retainer_design v1.0.0

name: retainer_design description: Design a consulting or agency retainer that holds up month after month: pick the retainer type (access, work, deliverables, or hybrid), scope the monthly work in hours before outputs, price from loaded delivery cost and target margin with a scope-creep buffer, set hours-bank and rollover rules with a cap and an expiry, fence out-of-retainer work with a written approval and overage rate, and write pause, renewal, and reporting terms. Ships a pricing and bank calculator, rollover and scope-fence rules, and a worked agreement skeleton. Use when someone asks to create, price, fix, or renegotiate a retainer, or when a retainer keeps losing money or absorbing extra work.

Retainer Design

Purpose

A retainer fails in predictable places. The fee was set from a competitor price instead of delivery cost. The scope was written as outputs with no hours behind them, so nobody knew the work was overrunning until margin was gone. Unused hours rolled over without a cap and arrived as a large debt in the busiest month. Extra requests were done first and argued about at invoice time.

This skill designs the retainer in the order that prevents those failures: type first, hours second, price third, bank rules fourth, scope fence fifth, then term and reporting. It ships bin/retainer_calc.py, which prices a retainer from loaded cost and simulates the hours bank, references/agreement-skeleton.md, and examples/worked-retainer.md.

This is commercial design, not legal advice. Have an attorney review the agreement once for high-value work, regulated industries, or work where ownership of intellectual property matters.

Workflow

1. Pick the retainer type, and write the choice into the agreement

The type decides what the fee buys and what a quiet month means.

• Access retainer: the client buys availability, priority, and a response time. The fee is earned in the period whether the client uses it or not. State that sentence in the agreement. If it is missing, a quiet month can turn into a refund argument.
• Work retainer: the client buys a block of hours or a defined set of tasks per period. Track use against the block and apply the bank rules in step 5.
• Deliverable retainer: the client buys named outputs per period, for example four articles and one report. Price it from the hours those outputs take, per step 2, even though the client sees outputs.
• Hybrid: a fixed core (a named deliverable or a standing meeting and report) plus an hours bank for requests. Name what belongs to each part. A hybrid with no split is a vague retainer with two labels.

Pick one primary type. If the client wants guaranteed access and a work block, that is a hybrid with an access premium, priced in step 3.

2. Scope in hours first, outputs second

List every recurring item and put hours on it before any price is discussed: each deliverable broken into its roles, standing meetings, reporting, revisions, account management, and the small requests that arrive between deliverables.

• Count revision rounds. Write the included number down. Extra rounds use the bank or the overage rate.
• Count meetings and reporting as delivery. A weekly hour-long call with prep and follow-up is often 6 to 8 hours a month, not 4.
• Use tracked history where it exists. Where it does not, estimate per item and mark the estimate as an estimate, then re-measure after three months.
• Add the items up. That total is the scoped hours. Keep a per-role split, because a retainer staffed by a senior at a junior price loses money on every hour.

3. Price from loaded cost and target margin

Never price from salary alone and never copy a market price without checking it against cost.

1. Build a loaded cost per hour per role: pay plus payroll taxes, benefits, insurance, and per-head costs, divided by productive hours (paid hours minus holidays, leave, training, and non-delivery time).
2. Add client-specific fixed monthly costs: tools, subscriptions, and subcontractors bought for this client.
3. Apply a scope-creep buffer to the hours before pricing. Working range from practitioner sources: 10 to 20 percent. Use the high end when the client has a history of extra requests, the low end for a stable, well-defined scope.
4. Divide total buffered delivery cost by (1 minus target delivery margin). Practitioner benchmarks put a healthy delivery margin at 55 to 60 percent or higher, with some agencies targeting 70 percent on individual retainers so the blended result survives overhead. Pick a target and write it down. A fee that only works at 100 percent utilization was priced on a fantasy, since healthy billable utilization for delivery staff is well below that.
5. Add an access premium only when the retainer really buys priority: faster response, senior-only staffing, or reserved capacity at short notice. Apply a commitment discount only for a longer minimum term, and keep it small: practitioner guidance puts a reasonable long-commitment discount at 5 to 10 percent. A deep discount gives away the security the client is paying for.
6. Run bin/retainer_calc.py case.json (see examples/worked-retainer.md for the input). Check the implied rate per scoped hour, the overage rate, and the effective margin if buffered hours are fully worked. If the effective margin at full buffer is below target, raise the fee or cut scope before presenting anything.
7. Present three tiers when the client is choosing size: a core tier, a recommended tier, and a larger tier. Tiers should differ in capacity, response time, or risk cover, not only in a longer list of tasks.

Sanity checks before the fee leaves the building:

• Fulfillment cost above about 60 percent of the fee leaves too little room. Reprice or reduce scope.
• One person carrying a retainer that takes more than about 60 percent of their monthly billable capacity has no room for other work or absence. Split staffing or raise the fee.
• If the client will not pay the cost-based fee, cut deliverables to fit their budget. Do not keep the scope and drop the price.

4. Write the monthly scope as an in list and an out list

Put the scope boundary in a schedule or appendix the two sides fill in together, not only in contract prose.

• In list: each recurring deliverable with quantity and hours, included revision rounds, channels the client can use to request work, and the response times in step 7.
• Out list: the work a reasonable client would assume is included but is not. Name it plainly: new builds, campaigns, redesigns, work for other departments, rush work outside agreed hours, third-party costs. The out list does more protective work than the in list.
• One-off project work that appears inside a retainer month is not absorbed. It gets its own quote or statement of work, per step 6.

5. Set the hours bank and rollover rules

Unused time is where access and work retainers split, and where uncapped rollover creates an unstaffable debt. Choose one rule, write the numbers in, and show the client the simulation before signing.

• Use it or lose it (default for access retainers, cleanest for work retainers): hours expire at period end, no carry. Explain why in one plain sentence: the fee reserved capacity in that period, and unused capacity in a past period cannot be delivered twice later.
• Capped rollover (use when the client pushes and the work is lumpy): carry at most a stated percent of the monthly allowance, practitioner examples run from 25 to 50 percent, the carried hours expire after one following period (or a stated number of days, for example 60), and banked hours are spent first so old hours cannot pile up behind new ones. Underuse never reduces the fee.
• Never allow unlimited rollover, and never allow rollover with no expiry. Both turn a monthly fee into an open-ended hours debt.

Run the bank before agreeing to any rollover request. Put the client's expected usage pattern into monthly_usage_hours in the calculator and read the expired, overage, and ending bank figures. If a quiet quarter followed by a spike produces a bank the team cannot staff, lower the cap.

Worked check from the calculator demo: 40-hour allowance, 50 percent cap, usage of 32, 18, 55, and 40 hours over four months gives 8 hours banked after month 1, 10 hours expired in month 2, no overage in the spike month because the bank covered it, and a 5-hour ending bank. That is a rollover rule doing its job: smoothing a spike without creating a large debt.

6. Fence out-of-retainer work

A fence only works if it triggers before the work starts.

1. Either side can flag a request as possibly out of scope. Flagging is normal business, not a complaint.
2. The provider confirms in writing whether it is in or out, with the cost if it is out: bank hours, a fixed quote, or the overage rate.
3. No out-of-scope work starts without the client's written approval. Email approval is enough if the agreement says so.
4. Bill overage at a stated rate. Common practice is the standard rate plus 10 to 20 percent for freelancers, or 1.25 to 1.5 times the implied retainer rate in agency agreements. Cheap overage trains the client to under-buy the retainer and treat the provider as on-demand cover.
5. Track time in stated increments (15 minutes is common) and report use in the monthly report, so an overage invoice is never the first time the client hears about it.
6. Apply the fence the same way every month. A boundary waived twice becomes the real boundary.

7. Set response times, pause, term, and renewal

• Response times: acknowledge requests within a stated number of business days, with a separate, faster rule for a defined emergency. Define emergency in words (for example site down, payments failing). State the hours during which requests are monitored. Silence on this point becomes an assumption of round-the-clock cover.
• Minimum term: 3 to 6 months is a common starting range, long enough to judge fit, short enough that a bad fit does not trap either side. Longer terms should trade for something: a discount in step 3, a rate lock, or a larger rollover cap.
• Billing: invoice in advance, on a stated day of the month, with stated payment terms. Work pauses when an invoice is overdue past a stated point. State whether the final month on termination is paid in full or pro-rated.
• Pause: if a pause option exists, cap it: for example one pause per year, up to one month, with a stated notice period, and the term extends by the paused time. An uncapped pause is a cancellation with extra steps. If no pause is offered, say so.
• Termination: written notice, 30 days is a common figure, plus handover duties: files, access, and documentation delivered within a stated number of days. Add termination for cause with a cure period, and a kill fee or payment for work in progress on a mid-deliverable exit if the work justifies it.
• Renewal and price review: state whether the retainer auto-renews, and add a scheduled review, at least annually, where fee, hours, and scope can change. Also add a usage trigger: if actual hours run above scoped hours for two or three periods in a row, either side can open a reprice or re-scope discussion using the tracked hours as evidence. One heavy month is noise. A run of them is a pricing problem.

8. Report what the retainer bought, every period

Send a short report with the invoice or just after period end. It has five parts:

1. Hours used against the allowance, by item or deliverable, plus bank in, bank out, expired, and overage hours.
2. Deliverables completed, named.
3. Requests handled and response performance against the agreed times.
4. Next period's planned work and any capacity warning (for example the bank plus expected work exceeds staffing).
5. One line on value the client can repeat internally: the outcome the work supported, stated plainly without inflated claims.

Reporting is the renewal tool. A client who can see use, output, and the bank balance renews on evidence. A client who only sees an invoice renews on mood.

9. Review the retainer quarterly

Re-run the calculator with actual hours per role for the last three months. Read the effective margin, the overage pattern, and the bank balance.

• Margin below the target for two quarters in a row: reprice, re-scope, or exit in that order of preference, using the decision sequence in the Operating Rules.
• Repeated overage in the same category: that category belongs inside a larger retainer or on its own quote, not in permanent overage.
• Repeated expiry with low use: offer a smaller tier or convert part of the fee to an access retainer. Do not let a client keep paying for capacity they never use without raising it in writing at least once. Quietly collecting for unused work reads badly when it surfaces.

Output Contract

Deliver a retainer pack with:

• The type decision (access, work, deliverable, or hybrid) with one sentence on why it fits this client
• The scope schedule: in list and out list, each recurring item with hours per role, included revision rounds, and request channels
• The pricing sheet from bin/retainer_calc.py: loaded costs, buffer percent, target margin, monthly fee, implied rate per scoped hour, overage rate, and effective margin at scoped and at buffered hours
• The bank rules: allowance, rollover cap in hours and percent, expiry, bank-spent-first rule, and the bank simulation for the client's expected usage pattern
• The scope fence: flag, written confirmation, written approval, and overage billing steps, with the rates filled in
• The commercial terms: minimum term, advance billing day and payment terms, pause rule or a statement that no pause is offered, termination notice and handover, renewal and scheduled price review, and the usage trigger for an early reprice discussion
• A filled agreement skeleton from references/agreement-skeleton.md, with no bracketed fields left blank
• The monthly report template, filled in for the first period or ready to fill

Operating Rules

1. Price from cost, then check against the market. Never price from the market alone. A fee with no cost build behind it cannot be defended, discounted safely, or repriced with evidence.
2. State the type in the agreement, in plain words. In particular, an access retainer must say the fee is earned each period regardless of use. Blurring access and work is how quiet months become refund disputes.
3. No uncapped rollover, ever, and no rollover without an expiry. If a client insists on unlimited rollover, convert the offer to an hours block sold in advance with a stated expiry, and price it as a block, not as a retainer.
4. No out-of-scope work without written approval given before the work starts. Work done first and priced later trains the client to skip the fence and puts the provider in a payment argument they created.
5. Underuse does not reduce the fee, and overuse does not get absorbed in silence. Both numbers appear in the monthly report. A provider who hides underuse and bills surprise overuse loses trust on both ends.
6. When the fee cannot cover the scope, cut scope. Dropping the price while keeping the deliverables converts a negotiation win for the client into a monthly loss that ends in resentment or quiet quality cuts.
7. Reprice on evidence, on a schedule or a stated trigger, never on a feeling. Show tracked hours against scoped hours by deliverable. Offer the client a choice alongside any rise: full scope at the new fee, reduced scope at the current fee, or extras priced per item.
8. Keep total commitments in view. Add up buffered hours across all retainers before signing a new one. A book of retainers that assumes every client is quiet in the same month is a capacity accident waiting for a shared busy month.
9. Track time even on deliverable and access retainers. Hours are the only evidence that settles a reprice, a scope argument, or a renewal. An untracked retainer can only be argued from memory.
10. Never invent a number. Loaded costs come from payroll or a stated estimate, hours from tracking or a stated estimate, and benchmarks are presented as ranges with their source named in the pricing sheet notes. An unsourced precise benchmark gets cut.