The first 90 days marketing agency contract, week by week

Most engagements do not fail in month six. They fail in week three, quietly, and nobody notices until the quarterly business review has nothing real to show. The first 90 days with an outside growth partner is the only window where you have real bargaining power: money has not fully changed hands, the relationship has no scar tissue, and you can still walk. Treat those 90 days as a contract with four checkpoints, not a courtesy onboarding period, and you will know by day 60 whether this was the right hire.

Why most 90 day plans are written by the seller, not the buyer

Nearly every agency pitch hands the prospect a 90 day plan slide. It has three phases: discovery, activation, optimization. It looks organized. It is also written to protect the seller's timeline, not to prove anything to the buyer. Discovery can run four weeks and still be called on schedule. Activation can mean one campaign went live in one channel. Optimization can mean a dashboard got a new tab. None of that is a lie exactly, but none of it commits the partner to a number you can hold them to.

The fix is not a longer slide. It is naming the four dates that matter, day 14, day 30, day 60, and day 90, and writing down in the contract what must exist on each one, in the buyer's language, before the first invoice is paid. If a partner resists putting these dates in writing, that resistance is itself the answer to whether you should sign.

Day 14: baseline pipeline math, tracking audit, ICP cut

Two weeks in, a partner who knows what they are doing has three things on your desk, none of which is a brand refresh or a creative concept. The first is pipeline math: your actual close rate, your actual sales cycle length, your actual cost per meeting by channel, pulled from your CRM, not assumed from a benchmark deck. The second is a tracking audit, a plain list of every conversion event that is broken, missing, or double counting, because no test run on broken tracking produces a result you can trust. The third is a narrowed ICP, with the accounts and segments that got cut named explicitly, along with the reason.

If day 14 arrives and you have a brand positioning workshop instead of these three things, the engagement has already drifted. Strategy work that cannot cite your own numbers is guessing with better formatting.

Day 30: first live tests, reporting the client actually reads

By day 30, something should be live and spending money, even a small amount, in at least the top one or two channels the pipeline math pointed at. A partner who is still in planning at day 30 is not being careful, they are avoiding the first data point that could contradict their pitch. Live tests do not need to be big. A single campaign with real creative and a clean tracking setup beats three channels launched half configured.

The reporting question matters just as much and gets less attention. Ask yourself honestly whether you read last week's report from your current or previous partner. If the answer is no, the format is wrong regardless of how much data is in it. A report you read has three things: what changed since last week, what that change means for pipeline, and what the partner wants to do about it. Everything else is filler that exists to make the deck look thorough.

Day 60: channel verdicts, kill decisions

By day 60 you have roughly six to eight weeks of live data in at least one channel, which is enough to make a real verdict, not a final one. A verdict at day 60 looks like this: one channel is beating its cost per meeting target and gets more budget, one channel is missing it and gets one more testing cycle with a specific change, and one channel is dead and gets killed outright. If day 60 arrives and every channel is still described as promising or early stage, the partner is avoiding the part of the job that creates friction, which is telling you something is not working.

Day 90: a pipeline number, not a deliverable count

This is the checkpoint most engagements quietly fail without anyone naming it. The day 90 review should produce one number: how much pipeline, sourced or influenced with a defined attribution rule stated up front, came from the program in the quarter. Not the number of campaigns launched. Not the number of assets produced. Not a slide that says brand awareness has improved, which is a claim nobody can falsify and therefore nobody should accept.

If your partner cannot produce a pipeline number with a stated methodology by day 90, one of three things is true: the tracking from day 14 was never actually fixed, the sales team is not returning lead quality feedback, or there is no pipeline yet and the review is being dressed up to avoid saying so. All three are legitimate situations. None of them should be hidden behind a deliverable count.

The tells that a partner is stalling

• The strategy phase runs past three weeks with nothing live and the explanation keeps changing.
• Weekly reports describe activity, deliverables shipped, posts published, rather than pipeline or cost per meeting.
• No channel is ever named as underperforming, in month one or month three.
• The named senior person from the pitch has been replaced by someone you have not spoken to before the day 30 call.
• Every question about a number gets answered with a promise to follow up, and the follow up does not name a date.
• The day 60 review reads almost identically to the day 30 review with different dates on it.

What the client owes the partner

None of this works if the client is the reason the timeline slips, and a meaningful share of stalled engagements stall on the client side and get blamed on the partner anyway. Access has to be granted before day one, not requested on day one and delivered on day 12. That means analytics, CRM, and ad account logins ready to hand over, not pending an IT ticket. It means a named internal owner who can approve a creative concept or a budget shift inside 48 hours, not whenever the CMO next has a free hour. It means sales gives honest feedback on lead quality even when that feedback is unflattering to the marketing program, because a partner cannot fix what they are never told is broken.

Exit criteria, written before anyone needs them

The right time to agree on how this ends is before it starts, while both sides are still being reasonable. Three exit triggers are worth writing into the contract directly. No live test launched by day 35, five days past the day 30 checkpoint to allow for real world slippage, is grounds for exit without penalty. A day 60 review that produces no channel verdict, meaning every channel is still described as too early to judge, is grounds for exit without penalty. A day 90 pipeline number that cannot be traced to a stated attribution methodology is grounds for exit without penalty.

A partner who has actually done this work before will not blink at these terms, because they expect to clear all three anyway. A partner who negotiates these clauses down before signing is telling you, in the only language that matters in a contract, that they do not expect to hit them.

1. 1

   Before signing: write the four checkpoints into the contract

   Day 14, 30, 60, and 90 deliverables named specifically, plus the three exit triggers, go into the statement of work, not a separate onboarding deck that nobody references again.

2. 2

   Day 1: grant full access and name your internal owner

   Analytics, CRM, ad accounts, and a person who can approve budget and creative changes inside 48 hours. This clock starts the moment access is actually usable, not the moment it was promised.

3. 3

   Day 14: read the pipeline math, tracking audit, and ICP cut yourself

   Do not accept a summary from the partner about their own work. Open the CRM export and the tracking audit and check the numbers against what you already know to be true.

4. 4

   Day 30: sit through the report, not just receive it

   Take the 15 minute call. A partner who cannot defend their own report in a live conversation has not actually thought about what it means.

5. 5

   Day 60: force a written verdict per channel

   Scale, hold with a specific change, or kill. Ask for the reasoning behind each one in a sentence you could repeat to your CFO.

6. 6

   Day 90: demand the pipeline number and its source

   If it is not there, treat that as data, not an excuse to extend goodwill. Decide whether the cause is fixable or whether this is the exit point you wrote into the contract.

What I would do with a partner starting Monday

If I were signing a new growth partner this week, I would put the four checkpoints and three exit triggers into the contract in plain language before I signed anything, grant full access the same day, and calendar the day 14, 30, 60, and 90 reviews before the kickoff call ever happens. Ninety days is short enough that neither side can hide behind ambiguity for long, and long enough to know, with real numbers instead of a feeling, whether this is a relationship worth extending into a second quarter.