The question is not which model pays more. It is whether the client has a recurring decision to make.
The decision rule
Choose a retainer when a site needs reading on a schedule. Somebody must answer a question each cycle, and the work must be small and repeated.
Choose a project fee when the outcome is a thing, not a process. The scope can be written in advance, and the work ends at delivery.
A worked price
Meridian Landscaping, a fictional client, one site with a repository.
inputs
sites: 1
cadence: 14 days, so about 2.2 cycles a month
changes per cycle: 1.5 (measured over the last three months)
hours per cycle
read 0.33 h
decide 0.25 h
ship per change 0.75 h
report 0.50 h a month
rate $95 an hour
platform cost $99 a month, Pro for the owner
monthly hours
read 2.2 x 0.33 = 0.73 h
decide 2.2 x 0.25 = 0.55 h
ship 2.2 x 1.5 x 0.75 = 2.48 h
report = 0.50 h
total = 4.26 h
price
4.26 x $95 = $404.70
platform = $99.00
rounded retainer = $505 a month
Two numbers decide that price: the cadence and the change appetite. One more shipped change per cycle adds about $157 a month. Nothing else in the model moves as fast.
What the retainer includes, and what it does not
Included: the reading, the decision, the shipping of agreed changes, and the monthly update. The record of what each change did stays with the site.
Not included: new pages, redesigns, brand work, paid media, hosting and licences. Work caused by an unauthorised change to the site is also excluded.
The boundary clause, ready to copy
scope
- two cycles a month, one site
- up to two shipped changes each cycle
- one written update a month
- access, triage feedback, and the search reading
outside scope, quoted separately
- new pages or sections
- redesign work
- a third-party integration
- work needed because of a change we did not make
review
- if shipped changes average more than two a cycle for two months,
we agree a new rate before the third month
Why the review clause protects both sides
A retainer that silently absorbs project work becomes a discount. A client who never sees the boundary assumes everything is included.
Acceptance: the quote is complete when
- hours come from your own last three months, not from a market average
- the cadence in the plan matches the cadence in the price
- platform cost is a line, not an absorbed expense
- the excluded list names at least four items
- a review trigger exists, with a number
Troubleshooting
| Symptom | Likely cause | What to do |
|---|---|---|
| The retainer feels unprofitable by month three | Change appetite grew | Trigger the review clause |
| The client disputes the hours | The plan never named the cadence | Show the cycle counts, then agree a rate |
| Every month needs a new quote | The scope is project work in disguise | Split the project out, keep the retainer small |
| A client asks for a guarantee | They want a different contract | Say so, and offer the record instead |
| Two clients, same price, different effort | One has two sources missing | Price the access gap into onboarding |
Limits
These hours are a model. Track one cycle honestly, then replace every estimate with your own numbers.
Neither model promises a result. A client who needs a guarantee is buying something else. Saying that early is cheaper than saying it later.