Every retail founder eventually stares at the same invoice and asks the same question. The marketing agency costs more than any single person on the payroll, the monthly report is a wall of green arrows, and revenue has been flat for two quarters. Is the agency the problem, or is the brand?
This is an account of one direct-to-consumer brand that answered that question by ending a five figure monthly retainer and hiring one in-house marketer instead. The company sells a replenishable home goods product, was doing roughly $6M in annual revenue at the time of the switch, and asked not to be identified because vendor relationships are still active. Figures below are rounded and some details are composited with patterns we have seen at similar brands, but the sequence of events and the cost structure are real.
The short version is that the switch worked, it worked more slowly than the founder expected, and the parts that broke were not the parts anyone had worried about. The debate over in house vs agency retail marketing is usually framed as a cost argument. In practice it is an argument about who owns judgment, and cost is downstream of that.
In short
- The retainer did not fail suddenly. It decayed over about nine months while reporting stayed positive, because the agency was measuring its own activity rather than the brand’s incremental revenue.
- The agency was genuinely better at three things: creative volume, platform-specific technical execution, and absorbing a bad month without anyone quitting.
- One generalist hire replaced roughly 60% of the scope, not 100%. The rest moved to freelancers or simply stopped happening, and some of what stopped happening did not matter.
- The break-even is around $3M to $8M in revenue for most consumer brands, though the real trigger is whether the founder can write a brief, not the revenue number.
- The first 90 days got worse before better. Paid performance dipped, creative output collapsed, and the recovery came from rebuilding a freelance bench rather than from the hire alone.
The retainer that stopped producing and why it took so long to see
The agency relationship started well. The brand signed in early 2024 at $9,500 a month for paid social, paid search, and what the scope of work called “growth strategy.” For the first seven months, blended return on ad spend held between 3.1 and 3.4, revenue grew about 40% year over year, and the founder considered the retainer the best money the company spent.
The decay began quietly. Return on ad spend slipped to 2.8, then 2.6, then hovered near 2.4 for a full quarter. Each monthly report explained the movement: rising auction costs, a seasonal softness, a competitor entering the category with aggressive pricing. Every explanation was individually plausible. That is exactly what made the pattern hard to see.
What the founder eventually noticed was structural. The reports measured campaign metrics, not business outcomes. Click-through rate was up. Cost per thousand impressions was benchmarked favourably against a category average the agency itself supplied. Creative testing velocity was described as “strong.” None of those numbers were fabricated, and none of them answered whether the ad spend was producing revenue the brand would not have earned anyway.
The incrementality gap nobody was measuring
The specific failure was attribution. Roughly 70% of the reported conversions were coming from branded search and from retargeting audiences built out of the brand’s own email list. Those are the cheapest, highest-converting placements in any account, and they are also the ones most likely to capture demand that already existed. The agency was, in effect, being paid a percentage of spend to buy customers who were already on their way to checkout.
This is not necessarily malpractice. It is what incentive structures produce. An agency compensated on retainer plus a percentage of media spend has a rational interest in spend growing and in reported return staying defensible. Nobody in that arrangement is motivated to propose spending less. The founder had never asked for a holdout test or a geo-based incrementality read, and the agency had never volunteered one.
The lesson generalises past this one company. Brands that grow through a channel rarely notice when the channel stops carrying them, because the reporting layer sits inside the channel. Founders who have rebuilt after a channel collapse describe the same delay in recognition, a pattern covered in more depth in our account of how a retail founder rebuilds after a category killer kills the channel. The warning signs are visible in aggregate business data long before they show up in a platform dashboard.
The conversation that ended it
The founder asked for a two week geographic holdout: pause non-branded paid social in four states, hold everything else constant, and compare revenue per capita. The agency pushed back, arguing the sample was too small and the test would damage account learning. That objection has real technical merit and it was also, in context, the wrong answer. The founder ran the test anyway using the agency’s own campaign structure.
Revenue in the held-out states fell about 6% against a national decline of 4% over the same window. The difference was inside the noise band for a test that size, which is to say the test did not prove the spend was worthless. It proved that a meaningful chunk of it was not clearly producing anything, and that after two years nobody could say which chunk. Thirty days later the retainer ended.
What the agency was genuinely better at
The honest post-mortem matters more than the grievance. Founders who leave an agency angry tend to under-hire, because they have convinced themselves the work was easy. Three things the agency did well became visible only once it was gone.
Creative volume. The agency shipped between 30 and 45 new ad creatives a month across static, video, and user-generated formats. That was not brilliance, it was throughput: a junior designer, an editor, and a template library. A single in-house marketer produced eight in her first month. Paid social performance is substantially a function of how many creative concepts enter the auction, and volume is the least glamorous input that actually moves the number.
Platform-specific technical depth. Server-side conversion tracking, catalogue feed hygiene, and the specific ways attribution settings interact after privacy changes are genuinely specialised knowledge. The agency had someone who did nothing else across a dozen accounts. Rebuilding that competence in one generalist takes months, and some of it never fully transfers. Retailers working through the measurement side of this will recognise the terrain from our walkthrough of how retail marketing campaigns are built from brief to launch.
Absorbing variance. When a month went badly at the agency, four people covered for each other and the account kept moving. When the in-house marketer took a two week holiday in month five, marketing output effectively stopped. Concentration risk is the least discussed cost of the in-house model, and it is the one that bites hardest at small headcounts.
What the agency was not better at
Against that, three things improved immediately and permanently once the work moved inside. Response time to a product or inventory change dropped from four days to about four hours. Understanding of why customers actually bought the product went up sharply, because the marketer sat close enough to customer service to hear the calls. And the brand stopped paying for a strategic layer it was not consuming, since the founder had been overriding agency strategy recommendations for most of the previous year anyway.
That last point is worth sitting with. A great deal of agency retainer value is theoretical value the client never uses. If a founder has strong opinions and exercises them, they are paying senior agency rates for execution while supplying the judgment themselves. That specific mismatch, more than cost, is the clearest signal that the model has stopped fitting. The broader question of when to buy expertise and when to build it sits at the centre of the retail business landscape covering funding, founders and exits.
Scoping the first in-house marketing hire realistically
The first job description the founder wrote was a fantasy. It asked for a marketer who could run paid acquisition, own email and SMS lifecycle, produce creative, manage influencer partnerships, write the content calendar, handle analytics, and contribute to brand strategy. That is four jobs at three seniority levels, and the candidates who claim to do all of it competently are, with rare exceptions, claiming something untrue.
The rewrite came from a simple exercise. The founder listed every marketing activity from the previous quarter, estimated hours, and sorted by revenue contribution. Roughly 60% of the total hours sat in three activities: paid social management, email and SMS lifecycle, and creative briefing. Everything else was long tail. The role was rescoped to own those three, with explicit language that the hire would brief and manage external specialists for the rest.
Generalist or specialist at the first hire
The debate is real and the answer depends on what the founder can personally cover. A specialist paid media hire is the right call when the founder is weak on acquisition and strong on brand and creative. A generalist is right when the founder is a competent operator who needs execution capacity and coordination more than deep channel skill.
This brand hired a generalist with a paid social bias, roughly five years of experience, at $95,000 base. That level matters. Hiring at $70,000 produces someone who needs direction the founder does not have time to give. Hiring at $140,000 produces someone who expects a team and a budget the company cannot supply, and who will be gone in a year. The band that works for a first marketing hire at this revenue level is generally $85,000 to $110,000 depending on market.
The true cost of the hire
Base salary understates the commitment by a wide margin. Employer payroll taxes, health benefits, equipment, software seats, and recruiting fees all sit on top. According to the US Bureau of Labor Statistics Employer Costs for Employee Compensation series, benefits have historically accounted for roughly 30% of total compensation for private industry workers, though the exact share varies by industry, region and reporting period and should be verified against the current release. Software the agency previously supplied out of its own stack added about $900 a month at this brand: creative tooling, a reporting layer, and an SMS platform seat.
Founders consistently underestimate that software line. Agencies bundle enterprise tool access into the retainer, and losing it is invisible until the first month without it. Build the tooling cost into the comparison before signing an offer letter, not after.
The first 90 days: what broke immediately
The transition plan assumed a two week overlap, a documented handover, and continuity of campaign structure. What happened was closer to a controlled fire.
Week one, the agency revoked access to the shared creative asset library. This was contractually permitted, since the assets had been produced under a work-for-hire clause that assigned ownership in a way nobody had read closely since signing. The brand lost roughly 200 edited video clips and had to re-shoot. Check the intellectual property clause in any agency contract before giving notice, not during.
Week three, paid social performance dropped about 25%. Some of this was the creative volume collapse described earlier. Some was a genuine account learning reset, because the new marketer restructured campaigns rather than inheriting a structure she did not understand. Both causes were predictable and neither had been budgeted for.
Week six was the low point. Return on ad spend sat at 1.9, the founder was three weeks into second-guessing the decision, and the marketer was working sixty hour weeks trying to cover a scope built for four people. This is the phase where most founders panic and re-hire an agency at worse terms.
What actually stabilised it
Recovery came from three moves, none of which were in the original plan.
- A freelance creative bench. Two video editors and a static designer on retainers of $1,200 to $2,000 a month each, briefed by the in-house marketer. Creative volume returned to roughly 25 assets a month by week ten, short of agency throughput but sufficient.
- A ruthless scope cut. Influencer partnerships, the blog, and organic social were paused entirely for a quarter. Revenue impact was undetectable. Two of the three never came back.
- A weekly founder review with a fixed agenda. Thirty minutes, three numbers, one decision. The marketer had been drowning partly because she had no forum to escalate trade-offs and was trying to resolve them alone.
By month four, blended return on ad spend was back to 2.7, above where the agency had left it, on about 20% less media spend. By month six it was 2.9. The gain came less from better campaign management than from cutting spend that the holdout test had already suggested was not doing much.
Freelancers and contractors as the missing middle
The framing of agency versus in-house is a false binary, and it is the reason so many founders get the decision wrong in both directions. The actual structure that works at most consumer brands under $20M is a small salaried core surrounded by a managed contractor bench.
Freelancers solve the specific weaknesses of the in-house model. They supply variable capacity, so a product launch month can absorb three times the normal creative load without a permanent hire. They supply specialist depth in narrow areas, such as feed management or lifecycle deliverability, where a generalist will never be strong. And they can be ended in thirty days, which makes experimentation cheap.
They also introduce a real cost that founders discount: management overhead. A bench of four contractors consumes roughly a day a week of the in-house marketer’s time in briefing, review, and payment administration. That is 20% of the salaried hire’s capacity, and it needs to be in the scope conversation from the start rather than discovered in month three.
How to structure a contractor bench that holds
Three practices separated the arrangements that worked from the ones that quietly fell apart at this brand and at others we have looked at.
Pay a small monthly retainer rather than purely per project. A $1,200 monthly commitment for a guaranteed volume window buys priority in the freelancer’s queue. Pure project work means the brand is always at the back of the line behind whoever pays a retainer, and turnaround times drift from three days to two weeks without any conversation happening.
Write briefs the freelancer can execute without a call. The single biggest predictor of contractor output quality is brief quality, not freelancer talent. A brief that specifies the audience, the objection being answered, the format, the reference examples, and the deadline produces usable work from a mid-tier freelancer. A vague brief produces unusable work from an excellent one.
Keep the account credentials and the asset library in-house from day one. Every access credential and every source file lives in a brand-owned drive and a brand-owned password manager. This is the single cheapest insurance policy in the entire marketing function, and it is the lesson the asset library incident taught expensively.
Where contractors are a bad answer
Contractors are poor at anything requiring accumulated institutional context: positioning decisions, pricing, customer research synthesis, and anything touching the product roadmap. They are also poor at work that requires being present when something breaks at 9pm on a Friday during a launch. Those responsibilities belong to salaried staff or to the founder, and trying to contract them out produces expensive disappointment.
Cost comparison at three revenue levels
The following comparison uses total annual marketing operating cost, excluding media spend, since media is roughly constant across models. Figures are drawn from this brand’s actuals plus market rate ranges for US consumer brands as of 2026, and they will vary meaningfully by city, category and negotiating position.
| Annual revenue | Agency retainer model | In-house plus freelance bench | Practical verdict |
|---|---|---|---|
| Under $2M | $3,000 to $6,000 a month, junior team, limited attention | $0 salaried, $2,000 to $4,000 a month freelance, founder-led | Founder plus freelancers. A salaried hire is premature and a retainer buys a junior you will manage anyway. |
| $3M to $8M | $8,000 to $15,000 a month, mid-tier team, percentage of spend | $95k salary plus roughly $30k loaded cost plus $4,000 a month freelance, about $173k a year | The genuine decision zone. Costs are within 15% of each other, so the choice turns on control and speed, not price. |
| $10M and above | $20,000 to $40,000 a month, senior team, specialist pods | Two to three salaried roles plus bench, roughly $350k to $450k a year | In-house core with specialist agencies retained for defined projects. Full outsourcing at this scale usually signals a leadership gap. |
The most important reading of that table is the middle row. At $3M to $8M the two models cost roughly the same, which means anyone arguing the decision on price alone has not done the arithmetic. The real variables are response speed, who owns judgment, and whether the founder is capable of directing a hire.
The cost the table does not capture
Three costs sit outside the spreadsheet and change the answer more than any line in it.
Founder attention is the first. An in-house hire consumes three to five hours a week of founder time in the first six months, in briefing, review and unblocking. Founders who are already the bottleneck on product or operations should price that honestly, because the hire will underperform in proportion to the attention they do not get. The same dynamic shows up across every early role, which is why hiring sequence matters as much as hiring quality, a theme explored in our piece on hiring the first ten roles at a scaling D2C brand.
Ramp time is the second. A new marketer costs full salary and delivers partial output for roughly 90 days. Budget one full quarter of reduced performance into the switch or the decision will look like a failure exactly when it is proceeding normally.
Concentration risk is the third. One marketer means one point of failure for holiday, illness and resignation. The mitigation is documentation and a contractor bench that already knows the account, not a second hire the company cannot afford.
When to go back to an agency for specific work
Going in-house is not a permanent ideological position. This brand re-engaged external partners twice within eighteen months, both times on narrow project scopes rather than open-ended retainers, and both engagements were straightforwardly good decisions.
The distinction that matters is between buying capacity and buying capability. Buying capacity to do work you already understand is usually better handled by contractors, because you can brief precisely and you keep the learning. Buying capability you genuinely lack, and which you do not need permanently, is exactly what an agency or a specialist consultancy is for.
| Work type | Best owner | Typical engagement | Why |
|---|---|---|---|
| Daily paid social management | In-house | Salaried | Needs product and inventory context that changes weekly |
| Email and SMS lifecycle | In-house | Salaried | Owns the customer relationship and the highest margin revenue |
| Ongoing creative production | Freelance bench | Monthly retainer | Variable volume, briefable, no institutional context needed |
| Brand identity or rebrand | Agency or studio | Fixed project, 8–14 weeks | Deep capability used once every few years |
| Measurement and incrementality build | Specialist consultancy | Fixed project, 4–8 weeks | Narrow technical expertise, leaves a durable system behind |
| New channel launch, for example retail media | Agency | 6 month engagement with handover clause | Buys a learning curve you would otherwise pay for in wasted spend |
| Positioning and pricing | Founder | Never outsourced | Requires accountability nobody external can hold |
Contract terms worth insisting on
Whatever the engagement, four clauses determine how painful the eventual exit will be. Insist that all creative assets and source files are assigned to the brand on payment rather than held under work-for-hire terms that keep ownership with the agency. Insist that all advertising and analytics accounts are owned by the brand entity with the agency added as a user. Insist on a defined handover deliverable at termination, including campaign documentation and asset transfer within fourteen days. And keep notice periods to thirty days rather than the ninety many agreements default to.
None of that is adversarial. Good partners agree to all four without argument, because they intend to be re-hired. The reaction to those clauses during negotiation is itself informative.
The signal that it is time to switch
Across the brands we have looked at, the reliable trigger is not a revenue threshold. It is the moment the founder starts routinely overriding the agency’s recommendations. At that point the brand is paying senior rates for execution while supplying the judgment internally, and the arrangement has quietly inverted. That is also true in reverse: a founder who cannot write a coherent brief or evaluate a marketing plan should not hire in-house yet, because they will hire badly and blame the hire. Founders who have built without outside capital tend to hit this decision earlier and with less margin for error, a dynamic we cover in how a retail founder bootstraps to seven figures without VC.
A narrow category position makes the in-house case stronger, because specialist product knowledge compounds inside a salaried employee and evaporates at the end of a retainer. That is one of the underrated arguments for picking a niche even when it feels narrow. For a broader map of how marketing sits alongside funding, hiring and eventual exit decisions, the wider retail business landscape guide puts this choice in sequence with the others a founder faces.
Where this brand landed
Eighteen months after the switch the marketing function is one salaried marketer, a bench of three freelancers, a specialist measurement consultancy engaged twice a year, and a founder who spends about two hours a week on marketing rather than the six he spent managing the agency. Revenue is up 34% on media spend that is 18% lower than the peak. The company also went through one resignation scare in month eleven that would have been genuinely dangerous without documentation, which is the honest counterweight to the numbers.
The switch was correct for this brand at this size. It would have been wrong two years earlier and it will probably need revisiting at $15M. That is the actual answer to the in-house versus agency question: it is a phase decision, not a philosophy, and it should be re-examined roughly every eighteen months.
FAQ on agency versus in-house
At what revenue should a retail brand hire its first in-house marketer?
Most consumer brands reach the decision zone between $3M and $8M in annual revenue, where agency retainers and a loaded salaried hire cost roughly the same. Revenue is a weak signal on its own though. The stronger test is whether the founder can write a clear marketing brief and evaluate the output, because a first hire without competent direction usually underperforms regardless of the revenue behind them.
Is in-house marketing actually cheaper than an agency?
Rarely, once fully loaded. A $95,000 salary typically carries an additional 25% to 35% in payroll taxes, benefits and equipment, plus software the agency previously bundled and a freelance bench to replace creative volume. Total cost often lands within 15% of a mid-tier retainer. The gains are speed, context and ownership rather than savings.
What breaks first when a brand leaves its agency?
Creative volume, almost every time. An agency team ships 30 to 45 assets a month through a production pipeline, while a single hire ships under ten. Paid social performance is closely tied to the number of creative concepts entering the auction, so the drop shows up in the ad account within two to four weeks. Building a freelance creative bench before termination is the mitigation.
How long should we expect performance to dip after the switch?
Plan for a full quarter. Ramp time for a new marketer, campaign restructuring, and the creative volume gap typically combine into 60–90 days of below-baseline performance. Brands that budget for the dip tend to hold their nerve through week six, which is where most reversals happen.
Should the first hire be a generalist or a paid media specialist?
It depends on the founder’s own gaps. Hire a paid media specialist when the founder is strong on brand and creative but weak on acquisition mechanics. Hire a generalist when the founder is a capable operator who mainly needs execution capacity and someone to coordinate external specialists. Hiring a generalist to fix a genuine acquisition problem is the more common mistake.
What contract terms should we secure before giving an agency notice?
Confirm that creative assets and source files transfer to the brand, that all ad and analytics accounts are owned by the brand entity, that a documented handover is a defined deliverable, and that the notice period is thirty days rather than ninety. Review the intellectual property clause specifically, since work-for-hire language can leave asset ownership with the agency. This is general information rather than legal advice, and an attorney should review any contract before termination.
Can freelancers fully replace an agency?
For execution work, largely yes. For strategy, judgment and anything requiring accumulated institutional context, no. A contractor bench also consumes roughly a day a week of the in-house marketer’s time in briefing and review, so the management overhead has to be scoped into the salaried role rather than assumed away.
How do we know whether our current agency is actually producing revenue?
Run a geographic holdout test. Pause non-branded paid activity in a set of comparable regions for two to four weeks, hold everything else constant, and compare revenue per capita against control regions. Also check what share of reported conversions comes from branded search and from retargeting built on your own email list, since those placements frequently capture demand that already existed.
When does it make sense to hire an agency again after going in-house?
When you need capability you genuinely lack and will not need permanently: a rebrand, a measurement infrastructure build, or a new channel launch such as retail media. Scope those as fixed projects with a handover clause rather than open-ended retainers, so the learning stays with the brand once the engagement ends.
Cost figures, salary bands and benefit ratios in this article reflect US market conditions as of 2026 and are illustrative rather than prescriptive. Employer cost ratios should be verified against the current Bureau of Labor Statistics release, and contract terms should be reviewed by qualified counsel before acting on them.