Service 17

Cloudflare set up properly — protection that lets customers through

Cloudflare is in front of your site in ten minutes — and misconfigured just as quickly. Then the bot protection blocks your payment provider, the cache serves one customer’s basket to the next, and the firewall protects a server that is still wide open next door. We configure Cloudflare every day for websites and shops we operate, and we know which rules fit TYPO3, WordPress and Shopware: stop attacks, let customers and interfaces through.

WAF rulesBot Fight ModeRate limitingCache rulesOrigin protection

What we find most often

Three mistakes come up again and again. The connection between Cloudflare and the actual server still runs unencrypted, although the visitor sees a valid certificate. The address of the server is still public, so any attacker simply walks past Cloudflare — the whole firewall is then decoration. And the bot protection is switched on but not set up: it stops scanners, but equally the callback from the payment provider, the synchronisation with the ERP or the monitoring. Whoever does not know this spends days looking for the reason why payments are not confirmed.

Add cache rules that, in a shop, deliver one customer’s basket to the next, convenience features that silently break storefronts, and an “Under Attack” mode that was never switched off after the attack. All of these look fine from the outside. That is exactly why we look first, before we touch anything.

Foundation: encryption, and a server that only Cloudflare can reach

We start with the zone itself: which records run through Cloudflare and which do not, whether encryption reaches all the way to the server, whether DNS is protected against manipulation, and whether the server address is somewhere it does not belong. That sounds like basics — these are the points where a setup stays open despite Cloudflare.

Then the server is sealed off: it accepts requests only from Cloudflare’s networks, and on request Cloudflare additionally identifies itself to the server with a certificate. We also make sure that logs, login throttles and statistics still see real visitors instead of Cloudflare addresses — a detail that is missing surprisingly often, and without which every server-side throttle runs into nothing.

Firewall and Bot Fight Mode: rules that hit attacks, not customers

Our web application firewall works in layers: first Cloudflare’s managed rulesets against known attack patterns, on top custom rules that fit your system — admin access only for those who need it, requests for files that do not exist on your site dropped immediately, paths for interfaces and payment providers explicitly allowed. Rate limiting caps login attempts, search queries and form submissions per sender without a normal visitor ever noticing.

For bot protection the plan decides the tools. The simple Bot Fight Mode has no exceptions; wherever interfaces, webhooks or monitoring are involved — that is, in practically every shop — you need Super Bot Fight Mode with verified bots on the allow list and custom rules that exclude certain paths from bot protection. We set this up so that the scanner sees the challenge page and the payment provider does not.

Rules are never finished. After going live we read the security events: what was blocked, what was only challenged, and was there anything that should have passed? False positives become exceptions, new patterns become rules. For forms we use Turnstile instead of classic captchas — no picture puzzles, verified on the server, as on this website.

Cache, operation and your account

Static files can be held by Cloudflare for a long time — this is where the biggest gain in load time and server load lies. For page content the system decides: a TYPO3 or WordPress page can be cached at the edge if publishing clears the cache deliberately; a shop needs rules that always bypass basket, checkout and customer account and work together with Shopware’s own cache. We tie cache clearing to your workflows so that neither editors nor the shop ever deliver stale content.

The Cloudflare account belongs to you — we work with access, not ownership, exactly as with ad accounts. Every setting is documented and justified so that someone else can understand it later. And because attack patterns, your interfaces and Cloudflare’s features change, the zone is part of the regular review cycle of our support.

A note for context: Cloudflare is a US provider, and your visitors’ traffic passes through its network. Using it belongs in your privacy policy and requires a data processing agreement, which Cloudflare provides. We supply the technical details — the data protection assessment stays with you or your data protection officer.

Frequently asked questions

Is the free Cloudflare plan enough?

For a company website without interfaces, often yes. As soon as payment providers, ERP systems or monitoring access the site automatically, the basic plan reaches its limit — above all because the simple Bot Fight Mode cannot be excluded for individual senders. Then the Pro plan with Super Bot Fight Mode and full managed rulesets is worth it. Cloudflare changes the limits per plan from time to time; we check them during setup.

Does Cloudflare block real customers?

Set up wrongly: yes. Set up properly: hardly ever — and if so, visibly. We work with graded actions: clear attacks are blocked, doubtful cases get a short, mostly invisible challenge, known partners are on the allow list. After the start we read the security events specifically for false positives and turn them into exceptions.

What exactly is Bot Fight Mode — and why does it interfere with interfaces?

Bot Fight Mode detects requests that look like a machine and challenges or blocks them. That hits scanners and scrapers, but equally every legitimate automation: the payment confirmation from your payment provider, the product synchronisation with your ERP, the monitoring service. In the simple mode there are no exceptions for this; Super Bot Fight Mode allows them. Exactly this fine-tuning is the difference between switched on and configured.

Do we need Cloudflare if you host our site?

Not necessarily. Our servers in Germany are hardened and monitored, and for many websites that is enough. Cloudflare adds a line of defence in front of the server: attacks and mass traffic are intercepted before they cost your resources, and rules can be adjusted in seconds when something new appears. For shops, exposed portals and projects with an international audience we recommend it; for a regional company website we discuss it openly.

Who owns the Cloudflare account?

You do. The zone runs in your account, and we are invited as a member with the necessary rights. Domains, certificates and rules stay with you, even if our cooperation changes one day. If you already have an account with a grown configuration, we take over the zone as it is and clean it up.

What happens during a DDoS attack?

Basic protection against mass traffic is included in every Cloudflare plan and works automatically. For attacks on the application level — many seemingly normal requests to expensive pages such as search or checkout — our rate limiting and firewall rules apply; in an emergency we switch on the “Under Attack” mode, which briefly checks every visitor, and switch it off again as soon as it is over. Permanently enabled, it belongs nowhere.

More answers in our FAQ

Cloudflare set up properly — protection that lets customers through: Let us get specific.

Tell us your starting point — you get an honest assessment, not a sales pitch.

Call Start a project