The two-week rebuild that beats a six-figure agency retainer.

A regional franchise group I did some work with a while back was eighteen months into a marketing agency retainer. Good people on both sides. Every month there was a report deck, forty-some slides, color-coded, with a slide near the front called "momentum." The retainer ran a little north of eight thousand a month. Do the arithmetic and that is six figures a year, every year, for a relationship that had quietly turned into a subscription nobody remembered signing up for.

I asked the marketing lead one question. What did the agency ship last month that your team actually uses? Long pause. She pulled the deck back up, scrolling, like the answer might be sitting on slide 31. It wasn't. Slide 31 was a chart of impressions.

That is not a knock on the agency. They did exactly what a retainer pays them to do. The trouble is what a retainer pays them to do.

The retainer is a great deal. For the agency.

Marketing retainers for small and mid-sized businesses commonly run between $2,500 and $6,000 a month, and businesses in the $1M to $5M range routinely pay $5,000 to $10,000 a month for multi-channel work, according to current agency pricing surveys. None of that is a scam. It is a pricing model, and the pricing model has a shape.

A retainer sells hours and activity. It does not sell shipped assets. That distinction sounds academic until you have paid for eighteen of those months. The agency's incentive, structurally, is for the relationship to continue. Not to finish anything. A finished thing is the end of the engagement. Momentum is the product, and momentum is a slide, not a sale.

So you get a cadence. Posts go out. Ads run. Reports arrive. The dashboard is busy. And if you strip out the activity and ask what got built that outlives the invoice, the honest answer at a lot of shops is: the reports.

The redesign that ships in month nine

The retainer's cousin is the big redesign project. Same story, different wrapper. You sign a five or six figure statement of work for a full site rebuild, and then you wait.

A typical small-business website redesign runs about 8 to 12 weeks when everything goes right, and projects routinely stretch 30 to 50 percent past that. The delays are usually not the build. They are slow feedback rounds, missing content, and a stakeholder who surfaces in week seven with opinions. By the time the thing goes live, the promotion that justified the rebuild is over, the seasonal window closed, and the person who commissioned it has half-moved on.

Nine months and a hundred grand later you have a prettier version of the site you already had. Prettier is not nothing. But prettier is rarely the problem. The problem is usually one page, or one flow, that leaks money, and you did not need to rebuild the other forty pages to fix it.

What a two-week rebuild actually is

Here is the alternative we keep coming back to. Instead of retaining an agency to generate activity, or commissioning a nine-month redesign of everything, you scope one load-bearing thing and you build it in two weeks. Done. Shipped. Yours.

Two weeks is not a gimmick. It is a constraint, and the constraint does the work. When you have two weeks, you cannot rebuild the whole site. You are forced to answer the only question that matters: what is the single page or flow that moves money, and what is wrong with it? That is the entire scope. The booking flow. The quote widget. The one landing page that eats your ad spend. The checkout that loses people at step three.

The upside on getting that one thing right is not small. Baymard's checkout research, one of the more rigorous bodies of work on the subject, puts the average conversion lift from fixing a broken flow at roughly 35 percent. On the traffic and spend a lot of operators are already running, a lift like that is worth more than a year of the retainer, and it lands in two weeks instead of month nine.

The math, laid flat

Put the two models side by side and the comparison gets uncomfortable for the retainer.

The point is not that the build is cheaper, though it usually is. The point is what you own at the end. After a year of retainer you have a stack of reports and a relationship you are afraid to cancel. After a two-week build you have a thing. It runs whether or not you keep paying anyone. That is the whole difference between renting and owning.

Why two weeks and not two months

People assume faster means sloppier. In practice the opposite tends to be true, because the deadline kills the scope creep that wrecks the nine-month project. Two months of runway invites a second opinion, then a third, then a "while we're in there" that doubles the surface area. Two weeks does not have room for any of that. You define the one flow, you build it, you ship it, you measure it.

It also changes who you talk to. A two-week build has one decision-maker and one builder in the same short conversation. No account manager translating your ask into a ticket into a sprint into a status update. The person who understands the business talks to the person writing the code. That is most of where the speed comes from, and most of where the quality does too.

Short timelines are not the right call for genuinely large systems. A full ERP migration is not a two-week job and we would not pretend otherwise. But the thing most operators actually need fixed this quarter is small, specific, and load-bearing. It has been sitting on the roadmap for a year precisely because it never justified a big project. Two weeks is exactly the right size for it.

When the retainer is actually right

To be fair to the model, retainers earn their keep in a few cases. If you genuinely need ongoing paid-media management, someone watching campaigns and adjusting bids every week, that is real recurring labor and a retainer is a reasonable way to buy it. Same with ongoing content production at volume, or a support function that has to exist every day. Recurring work is a fair thing to pay for on a recurring basis.

The failure is using a retainer to buy things that should be finished. A website. A tool. An integration. A calculator. Those are assets. You should build them, own them, and stop paying. Renting an asset by the month is how you end up eighteen months in, staring at slide 31, unable to name the last thing that shipped.

The test before you renew

Next time a retainer comes up for renewal, or a big redesign lands on your desk, run one test. Ask what you will still have if you stop paying the day it is delivered. If the answer is a working asset that keeps earning, good, that is a build worth funding. If the answer is that the activity stops and you are back where you started, you were renting, and you should think hard about buying instead.

We like the two-week version because it is honest about what it is. It is not a relationship. It is a scope, a build, and a handoff. The data stays yours. The tool stays yours. When we walk out, the thing we built does not leave with us. That is the point.