Most marketing retainers written for software companies are a list of outputs. Four articles a month, two landing pages a quarter, a report. That tells you what will arrive. It tells you nothing about whether any of it will move the thing you are actually worried about, which is usually that qualified people are not finding you, or they are finding you and not understanding what you sell.
If you are scoping your first retainer, the proposal in front of you is the only artefact you get to judge before money moves. So it is worth knowing what a good one contains. If you are still deciding between a retainer and a hire at all, what a hire, a fractional engagement and an agency each actually cost is the comparison to read first.
A scope should name the problem, not just the deliverables
The first page of a retainer proposal should be able to finish this sentence: based on what I saw, the constraint on your growth right now is X, and here is the sequence I would work in because of it. If it cannot, the deliverables underneath it are guesses with a price attached.
That sentence is hard to write without having looked at your analytics, your search data, your CRM records and your site. Which is the honest argument for doing a short, paid, fixed-scope diagnosis first rather than starting a twelve-month engagement on a hunch. A founder who has bought a diagnosis knows what they are buying next.
A diagnosis is not a retainer, and should not be priced like one
These are two different products and blurring them is the most common way a first engagement goes wrong. A diagnosis is finite: read access to your stack, a written assessment, a sequenced plan, one call to argue about it. A retainer is ongoing execution against that plan. The first should be cheap enough to say yes to and specific enough to be useful even if you then hire someone else. The second should be month to month.
If a proposal folds a three-week assessment into month one of a twelve-month commitment, you are paying for the assessment either way and losing the option to walk after it.
Six things to get in writing before you sign
Who does the work. Not who is on the call. Ask for the name of the person writing the articles and the person in the ad account, and ask whether that changes after month two. At a startup this matters more than at a larger company, because your product is hard to write about and handover cost is real.
What the report contains and when it arrives. A monthly report should say what was done, what it cost, what changed and what happens next. If the sample report is a screenshot of a traffic graph, you will spend a year unable to tell whether the retainer is working.
Account access and who owns it. Every analytics property, search console, ad account, CRM and content management system should be owned by your company, with the agency granted access rather than holding the account. Ask what happens to that access on the last day.
What happens to lifecycle email. If the retainer touches onboarding email, product email, nurture sequences or CRM automation, get it written down who is responsible for consent records, sender identification and unsubscribe handling. More on that below.
What is explicitly out of scope. A good proposal has an exclusions list. Website rebuilds, design systems, sales collateral, event work and paid media spend itself are all things that commonly sit outside a retainer and commonly get assumed into one.
Contract length and notice. Ask for the notice period in days and whether there is any minimum term. An annual contract is not a red flag on its own, but you should know you are signing one.
A content quota is not a strategy
Two articles a month is a capacity statement. It describes how much of someone's week you have bought. It says nothing about what those articles are for, who they are aimed at, or what happens if the first four do not work.
The question that separates the two is simple: what would make you change the plan? An agency running a quota has no answer, because the quota is the plan. An agency running a strategy will tell you which numbers they are watching and what they would do differently at month three if those numbers have not moved. Ask it in the first call and listen for whether the answer is specific.
Related to this: be wary of a proposal that promises a ranking, a signup volume or a pipeline number. Nobody can promise those, and the proposals that do are competing on the one dimension where lying is free.
The CASL part, because it always comes up and rarely appears in the proposal
Under Canada's Anti-Spam Legislation, a commercial electronic message needs three things: consent, identification information, and an unsubscribe mechanism (sourced from the CRTC's CASL frequently asked questions). Express consent has to be obtained through an opt-in, not an opt-out, and the recipient has to take a positive action. Implied consent is narrower than most founders assume: a conspicuously published business address with no statement declining messages can support it, and only for messages relevant to that person's business role. The unsubscribe mechanism has to stay valid for at least 60 days after the message is sent, and a request has to be processed without delay and no later than 10 business days.
None of that is optional because an agency is doing the sending on your behalf. Your company is the sender. So the retainer should say who keeps the consent records, who owns the unsubscribe list, and what happens to both when the engagement ends. This is general guidance and not legal advice, and anything consequential is worth a conversation with counsel.
If none of that machinery exists yet, the scope question changes shape: you are buying the first version of it rather than maintenance of it. Where lifecycle email and CRM actually start sets out what that first version reasonably contains.
What mine looks like, since I am asking you to judge someone else's
It would be strange to write all that and keep my own scope vague. There are two retainers, both month to month with no annual contract ever, and a one-time diagnostic that credits in full against month one if you start a retainer within 30 days of delivery. The lower retainer covers positioning and messaging maintained as the product changes, two researched articles a month, technical SEO, a landing page a quarter, a lifecycle email flow a quarter, a written monthly report and a recorded call at regular intervals. The higher one adds fractional product marketing, managed paid acquisition (aka ads) and one end-to-end launch a quarter supporting a new feature or a new product. If you want to compare a proposal against something concrete, what each retainer actually includes is published on the page rather than quoted on a call, which is the only reason I get to make the argument above with a straight face.
So: read the proposal on your desk and find the sentence that names your constraint. If it is not there, that is the thing to ask for before you ask for a discount.