Worth

Worth reads a UK job ad, estimates what the role should pay based on thousands of real ads that publish a salary, and gives you sharp questions to ask at interview. It also looks the employer up at Companies House. Free, no account, nothing stored about you.

Job ads tell you plenty about what they want and very little about what they'll pay. Seven years' experience, three direct reports, full budget responsibility. Salary: "competitive." And when there is a number, it's hard to know whether it's fair for the work involved.

I've worked in tech and marketing for over twenty years, and some of those jobs paid less than the work deserved. I also watched friends, family and people on LinkedIn get worn down by vague ads, missing salaries and endless interview rounds. When you need a job, it's easy to anchor on what you earn now. If you were underpaid before, that number follows you.

So I built Worth. You paste a link to an ad (or the ad's text) and an AI model answers the same 25 questions about it: how much work is involved, whether you'd manage people, how rare the skills are. The salary is hidden from the model while it does this so it can't sway the answers. Those answers, plus location, become around 70 markers that are compared against real UK ads asking for the same things.

The data comes only from what employers published themselves. Over 22,000 UK ads with a stated salary so far, and nothing self-reported. Tested on ads it hasn't seen, the estimate is typically within about 11% of the real salary. Every result shows its working: how many similar ads it rests on, a warning when that's only a few, and an ONS fallback when an ad can't be read properly.

A few rules are fixed. Worth won't sell your data, won't run job listings or take referral fees, and no one can pay to change an estimate. Employers can reply to their record for free. There's also a Chrome extension, so any ad you open can be checked in one click.

• Try an example →
• How the estimate is tested →
• Chrome extension →