Budgeting a network by cost per site per month

Judge every option on one figure: fully loaded cost per site per month. That is the hosting share, plus the domain spread across twelve months, plus the certificate where it is billed separately, plus the line almost nobody writes down, which is your own labour valued at whatever an hour of it is worth. Sticker price alone ranks your choices in the wrong order.

The four lines that make up the figure

Split the time line in two, because the halves scale differently. Estate overhead (backup checks, patch rounds, invoice reconciliation, uptime watching) barely moves whether you run four sites or forty. Per-site work (content edits, mailbox fiddling, occasional breakage) grows one for one.

The model

The placeholders below carry no units on purpose; substitute your own figures.

H = hosting charge per month per plan   say 24
A = addresses each plan carries         say 6
n = sites actually live
d = domain renewal per year             say 14
c = certificate per year                say 0 when included
R = your hourly figure                  say 50
E = estate overhead, minutes per month  say 120
P = per-site work, minutes per month    say 12

per site per month =
    (H * plans needed) / n  +  d/12  +  c/12  +  R * (E/n + P) / 60

Worked example as the estate grows

                       n = 1      n = 6      n = 30
hosting share          24.00       4.00        4.00
domain (14 / 12)        1.17       1.17        1.17
certificate             0.00       0.00        0.00
time                  110.00      26.67       13.33
                     -------    -------     -------
per site per month    135.17      31.84       18.50

At one site the whole plan and all 120 minutes of overhead land on it: 132 minutes at 50 per hour is 110.00. At six the plan is full, so the hosting share falls to 4.00 and overhead splits six ways. At thirty (five plans) the hosting share holds at 4.00 while overhead drops to 4 minutes per site. The figure falls roughly sevenfold, and the plan charge is not what caused it.

Why the marginal site is cheaper than the first

Adding site 31 into spare capacity on a plan you already pay for costs 1.17 for the domain plus 10.00 of per-site work, so 11.17. That sits well under the 18.50 average, which is why averages flatter a growing estate. The exception is a site that crosses a plan boundary: it carries the whole new plan charge in the month it lands, 35.17, and that step amortises as the plan fills.

Two consequences. Buy addresses in block sizes that match your real intake rate, so boundaries get crossed deliberately rather than by surprise; the trade-offs sit under buying decisions. And attack the time line before the invoice line: halving overhead from 120 minutes to 60 saves more per site at n = 6 than any plausible change of plan.

Sanity checks

• Divide the plan by the sites you are running now, never by paid-for capacity sitting idle.
• Use renewal figures for the domain and the certificate, not introductory ones. A model built on a one-off figure is wrong from month thirteen.
• Log your actual minutes for one month rather than guessing. Self-estimates of admin time run low, consistently.
• Add a line for anything else that renews: paid themes, licences, monitoring, off-site backup storage.
• Recompute when n moves by more than about a quarter. Below that, the figure shifts less than the noise in your time log.

Estates that look expensive per site are usually small estates carrying fixed overhead, not badly bought ones. Before you switch anything, check whether the number simply needs more sites underneath it.