Analytics and CRO, measurement that earns spend.

Why is my GA4 data unreliable?

Bad data is expensive because it steers budget: eBay’s experiments showed paid-search returns were “a fraction of conventional non-experimental estimates” (NBER Working Paper 20171). Attribution built on duplicated events makes that gap worse.

“…brand-keyword ads have no measurable short-term benefits.”

Blake, Nosko & Tadelis, NBER Working Paper 20171 (eBay field experiments)

• GA4 set up in a hurry: conversions firing on the wrong pages, duplicate events, broken e-commerce tracking.
• Platforms over-claiming: several ad platforms counting the same sale. See the ROAS that lies.
• The CRM gap: platforms see leads; only the CRM knows which leads became revenue.
• Privacy changes: iOS tracking limits and consent requirements leave gaps that server-side and modelled approaches have to fill.

What does a reliable measurement setup include?

1. GA4 and Tag Manager done properly: documented tags, deliberate conversion events, custom dimensions and reports built around business questions.
2. Server-side tracking: Meta Conversions API, Google Enhanced Conversions and TikTok Events API, deduplicated, under consent mode.
3. CRM integration: HubSpot, Salesforce, Pipedrive or Zoho, with source tracking and lifecycle stages, and revenue fed back to marketing.
4. Attribution: credit spread across the journey rather than handed to one click.
5. Media mix modelling for accounts with enough spend and history, typically from around AED 100,000 a month.

How should conversion tests be run?

1. Audit: funnel analysis, heatmaps and session recordings (Microsoft Clarity, Hotjar) produce a ranked list of opportunities.
2. Hypotheses: each test states what changes, which metric should move and why.
3. Statistics: sample sizes calculated beforehand; inconclusive results reported as inconclusive.
4. Rollout: winners shipped and monitored; losers documented so they aren’t retried.

Where are the biggest conversion gains?

Speed is usually first: Google rates a Largest Contentful Paint under 2.5 seconds as “good” (web.dev), and the UAE’s 23.0 million mobile connections (DataReportal) mean most sessions are on phones.

• Forms: fewer fields, better order, clearer errors and a better mobile experience.
• Speed: slow mobile pages lose visitors before they see anything. See web development.
• Mobile: most traffic is on phones, so tap targets, sticky calls to action and mobile forms matter.
• Pricing pages: anchoring, comparison and how value is explained.
• Proof: genuine reviews and case material placed where decisions are made.

The engagement

• Week 1: foundations audit with a prioritised fix list.
• Weeks 2–6: fixes, which often improve results before any testing begins.
• Ongoing: monthly testing cycle with quarterly strategy reviews; media mix modelling where it fits.