Content Security Policy for a small business site: how-to
Short answer: A Content Security Policy (CSP) is a response header that tells browsers which scripts, styles, images and frames your pages may load, mainly to limit the damage if someone manages to inject malicious code. Start in report-only mode, collect violations for a week or two, adjust the policy to cover the tools you genuinely use, then switch to enforcement. Small sites break CSPs most often through inline scripts and third-party widgets.
Security scanners, hosting dashboards and agency audits regularly flag "missing Content-Security-Policy header". It's a reasonable recommendation, but switching it on carelessly can break your contact form, analytics or booking widget in one go. This step-by-step guide shows how to roll one out safely on a small business site.
What CSP does
MDN's guide to Content Security Policy describes it as a set of instructions from a website to the browser, restricting what the site's code is allowed to do. Its primary use is controlling which resources, especially JavaScript, a page can load, mainly as a defence against cross-site scripting (XSS), where an attacker injects malicious code into your site. MDN adds that a CSP can also help defend against clickjacking and help ensure pages load over HTTPS.
A policy is a list of directives. MDN's example:
Content-Security-Policy: default-src 'self'; img-src 'self' example.com
Here default-src 'self' allows resources only from your own origin by default, and img-src additionally allows images from example.com.
A strict policy can block inline script tags, inline event handlers and javascript: URLs, and allow only scripts carrying the correct nonce or hash. MDN notes the 'unsafe-inline' keyword can override the inline restriction, at the cost of much of the protection.
Step 1: inventory what your pages load
Before writing a policy, list what each type of page actually uses:
- Your own scripts and stylesheets
- Analytics and tag manager
- Fonts (self-hosted or from a font service)
- Embedded maps, videos, booking widgets, chat widgets, review badges
- Payment or checkout scripts
- Forms posting to third-party services
- Images from a CDN
Your browser's developer tools (Network tab) show the domains each page contacts. Do this for the homepage, a typical content page, the contact or booking page and the checkout if you have one.
Step 2: write a first draft policy
Start with your own origin as the default, then add sources directive by directive:
script-src: your domain plus analytics, tag manager, payment and widget domainsstyle-src: your domain plus any font or widget style sourcesimg-src: your domain, CDN, analytics pixels, map tilesfont-src: self-hosted or font service domainsframe-src: embedded maps, videos, booking or payment framesconnect-src: endpoints your scripts call (analytics collection, form handlers, chat)frame-ancestors: who may embed your site in a frame. MDN describes this as an effective protection against clickjacking;'self'or'none'suits most small sites.
Vendors often publish the domains their tools need for CSP; check each tool's documentation.
Step 3: deploy in report-only mode
MDN explains that CSP can be deployed in report-only mode: the policy isn't enforced, but violations are reported. The Content-Security-Policy-Report-Only header exists for this, letting you test or repair violations before applying an enforced policy. MDN notes that the report-to directive must be specified for reports to be sent, with endpoints defined in a Reporting-Endpoints response header; the older report-uri directive is deprecated but still seen. It also notes the report-only header isn't supported inside a <meta> element, so set it as an HTTP header at your server, host or CDN.
Even without a reporting endpoint, violations appear in the browser console as you browse the site, which is enough for a small site's first pass.
Step 4: read the violations and adjust
For each reported violation, decide:
- Legitimate and needed? Add the source to the right directive.
- Inline script? Move it into a file, or use a nonce or hash if your platform supports them. Avoid reaching for
'unsafe-inline'as a first fix. - Unknown or unwanted? Investigate. It may be an old tracking script, a browser extension (those can be ignored), or something that shouldn't be there.
Repeat until normal use of the site produces no violations you care about. Give it at least a week of real traffic if you can.
Step 5: enforce
Switch the header name to Content-Security-Policy, keep a report-only copy of any stricter future policy if you want to keep improving, and test key journeys straight away: forms, checkout, bookings, maps, chat, analytics.
Step 6: maintain it
- Adding a new tool? Update the policy as part of the installation, not after something breaks.
- Removing a tool? Remove its domains so the policy stays tight.
- Platform updates? CMS, theme or plugin updates can introduce new inline scripts; recheck after major updates.
Troubleshooting common breakages
- Contact form stops sending: the form posts to, or loads scripts from, a domain missing from
form-action,connect-srcorscript-src. - Analytics shows a drop: analytics collection endpoints are missing from
connect-srcorimg-src. - Embedded map or video is blank: add the provider to
frame-src. - Fonts fall back to defaults: the font host is missing from
font-src(and sometimesstyle-src). - Page builder features broken: many builders rely on inline scripts or styles; check what nonce or hash support your platform offers.
Other headers often flagged alongside CSP
Scanners usually list a few companions, such as HTTP Strict Transport Security, X-Content-Type-Options and Referrer-Policy. Each is simpler than CSP; read MDN's page for each before switching it on.
FAQ
Will a CSP slow my site down?
No meaningful effect on speed; the browser simply checks resources against the policy. The work is in getting the policy right.
Can I set CSP with a meta tag?
Some CSP directives can be delivered in a <meta> element, but MDN notes the report-only header can't, and features such as frame-ancestors need the HTTP header. Prefer the header.
My WordPress site uses lots of plugins. Is CSP realistic?
Yes, but expect more report-only iterations. Plugin-heavy sites often need 'unsafe-inline' for styles at first; tighten over time.
Does CSP replace keeping plugins updated?
No. It limits damage if something is injected; it doesn't fix vulnerabilities. Updates and good access controls still come first.
Got a scanner report flagging missing security headers and worried about breaking your site? Send it over and we'll plan a report-only rollout with you.