A Pragmatic Guide to DeGoogle

Let's be clear about something upfront: this guide is not about Google.

Not exclusively, anyway. "DeGoogling" has become shorthand for a broader project — moving away from the surveillance economy, reclaiming ownership of your data, and reducing your dependence on centralized platforms that treat your behavior as their product. Google is the most visible example, but the same logic applies to Meta, Apple, Microsoft, and most of the dominant SaaS ecosystem.

This is also not a guide that demands a complete overhaul. There is a vocal faction in the privacy community that will tell you to run your own mail server, compile your operating system from source, and communicate exclusively through encrypted mesh networks. That approach is genuinely admirable, however that advice is mostly out of reach for people who are not tech savvy, are busy, and have finite time. If that is the standard, most people will correctly decide it is not worth it and change nothing.

The business model of most "free" digital services is data extraction. Your emails, search queries, calendar events, location history, and browsing patterns are harvested, aggregated, sold to data brokers, used to build behavioral profiles, and fed into advertising and influence systems. You are not the customer. You are the product.

The harm is not always obvious or immediate. But the cumulative effects are significant:

• Surveillance infrastructure built for advertising can be repurposed — by governments, by bad actors who breach platforms, or by the platforms themselves under changing ownership.
• Data broker ecosystems mean information collected by one service ends up aggregated with dozens of others, creating profiles far more detailed than any single platform holds.
• Behavioral manipulation — recommendations, feeds, notifications — is engineered to maximize engagement, not your wellbeing.
• Lock-in means switching costs rise over time, concentrating power with a small number of platforms.

Before touching a single service, accept the following:

The problem with Gmail

Your Gmail address is not just a communication channel. It is your identity across the internet. Every service you have ever registered with that address is linked to a Google-controlled namespace. Google can read your emails, cross-references them with your search history and location data, and the address itself belongs to them, not you.

The solution: own your identity at the domain level

The single most important infrastructure investment you can make is buying a personal domain. It costs roughly 10–15 EUR per year. What it buys you is portability: your email address becomes you@yourdomain.com, which you control regardless of which provider handles delivery.

Email aliases

Once you have a domain, use email aliases for every service registration. Tools like Proton Pass or SimpleLogin generate disposable addresses like random-alias@some-domain.com that forward to your real inbox.

A better approach: subdomain catch-all

Using Proton Pass aliases works, but comes with a subtle risk: those aliases live under Proton's domains (like passmail.net or passfwd.com). If you ever want to leave Proton, you would need to manually update every service you registered with those addresses. That is vendor lock-in by the back door.

A cleaner solution if you have a custom domain: create a dedicated subdomain for consumer services — for example alias.yourdomain.com — and add it as a secondary email domain in Proton. Then enable catch-all on that subdomain. Any email sent to anything@alias.yourdomain.com lands in your inbox, with no need to create addresses in advance.

Reusing passwords is the most common vector for account compromise. A password manager generates and stores unique, complex passwords for every service. If one service is breached, no other account is at risk.

A layered approach by sensitivity

Rather than choosing one tool, split credentials by sensitivity level:

The Google Docs problem

Google Docs and Sheets are genuinely excellent products. Moving away from them is probably the hardest practical concession in this entire guide. The collaboration features, the spreadsheet functionality, the ubiquity — nothing else matches them without friction.

Storage alternatives

Migration from Google

Browser

Uninstall Chrome. It is, functionally, a telemetry and advertising platform that also renders web pages.

Search

What a VPN does and does not do

A VPN encrypts traffic between your device and the VPN server, preventing your ISP from seeing your browsing. It replaces your IP address with the VPN server's IP. What it does not do: make you anonymous, protect against tracking cookies and fingerprinting, or secure traffic between the VPN server and its destination. A VPN shifts trust from your ISP to the VPN provider.

WireGuard for home network access

If you run a home server and want to access self-hosted services while traveling, a WireGuard tunnel is the right approach. WireGuard is modern, fast, and cryptographically clean.

The setup: run a small VPS at a provider like Hetzner (German provider, EU-regulated), install WireGuard on it, and configure your home network to tunnel through it. Your home services become accessible from anywhere without being exposed directly to the internet.

Pi-hole

If you have any home server hardware, Pi-hole is one of the highest-value installations you can make. It acts as a DNS sinkhole at the network level — blocking ad and tracker domains before they ever load, for every device on your network, without configuring anything on individual devices. Running it on a Raspberry Pi is the canonical setup, but any Linux machine works.

Messaging

Social media

There is no privacy-respecting equivalent of Facebook, Instagram, or Twitter with comparable network effects. The mitigation options are limited:

• Use Firefox Multi-Account Containers to isolate social media tabs — social sites cannot track your browsing outside their container.
• Log out when not actively using platforms.
• Be deliberate about what personal information you share.

For longer-term migration, the Fediverse — Mastodon, Pixelfed, Bookwyrm — is a federated network of open source social platforms. The model is sound: decentralized, user-controlled, no advertising. The network is growing. Worth building a presence there even if you maintain legacy accounts.

Why local AI matters

Cloud AI services — ChatGPT, Claude, Gemini, and most others — send your queries to third-party servers. For many tasks this is acceptable. For sensitive work, confidential documents, private research, or situations where you simply do not want your queries logged, local models are substantially better.

The current state of open-weight models (Llama 3, Mistral, Qwen, Gemma) is genuinely capable for a wide range of tasks — summarization, drafting, coding assistance, research questions, document analysis. You will not match the largest frontier models on home hardware, but for everyday work, local models are more than adequate.

Starting point: Raspberry Pi 5 (16 GB)

A Pi 5 with 16 GB RAM is a reasonable entry for lightweight local AI. It handles 7B-parameter quantized models for everyday tasks at acceptable speed. Not fast — but functional, and runs continuously on minimal power.

The next step: a dedicated home server

For more demanding workloads — larger models, image generation, document processing, running multiple services simultaneously — a dedicated machine with a GPU is the right investment. Current pragmatic options:

Either machine can run: Ollama with larger models (13B–34B range depending on VRAM), Nextcloud for file storage, Immich for photos, and any other self-hosted services you want. Run everything in Docker containers for portability and easier maintenance.

NAS for long-term storage

Pair a home server with a NAS running TrueNAS or OpenMediaVault for resilient, self-owned storage. A Raspberry Pi with attached USB drives running OpenMediaVault is a low-cost starting point. Purpose-built NAS hardware is more reliable for long-term use.

A realistic sequence that prioritizes impact and manages disruption:

Week 1 — Identity foundation

Week 2 — Storage and browsing

Week 3 — Communications and social

Week 4+ — Infrastructure