquia from the inside
UNICEF runs on Drupal and Acquia. So do PerkinElmer and IRONMAN, and OHCHR runs on Drupal. We know where dataLayers, consent banners and multisite launches break.
UN experience
OHCHR in six UN languages, including GTM risk review for a UN-wide rollout.
We fix, not just report
Findings ship as owned, evidenced backlogs, and many are fixed by us.
AI-search measurement already running
AI-visibility baselines for OHCHR, the AMA and AMA Insurance, and weekly tracking across five engines for IRONMAN.
Also in the practice: an SEO and analytics revamp for ESC and an SEO workstream for PADI. Working with clients since 2005.
14 / 15What you can hold us to
In 90 days, you will know what every office measures, and whether to trust it.
Every GA4 property, GTM container and Search Console account inventoried, with an owner and a risk rating.
Tag, consent and data-quality findings across the estate, ranked and routed to your platforms team.
An AI-visibility baseline across AI Overviews, ChatGPT, Gemini, Perplexity and Copilot, tied to the KPIs in your 2026 roadmap.
Your questions →
[Presenter name, role] · [Presenter name, role] · Axelerant
15 / 15