Skip to content

The website care plan guide for agencies

Short answer

A website care plan is a recurring service covering updates, backups, security, monitoring and support for a site you built or inherited. Good plans define exactly what's included, separate monitoring from fixing, set response times, and show the client evidence each month. Automate the checks, keep humans for fixes, and report what actually happened.

What a care plan is for

Most small-business websites are launched and then left alone until something breaks. A care plan turns that into a predictable service: the client pays monthly, the agency keeps the site healthy and responds when something goes wrong. For the agency it's recurring revenue that smooths out project work; for the client it's someone accountable.

The difficulty is that a well-run care plan looks like nothing happening. When the site is up, mail is delivered and pages load, clients start to ask what they're paying for. The rest of this guide is about defining the work clearly and making it visible.

What to include

Write the scope as a list of specific, checkable activities rather than 'maintenance'. A typical baseline:

  • Software updates: CMS core, themes and plugins, on a stated cadence, with a rollback plan.
  • Backups: frequency, retention and where they're stored, plus a restore test at least quarterly.
  • Availability checks: the site is requested on a schedule and someone is alerted when it fails.
  • Email DNS health: SPF, DKIM and DMARC records exist and haven't changed unexpectedly.
  • Certificate and domain expiry: renewals are tracked, not assumed.
  • Broken links: periodic crawl of the site and a list of links to fix.
  • Security: login hardening, malware scanning where the platform supports it, and a response process.
  • Support time: a defined number of minutes or small tasks per month, with what counts as a task.
  • A monthly report.

Equally important is what's excluded: new features, content production, redesigns and third-party outages are usually billed separately. Writing that down prevents most care-plan disputes.

Structuring tiers

Three tiers are common because they let a client pick a middle option. Differentiate them on things that cost you time, not on cosmetic extras:

  • Check frequency and response time: daily checks and next-business-day response at the entry tier; more frequent checks and same-day response higher up.
  • Included support time per month.
  • Update cadence and whether updates are tested on staging first.
  • Report depth: a short summary versus a detailed report with recommendations.

Price from your costs: tools per site, your time per month for updates and reporting, and a buffer for incidents. Avoid promising response times you can't staff, especially out of hours.

Monitoring versus fixing

Separate the two in your contract and your tooling. Monitoring is automated detection: requests to the site, DNS lookups, expiry checks, crawls. It should run whether or not anyone is at their desk and produce a record of every check. Fixing is human work triggered by what monitoring finds.

Be careful with the language you use about monitoring. A check every few hours doesn't measure continuous uptime; if you tell a client '99.9% uptime', you need data that supports it. It's more defensible to report what you measured: checks performed, checks failed, issues opened and how quickly they were resolved.

Proving the work each month

The monthly report is the product the client actually sees. Keep it short and factual:

  • Period covered and sites included.
  • Checks performed and the success rate.
  • Issues found, with when they were opened and resolved.
  • Email DNS status: whether SPF and DMARC are present and valid, and any changes.
  • Domain and certificate renewal dates where known.
  • Updates applied and support tasks completed.
  • Two or three recommendations, such as moving DMARC from p=none to quarantine once reports are clean.

If a section has no data, say so. 'No checks recorded in this period' is honest and prompts a conversation; a made-up number undermines every other figure in the report.

Send reports in your brand, through a link the client can open without an account, and keep the link expiring so old reports don't circulate indefinitely.

Onboarding a new care-plan client

  • Get access: hosting, DNS, registrar, CMS admin and any email provider, stored in your password manager.
  • Verify ownership of the domain for your monitoring tools, typically with a DNS TXT record.
  • Baseline: record current DNS records, certificate issuer, domain expiry and a first crawl.
  • Fix the obvious: duplicate SPF records, missing DMARC, broken links from the first crawl.
  • Agree the alert path: who you notify on the client side, and when.

The baseline is valuable on its own. The first report showing that a client's SPF record exceeded the lookup limit, and that you fixed it, often justifies the plan for the year.

Response times and incident handling

Response time is the promise that costs the most, so define it precisely. State whether it's measured from the alert or from the client's message, whether it covers weekends and public holidays, and what 'response' means: acknowledging the problem, not necessarily fixing it. A small agency can credibly offer next business day at the entry tier and four business hours higher up; promising one-hour response around the clock means someone must always be on call.

Write a short incident routine and follow it every time: confirm the problem from a second location, check the host's status page, tell the client what you know and when you'll update them, fix or escalate to the host, then record cause and resolution. The record becomes the issue entry in that month's report.

Use severity to decide who gets woken up. A site returning errors on the homepage is urgent; a changed MX record might be urgent or might be the client's planned email migration; a broken outbound link can wait for the weekly review. Alert rules with a severity threshold and quiet hours keep the urgent channel meaningful.

Tools and costs per site

Keep a simple per-site cost sheet: monitoring, backups, security scanning, any premium plugin licences you resell, and your time. Time is usually the largest item. Estimate minutes per month for updates, reviewing alerts, the report and a share of incident time, then multiply by your internal hourly rate.

Automate whatever is checking rather than judging. Requesting a site, reading DNS records, comparing them with last month, looking up registry expiry and crawling for broken links are all mechanical. What you sell is the judgement and the fix, and the evidence that both happened.

Review the plan yearly with each client. A care plan that started with five pages and one mailbox may now include a shop, a booking tool and three services sending email, which changes both the risk and the work.

When to end or change a plan

Sometimes the right move is to stop. If a client repeatedly declines security updates or recommendations, put the risk in writing in the report and offer a lower tier that excludes the work they won't approve. A care plan that can't be delivered as written is worse for both sides than no plan.

Frequently asked questions

Try it on a real domain — free, no account needed.