An agency that manages multiple WordPress sites can host them in four ways: an agency plan on a managed WordPress platform, its own reseller account or cPanel server, cloud servers it builds itself, or its existing hosts operated by a white-label infrastructure partner. The right choice depends on three things more than on the host: who carries the pager, who talks to the client, and what happens to the agency’s margin. A small portfolio of similar brochure sites usually belongs on a platform agency plan. A portfolio that has turned a senior developer into the de facto system administrator needs a support model, which a new host alone will not supply.
What are the hosting models for an agency with multiple WordPress sites?
An agency plan on a managed WordPress platform. Managed WordPress platforms such as WP Engine and Kinsta sell plans built for agencies. On this kind of plan the platform runs the servers, the caching layer and platform-level backups. The agency typically gets one dashboard and no root access. Platform support covers the platform. Site-specific work, such as a plugin conflict, a failed release or the conversation with the client, generally stays with the agency.
A reseller account or the agency’s own cPanel server. Reseller hosting is an account on a hosting company’s server that lets the agency create and bill separate client accounts. A whole server running cPanel and WHM (WebHost Manager, the administration layer of cPanel) gives the agency the same tools with full control. Both keep a hosting margin inside the agency. On a whole server the agency is also the system administrator, responsible for patching, mail, malware cleanup and restores. The knowledge base explains when a reseller account is enough and when you need a whole server, and what a white-label hosting business involves on cPanel.
Cloud servers the agency builds and runs. The agency rents a VPS (virtual private server) or an account with a public cloud such as AWS (Amazon Web Services) and designs the stack itself. Control is complete, and so is responsibility for everything above the virtual machine. The knowledge base covers how agencies manage client AWS accounts without a cloud team.
A white-label infrastructure partner. An operations partner runs the hosting layer behind the agency’s brand: monitoring, patching, backups, on-call and incident response. White-label means the client sees the agency’s name while another company does the work. The sites can stay on their current hosts. The agency keeps the client relationship, the development work and the care plan. A care plan is the monthly maintenance retainer an agency sells to its clients.
WordPress multisite is sometimes raised as a fifth option. Multisite runs many sites from one codebase and one database, which removes the separation between them. It rarely suits unrelated clients, who expect their own data, plugin set and a clean exit. The knowledge base covers running WordPress multisite in production.
How do the four agency hosting models compare?
The four agency hosting models differ mainly on who carries the pager, who talks to the client and where the agency’s margin comes from.
| Criterion | Platform agency plan | Reseller or own cPanel server | Self-built cloud | White-label infrastructure partner |
|---|---|---|---|---|
| Who carries the pager | The platform for platform faults, the agency for each site | The agency | The agency | The partner, within the written scope and coverage window |
| Who talks to the client | The agency | The agency | The agency | The agency by default |
| Server access | No root access | Limited on a reseller plan, full on a whole server | Full | Depends on the existing host |
| Where margin comes from | The care plan, sometimes a markup on the plan | Hosting resale plus the care plan | Hosting resale plus the care plan | A care plan backed by documented operations work |
| Main risk | Platform limits on plugins and server settings | One person holds all the knowledge | One person holds all the knowledge, plus design mistakes | A second vendor, and a scope that must be written down |
| Best fit | Similar sites, no need for server access | Agencies that want hosting as a product line and can staff it | Agencies with real operations skills in-house | Agencies with sites clients depend on and developers who should be building |
Who carries the pager when a client site goes down?
Carrying the pager means being the person who is alerted, and expected to act, when a site fails outside office hours. On a platform plan, a reseller account, an agency-run server or self-built cloud, the agency carries the pager for anything specific to a site. A platform or a reseller host fixes its own faults. It generally does not investigate why one client’s checkout broke after an update.
In many agencies the pager is an unwritten arrangement: a founder’s or a lead developer’s mobile phone. An agency can test its arrangement by writing down four answers for a Saturday night outage:
- Who is alerted, and by what system, before the client notices.
- How quickly that person is expected to acknowledge the alert.
- Whether that person has the access needed to act.
- Whether a runbook exists. A runbook is a written procedure for a known failure on a specific environment.
If an agency’s clients do not sell or publish at night, business-hours coverage can be the honest answer, provided the care plan says so. The trade-off between staffing a rotation and contracting one is covered in Hire a DevOps Engineer or Outsource Operations.
What happens to agency margin under each hosting model?
An agency earns two different margins on hosting. The first is resale margin, the difference between what the host charges and what the client pays. The second is care-plan margin, the difference between the retainer and the hours the agency spends keeping the site healthy. Only the first shows on an invoice.
An agency can measure the hidden cost in one month. Log every hosting-related interruption: tickets, renewals, restores, host support chats and after-hours alerts. Multiply the hours by the billable value of the person who handled them. Then compare that figure with the hosting margin for the same month.
The cost of any model moves with the same drivers: the number of client sites and how much they differ, traffic and storage, the isolation each site needs, the coverage window, backup retention and restore testing, licences, and migration volume.
A white-label partner’s fee is a real cost, so the margin moves from resale to the care plan. The model pays when the recovered senior hours are sold as project work, or when care plans are repriced around documented operations work. No model promises that outcome. In one example scenario, a 22-person agency with 14 WordPress sites consolidated five care-plan retainers and repriced them at an average 12% uplift, and every client renewed. That uplift was the agency’s own pricing to its clients, not a Stack Harbor fee or a typical result. For the structure behind provider quotes, see how managed cloud hosting is priced.
What are the risks of hosting many client sites yourself?
The risks of agency-run hosting come from operations, not from the choice of server.
- Blast radius. Blast radius is how far one compromised or overloaded site can affect its neighbours. Sites inside one hosting account share a user, a quota and a backup. The knowledge base weighs one cPanel account per site against many sites in one account.
- Update discipline. One bad plugin release can break several client sites in the same afternoon. The knowledge base describes how to manage WordPress updates across a fleet of sites.
- Backups nobody has restored. A backup is proven only by a test restore.
- Knowledge in one head. When the developer who built the server leaves, the agency inherits an undocumented system.
- Promises in the care plan. Clients hold the agency responsible, whatever the host’s terms say.
When should an agency choose each model, and when not?
Platform agency plan
Choose a platform agency plan when the sites are similar, none needs custom server settings, and nobody on the team wants server access. Avoid it when several clients need plugins or configurations the platform does not allow, or when clients expect the agency to answer at night.
Reseller account or own cPanel server
Choose a reseller account or a cPanel server when hosting is a deliberate product line with its own pricing, and the agency can staff patching, mail and abuse handling. Avoid this model when the system administrator is also the agency’s best developer.
Self-built cloud
Choose self-built cloud when the agency employs people with real operations experience and its clients need architectures a control panel cannot express. Avoid it when the only motivation is a lower server bill, because staff hours often outweigh the saving.
White-label infrastructure partner
Choose a white-label infrastructure partner when client sites carry revenue or reputation, the portfolio is spread across several hosts, and senior developers are losing project time to tickets. Avoid a partner when the portfolio is a few low-risk sites, because a platform plan is usually enough and adds no second vendor. Avoid it also when the business model is reseller margin from a public price grid, or when the team wants to own operations. A partner operates infrastructure and does not build or fix client sites.
Does moving every client site to a better host solve the problem?
Moving every client site to a better host often leaves the problem in place, because the problem is usually the missing support model. One agency portfolio consolidation example scenario says it directly: “the host is rarely the actual problem”.
In that one example scenario, the agency ran fourteen sites across four hosts, and consolidation took ten weeks. Average measured edge downtime was 22 minutes per site, against a four-hour window communicated to each client. Inbound client support tickets were down 68% at 90 days. Those figures describe one portfolio and are not a promise for another.
Decide who owns support first, then consolidate only where a move lowers cost or risk. The difference between buying servers and buying operations is explained in Managed WordPress Hosting vs Managed WordPress Operations.
What should an agency ask before choosing a hosting model or provider?
Take these questions into an internal review or a vendor call.
- Which sites would cost a client money or reputation within an hour of failing?
- Who is alerted when one of those sites fails at night, and how quickly must that person acknowledge?
- Does each site have its own account, database and resource limits, or do sites share them?
- When was a backup last restored as a test, and who holds a copy outside the host?
- What does each care plan promise in writing, and does the hosting setup support that promise?
- Does the provider require a move, or will it operate the hosts already in place?
- Is the price tied to a written scope, and how is work outside that scope quoted?
Where Stack Harbor fits
Stack Harbor is a Canadian-owned managed cloud operations provider (MSP) for production environments, headquartered in Montréal. It serves agencies, SaaS companies and publishers, and it is not an office IT provider. Its Agency Infrastructure solution is written for agencies of 5 to 30 people operating 5 to 30 client sites across WordPress, cPanel and cloud hosting.
Stack Harbor operates the agency’s existing hosts and consolidates only where a move saves money or risk. It does not replace the agency’s developers. White-label is the default of three partnership models: white-label, co-managed and direct referral. Within the coverage window the scope sets, on-call goes to Stack Harbor, with documented escalation, monitoring with alerting and a runbook per environment. A trained on-call rotation works to runbooks designed by senior engineers, and senior engineers own Severity 1 incidents. Technical requests run through the support desk, and the telephone reaches the account manager. On plans that include incident coverage, a Severity 1 incident receives an acknowledged response within 15 minutes. Resolution is a target, not a commitment.
For agencies that run their own servers, cPanel and WHM server management has a 24/7 Operations tier that includes agency portfolio support. It is the only tier of that solution with 24/7 monitoring and on-call. Where a move does make sense, sites migrate one at a time with staging, and every project under migration services ships with a written rollback plan. Client sites that need Canadian hosting can be placed in Stack Harbor’s two Montréal data centres. The trust page covers data residency and the service commitments.
Stack Harbor does not publish prices. Agency pricing pairs a base retainer with per-site or resource billing. An infrastructure consult and a portfolio walkthrough lead to a fixed scope and a fixed quote.
How should an agency decide on a hosting model?
An agency choosing a hosting model should start with the pager, not the server. An agency with a few similar, low-risk sites should stay on a platform agency plan. An agency that wants hosting as a product, and can staff it, should run its own server. An agency whose senior developers are absorbing hosting work should measure those hours for a month. If the hours cost more than the hosting margin earns, the agency should hand the hosting layer to an operations partner and keep the client relationship.
