Why AI lead scoring needs an Arabic explanation

If your sales rep speaks Arabic and the AI score comes back with English-only reasons, the rep cannot act on it. That's not a UX problem — it's a coaching problem. In every pipeline review we sat in on across UAE field-sales teams, the score was either ignored or argued about. Never coached on.

What SHAP actually means

SHAP (Shapley Additive exPlanations) is the standard technique for explaining what a machine-learning model is doing. For each prediction, SHAP attributes the score back to the inputs that drove it: +12 from deal value, +15 from stage, -2 from industry. It's the explainability layer that lets a non-engineer reason about a number a model produced.

Why English-only SHAP fails in the UAE

The label "deal value above org average" lands in an English-speaking head differently than its Arabic equivalent "قيمة الصفقة أعلى من المتوسط". Direct translation is not enough — the sentence structure, the emphasis, and the framing all need to be authored, not pasted. We learned this the hard way during early SOOMA pilots when reps consistently flagged the Arabic reasons as "robotic" until we rewrote them.

The three things we changed

First, we stopped translating English reasons word-for-word and started authoring each Arabic reason from the underlying signal. Second, we standardized the verb forms used across reasons so the cadence is consistent. Third, we let managers customize the reason wording per org — what reads naturally for an Abu Dhabi auto dealer doesn't read the same way for a Riyadh logistics fleet.

What changed in pipeline reviews

Managers stopped saying "the AI scored this 87" and started saying "the deal value signal is strong but the industry win-rate is weak — let's focus there." That's the whole point. The score should be the start of the conversation, not the end of it.

See the bilingual SHAP reasons running in the product on the AI lead scoring page.