Analytics03 Aug 2026 · 6 min read · explainer

Attribution Models Explained Simply

Make sense of marketing attribution for UAE journeys across Google, Meta, WhatsApp, and calls—without worshipping any single platform’s ROAS.

Attribution models explained simply (for multi-path UAE journeys)

Attribution is how you assign credit when a customer saw Instagram, searched Google, asked a friend, WhatsApped you, and only then booked. Everyone wants a single pure model. Nobody has one. The goal is decision-grade fairness, not metaphysical truth.

Why attribution fights break out in Dubai teams

Meta reports a purchase
Google reports a conversion
GA4 reports a different path
Finance sees one invoice

All can be “right” inside their rules. Conflict is designed in.

UAE twists:

WhatsApp and phone cut web paths short
Multi-emirate targeting with one creative
Bilingual behaviour
Influencer → brand search halo

If you only trust last-click web conversions, you under-fund upper funnel and over-trust branded search.

Core models in human language

Model	Idea	Tends to favour	Weakness
Last click	Final touch before conversion wins	Brand search, retargeting	Ignores discovery
First click	First touch wins	Prospecting, influencers	Ignores closer channels
Linear	Equal split across touches	“Everyone did something”	No prioritisation
Time decay	Recent touches get more	Lower funnel	Still web-path biased
Position based	First + last get more	Hybrid compromise	Arbitrary weights
Data-driven (platform)	Algorithmic split	Whatever the platform sees	Black box; incomplete offline

GA4 and ads UIs ship variants of these. Platform-reported conversions are not a cross-channel ledger. They are optimisers’ scoreboards.

The multi-touch fantasy vs SME reality

True multi-touch with identity resolution across devices, WhatsApp, and calls is enterprise territory. Most Dubai SMEs should not buy a CDP to settle an argument about AED 20k/month media.

What works instead (practical stack)

Blended CAC and revenue as the court of final appeal
Channel diagnostics with honest caveats
Geo / time holdouts when the decision is expensive
Offline conversion truth for closed-won
Annotate major creative and landing changes

How to read Google vs Meta without rage

Expect duplicate credit across platforms
Summing platform conversions will exceed real orders — always
Compare trends and cost per offline outcome, not absolute conversion totals
Use platform models for in-platform bid decisions
Use finance + CRM for budget allocation across platforms

A useful weekly question: If we cut Channel X by 30% for two weeks, what happens to blended new customers? That is incrementality thinking without a PhD.

Paths common in the UAE (illustrative)

Service / trades: YouTube or Meta discovery → Google Search “near me” → call → WhatsApp quote → job. Last-click = call/search. First touch = video. Cutting video may quietly raise Search CPCs later.

Clinic: Instagram → WhatsApp → visit. Web attribution almost empty. CRM stages matter more than GA4 paths.

Ecommerce: TikTok → Instagram retarget → Google brand → purchase. Last-click over-credits brand. Need new-customer filters.

B2B free zone: LinkedIn → webinar → email → sales call → deal 45 days later. Offline import or CRM revenue is the only adult metric.

Building a “good enough” attribution policy

Write a one-pager:

Primary company metric: new customers, contribution margin
Primary in-channel metric: platform conversion that best correlates with primary (validated quarterly)
Credit rules for reporting: e.g. “We do not add Meta + Google conversions; we report each separately and reconcile to CRM”
Brand vs non-brand Search split always
Influencer measured with codes + UTMs + lift notes

Share it with every agency. Arguments without a policy are sports talk.

Incrementality tests you can actually run

When monthly spend on a channel exceeds roughly AED 10–15k and debate is stuck:

Geo split: pause prospecting in one similar area cluster (e.g. compare certain Dubai communities carefully)
Time-based: 14-day 30% budget cut
Creative holdout: stop a heavy influencer flight and watch branded search + direct

Measure new customers and revenue, not only platform ROAS. Document. Small samples are noisy — look for directional truth, not false precision.

GA4’s role

Use GA4 paths and model comparison as hints:

Which landing pages assist
Whether organic and paid cannibalise
Where Arabic vs English paths differ

Do not crown GA4 data-driven attribution king if WhatsApp closes half of sales invisibly. Fix offline import and call tracking first.

Agency reporting red flags

Single screenshot ROAS without CRM
“We use data-driven attribution” as a conversation ender
Refusing to separate brand Search
Summing conversions across platforms into one “total leads” superhero number
No mention of view-through vs click-through definitions

See agency reporting.

Decision framework (print this)

Is blended CAC inside target? If no, fix before model debates.
Is a channel claiming credit without click IDs or codes? Discount heavily.
Is the channel mostly last-click brand? Protect it, do not scale it like prospecting.
Is offline path dominant? Invest in CRM attribution first.
Is the budget change >20%? Require a holdout or phased rollout.

Illustrative conflict resolution

Marketing wants more Meta; sales says leads are tyre-kickers; finance wants cuts. Offline labels show Meta produces 55% of chats but 25% of won jobs; Google Search inverse. Solution: Meta creative and qualification tighter; Google budget protected; shared KPI = cost per won job, not chat. Attribution model arguments end when the KPI matures.

What not to do

Do not pay for a black-box multi-touch tool before UTMs work
Do not change models every week to pick the flattering one
Do not ignore view-through entirely for pure branding, but do not treat it as cash

Closing

Attribution is a set of agreements + measurements, not a divine algorithm. In the UAE’s WhatsApp-heavy, multi-platform reality, the winners reconcile platforms to AED outcomes and stay humble about precision.

Part of the Dubai Marketing Playbook by Shabang — practical marketing for UAE businesses.

Attribution Models Explained Simply