I run SalarySouq, a small site that publishes salary guides for professions in the UAE. The target list is 119 professions; 60 guides are live so far. Every guide is built the same way: open the published salary pages (GulfTalent, Naukrigulf, Indeed UAE), record every figure with its source, sample size and check date, cross-check the publishers against each other and against live job postings — then print the disagreements instead of averaging them away.
That process taught me more about salary data than any single dataset could. Four lessons keep repeating.
1. The same job title is often three different jobs
"Insurance Agent" has no salary page of its own on GulfTalent or Naukrigulf. The closest page is "Insurance Advisor" — Naukrigulf averages it at AED 5,477/month (n=63) — while Indeed prices "Insurance Advisor" at AED 3,537 (n=62) and the exact title "Insurance Agent" at AED 4,078 (n=83). Same work, three wordings, three numbers, AED 1,940 apart between publishers on the same wording.
The honest response is not to pick one. Our insurance agent salary guide labels its headline as an advisor-wording headline wherever it is used, and prints every wording's figures side by side. If your data product silently maps titles to each other, you are publishing fiction with decimal points.
2. Sample size is not metadata — it is the story
The thinnest page in the whole series is Journalist. Indeed's exact-title page rests on 4 salaries: average AED 9,945, range AED 4,262–23,208. One new posting can move that "average" by thousands. The neighbouring desk title, Editor, averages AED 5,000 on GulfTalent (n=11, and the crawl is about a year old).
So the journalist salary guide leads with the warning, builds its bands across all the retrieved evidence with each edge named, and adds the only other sourced evidence that exists for the freelance tier — a peer-reviewed University of Sharjah study (15 interviews) reporting retainers of AED 4,000–7,000 and full-time contracts at AED 12,000–25,000. An n=4 average printed without its n is worse than no average.
3. Sometimes the publishers agree — and that is worth checking too
Interior Designer is the healthy end of the series: GulfTalent averages AED 6,000 (n=131, and it publishes an experience staircase — 5,000 at 1–5 years, 7,000 at 6–10, 11,000 at 11–20), Indeed cross-checks at AED 5,300 on 1.4k salaries. Two independent methods landing AED 700 apart is what corroboration looks like. Our interior designer salary guide prints both, unblended — because the distance between sources is itself information about how much to trust the centre.
4. The fixed salary is often only part of the pay
Salary pages record fixed pay. In insurance, the postings state the rest in plain sight: "plus commission", an "attractive commission scheme", one direct insurer posting "Commission / Bonus: 45%" with no base named, and a specialist medical-broking desk at AED 20,000–30,000 with a stated threshold (USD 175,000–210,000) and a 22.5% rate after it. No source publishes a standard commission rate — so we publish none. We publish the structure, plus the one legal fact that makes the split matter: UAE end-of-service gratuity is calculated on basic pay only, so a small basic under a large commission quietly builds a small gratuity.
The method, in one place
Every figure on the site carries its source, sample and date; disagreements stay printed; unverifiable figures are dropped rather than smoothed. The full method is on our how we calculate salaries page, and the guides themselves are indexed on the professions page. The site also runs three free calculators — salary comparison, savings, and city comparison — because a range is more useful when you can place your own offer inside it.
If you build data products from other people's published numbers: print your sources, print your sample sizes, and let the disagreements show. It is slower. It is also the whole job.
Top comments (0)