Twitter knows your age, gender, salary and interests - but is it accurate?

Twitter has updated its privacy policy – and while usually that wouldn't be interesting, this time things are different. The company has opened up a range of tools for users to better understand – and control – how Twitter and its advertisers target ads at you based on your lifestyle.

But users quickly found out that Twitter isn't that smart. Guesses at age vary by decades, guesses about gender are often wrong, Twitter's understanding of what languages we speak is baffling and interests shared with advertising partners are regularly wide of the mark.

To highlight the many errors in what Twitter thinks of us, this reporter, who has tweeted over 35,000 times since 2008, investigated.

I am male and 27 years old. Twitter gets the first bit right, but then only manages to narrow my age down to between 13 and 54. Technically this is correct, but given Twitter knows I joined in January 2008, it seems unlikely that I'd be 13 now, nine years later.

Twitter has my main language as English (US), which is close enough, but then lists my second languages as Spanish (I can order two beers, at best), Portuguese (nope) and Tagalog, which Wikipedia reliably informs me is the language of central Luzon, in the Philippine Islands.

Next up is the 'interests from Twitter' page, which lists things Twitter believes you are interested in based on your profile and activity on the site. From 'car culture' and 'cell phones' to 'politics' and 'startups' this is a fairly solid list of my interests. I'm not sure what having an interest in 'amazing' means, but nothing stands out as being offensively wrong.

The good news doesn't last long though, as we head into the 'interests from partners' section, which takes a deeper dive into your demographic information and topics of interest. There are a lot of car-related topics, which is fair enough, but then I'm told my car is worth between £23,000 and £30,000 – when it's more like £5,000.

Twitter and its advertisers then claim I'm a homeowner (I'm a millennial in London, so no), and my household income is at least £300,000 (not even close). The value of my home is guessed at £380,000 or over, which is correct. However, the "or over" bit suggests this is the highest possible option, so my three housemates and I are lumped into the same category as single hedge fund managers in multi-million-pound mansions.

The system claims I have a family member aged between 50 and 59, which is correct, and one who is between 20 and 29, which I guess could be me?

A section about brands I'm supposedly interested in shows some promise. Apple, Samsung, Bose, Nikon, and LG are all there, and I own products by all five.

But the good news in short-lived. Under 'proximity' are US retailers like 7 Eleven, Target, Walmart and Walgreens; I haven't stepped foot in any of them, ever.

Finally, 'retail categories' starts strong with 'automotive products' but quickly falls off a cliff with 'cat products' (I do not own any pets), 'dog products', 'home renovation' (I live in a rented house), 'jewelry' (for who?), 'women's accessories' (also, for who?), and 'women's fashion and apparel' (why is this a theme?).

If you want to see how well Twitter knows you, go to Settings -> Your Twitter Data, then enter you password if prompted.

While we found this experiment interesting, others will quite rightly feel uncomfortable with Twitter knowing (or at least trying to know) so much about them. Thankfully individual interests can be switched off – or edited in the case of your age, gender and languages – and personalised ads can be switched off if you go to Settings -> Privacy and safety then uncheck each box or click 'disable all'.