Which Tweets Should You Clean? Rank by Risk

Last week a friend sent me a screenshot of his X account's privacy checkup: seven years, 23,000 tweets, 61 red high-risk labels. He stared at the screen and asked, "I can't delete everything, can I? What tweets to delete first?"

Fair question. Nobody reads twenty thousand tweets one by one, and most of them honestly don't need to go. Cleaning your timeline shouldn't run on vibes. It should run on priority. The tiers below are what I landed on after scanning 50,000 of my own historical tweets, and you can copy the order directly.

P0: Anything that exposes who you really are, clean today

One test for this tier: **could a stranger reading this tweet point straight to the real-life you?**

Typical examples:

• A phone number, even if posted once, even in a reply
• A personal email, especially one built around your real name
• Home address, door number, or something like "waiting for my package tonight" that implies you're home
• Photos of ID cards, passports, driver's licenses
• A real-name company paired with start and end dates, which lets people reconstruct your work history

The rule is simple. Before deleting, ask yourself: when I wrote this, did I forget strangers were watching? If yes, P0.

P1: Things that map your daily life, handle soon

This tier doesn't expose identity directly. It exposes patterns. The risk labels in your report will flag this kind of content for you.

• Check-in geotags, like "just landed at the station" or "in the office, traffic is a nightmare"
• Anything around your home, the pancake stand downstairs, the neighborhood water outage
• Fixed commute routes, the gym and hospital you always go to
• Travel plans, "flying to Tokyo on Friday, staying in Shinjuku"

One post alone means nothing. Fifty of them are a very complete map of your movements. These usually carry location-related labels, and the report marks them for you.

P2: Person-related and easy to weaponize

This tier is softer, but when it bites, it hurts more.

• Negative takes on former employers or coworkers, because your next HR will search
• Screenshots of politically sensitive posts or jokes about regions and genders, which get taken out of context and shared on their own
• Old fights with people, especially anything that got quote-tweeted
• Complaints about clients or suppliers, even when you were right

Try this test instead: if this tweet got screenshotted into your new team's group chat, could you smooth it over? If not, it goes.

P3: Embarrassing but harmless, your call

Angsty poetry from your twenties, 2 a.m. sorrow, eight hundred reposts for your favorite idol. These don't pose real risk, and keeping them actually makes the account look human. Delete them if you want, or leave them alone. If one tweet makes you cringe every time you see it, deleting costs you nothing.

Risk labels are your sorting tool

You don't have to judge each tweet yourself. Upload your X data archive, and the report tags every high-risk tweet with labels like phone, email, address, location, and sensitive, ranked from highest to lowest risk. Your job is to sweep from P0 downward, batch-select, delete. Takes about a minute.

What this looks like in practice: a five-minute flow

Let me get concrete, because theory only gets you so far. This is how I run a cleanup myself:

1. Export your X data archive from the settings menu. It arrives as a ZIP, and depending on account age it can take anywhere from twenty minutes to a few hours to generate.
2. Drag the ZIP into the checkup tool and let it parse locally.
3. Filter by P0 first. Read the original text of those few high-risk tweets, make sure nothing was mislabeled, then batch-delete the whole tier.
4. Move on to P1 and clear out the location and home-address related posts while you're in the zone.
5. Save P2 for the weekend. Go through them one by one when you're in a decent mood, and if reading the original screenshot makes you wince, it goes.

All in, about ten minutes. Don't aim for one perfect sweep. Getting P0 and P1 to zero sets the safety floor for your whole account, and everything after that can wait.

One thing worth stressing: **the deletions run on your own device, using your own account credentials**. The tool only tells you what to remove. You press the buttons. That keeps control in your hands and leaves room for second thoughts, since some platforms keep a short undelete window if you do hit the wrong thing.

Priority table:

Tier

Risk labels

Typical content

Suggested action

Example

P0

phone / email / identity

Phone number, personal email, ID photos, address

Delete today, all of it

Posted your number in a reply

P1

location / address

Geotag check-ins, home surroundings, trips

Handle within a week

"Just arrived at the station"

P2

sensitive

Employer rants, political posts, flame wars

Clean within a month

A coworker rant that got shared

P3

No high-risk labels

Angsty old tweets, late-night feelings

Optional

A poem from ten years ago