The First 90 Days With a Social Media Agency: What Good Looks Like Week by Week

The engagement starts before the kickoff call. Four things should be written down and signed off before any work begins, and their absence is the strongest early predictor of a bad quarter.

The objective, stated commercially. Not "grow the brand" but something measurable: booked consultations, qualified enquiries, demo requests, applications. If the agency accepted a vague objective without pushing back, they have accepted that success will be argued about later rather than measured.

The deliverable schedule, with counts and formats. How many short-form pieces, how many long-form, how many static assets, on which platforms, by when each month. Ambiguity here becomes the thing you argue about in month two.

Ownership of the eight jobs: strategy, scripting, filming direction, editing, packaging, publishing, community management, and reporting. Every one of them should have a named owner — agency or client. The jobs nobody claims are the jobs that do not happen.

The review and approval loop. Who approves, in what timeframe, through which tool. Client-side approval latency is the single most common cause of missed schedules, and it should be contractually acknowledged: if approval takes six days, the calendar slips six days and that is not the agency's failure.

The first fortnight should feel like consulting, not production. If videos are being published in week one, that is a warning sign, not a sign of speed — it means they are producing before they understand your buyer.

What a good agency does in weeks one and two: a deep positioning session covering who you sell to, what they believe before they buy, the objections that kill deals, your pricing and your differentiation. A competitive teardown of five to ten accounts in your space with an explicit account of what you will do differently. An audit of your existing content and analytics for whatever signal exists. A technical setup pass: analytics access, scheduling tools, asset library, brand kit, shared drive structure, approval workflow.

The deliverable at the end of week two should be a written strategy document. Not a slide deck of platitudes — a document containing your positioning statement, the three to five content pillars, the format library you will test, the posting cadence, the measurement framework with baselines, and the ninety-day milestones.

What the client must provide in this window: two hours of interview time, access to analytics and accounts, brand assets, examples of past client work you are proud of, and honest answers about pricing and what has failed before. Clients who withhold the failures cost themselves a month.

Now production starts, and the first month's content should be explicitly framed as a test rather than a campaign. A competent agency will produce across three to five distinct formats, deliberately varied, so that by day sixty there is legible evidence about what your market responds to.

The first filming session happens here. It should come with a prepared shot list — twenty to forty angles, each with a hook written in advance — and direction on the day. If you were handed a camera and told to "just talk about your expertise," the agency has offloaded the hardest job back to you.

The first drafts arrive around week three. Expect to give heavy feedback on this batch; that is normal and healthy. What matters is whether the second batch reflects the first batch's notes. An agency that repeats the same mistake after specific written feedback is not going to improve at month five either.

By the end of week four you should have: a published cadence running, ten to twenty pieces live or scheduled, a brand style locked (captions, fonts, colours, pacing, first-frame treatment), and a reporting dashboard you can access yourself rather than only seeing in a monthly PDF.

What you should not expect: leads. Four weeks of content on a cold audience produces data, not pipeline. Anyone promising otherwise is either buying attention with paid spend or setting up a disappointment.

Month two is where an agency demonstrates whether it is running a system or a content factory. The defining activity is not production; it is the decision to kill things.

At the thirty-day review you should receive an honest read of the format test: which formats produced retention, which produced profile visits, which produced nothing, and what the plan is as a result. Specific numbers, specific decisions. "Engagement is trending up" is not a read.

Production in month two should visibly re-weight toward what worked. If the calendar for month two looks like month one's with different topics, no learning has occurred. Expect roughly 70% of output in the two strongest formats, 20% in new experiments, 10% in explicit offer content — the videos that state what you sell, for whom and at what cost.

This is also when the conversion layer must be fixed if it has not been. Profile bio, pinned content, landing page, booking flow, DM response scripts and response-time standard. Agencies that only produce content and never touch conversion routinely deliver views and no pipeline, then blame the market.

Expected metrics by day sixty for a business starting cold: median views climbing meaningfully over month one, profile visits measurably up, the first inbound DMs and comments containing buying language, and possibly the first one or two booked calls. Not a full pipeline — a working signal.

Month three is where a good engagement starts to look boring in the right way. The system runs, the format is known, the filming rhythm is established, and improvement comes from iteration rather than reinvention.

Expect three things in this window. First, output at full contracted volume with the quality floor consistently met — no more "we are still getting up to speed." Second, at least one genuine breakout piece; with sixty to a hundred pieces published across the quarter, the distribution of outcomes should have produced an outlier, and if it has not, the hooks are too safe. Third, a documented system: brand kit, hook library, format templates, shot-list templates, posting SOP. This is the asset you are actually buying, and it should exist in a shared location you own, not in the agency's private notion.

Commercially, day ninety is the first fair point to assess pipeline. For a service business with a functioning conversion layer, a realistic ninety-day outcome is a handful of qualified enquiries and a small number of booked calls, with the trajectory more informative than the total. Deal cycles matter here — if your sales cycle is sixty days, revenue from month-one content appears in month four at the earliest, and judging the engagement on closed revenue at day ninety is a category error.

The ninety-day review should cover: performance against the baseline set in week two, what was learned, what is being changed, what the next quarter's plan and volume are, and an honest assessment of what is not working. An agency that presents a ninety-day review with no failures in it is managing your perception rather than the programme.

Some failures are only visible early, and catching them at week three costs far less than discovering them at month six.

Week two warning signs: no written strategy document; discovery consisted of a single one-hour call; no baseline metrics recorded; you cannot access your own analytics; the strategy could apply to any business in any industry.

Month one warning signs: content published before strategy; no filming direction provided; all pieces in one format; the same feedback given twice without correction; no access to a live dashboard; the account manager is the only person you have ever spoken to and cannot answer specific creative questions.

Month two warning signs: the calendar did not change after the format test; reporting shows impressions and follower counts but no funnel metrics; nobody has touched your profile, bio, landing page or DM process; the agency cannot tell you which pieces produced profile visits.

Month three warning signs: no documented system exists; every deliverable is late but the invoice is not; the ninety-day review contains no failures; when you ask what changes next quarter, the answer is "more of the same"; the team that pitched you is not the team doing the work.

One structural warning that applies throughout: if you cannot get a straight answer about what is not working, you are being managed rather than served. Agencies with functioning programmes are relaxed about naming failures, because the failures are the evidence that they are testing.

Roughly half the failed engagements we see involved an agency doing acceptable work for a client who could not feed it. The client side has four non-negotiable jobs.

Access to the expert. Content that sells professional services requires the person with the expertise on camera or in an interview chair, and no agency can manufacture that. Two hours monthly is the realistic minimum: one filming session plus one review. Founders who cannot commit two hours should not buy a content programme.

Fast approvals. A 48-hour approval turnaround keeps the machine running; a week-long one halves your effective output. Nominate a single decision-maker with authority. Committee approval is where content goes to die, and it is the reason most legal, medical and enterprise programmes underperform.

Raw material. Client results, objections you hear on calls, questions you answer repeatedly, deals you lost and why, the actual numbers you can share. The agency cannot invent these and they are what makes content specific enough to convert.

Patience calibrated correctly. Not infinite patience — the milestones above are real and you should hold the agency to them. But organic content compounds on a six-to-twelve month curve, and the businesses that win are the ones that judged the system at ninety days and then let it run.