Paper trail.
How we actually do the work, written out. If you read one of these and decide to do it yourself, that is a fine outcome.
Custom AI tool, or off the shelf
Three questions that settle the decision, an honest two-year cost comparison, and the reason buying wins more often than a studio will tell you.
The short versionBuy unless the process is the thing you actually compete on, nothing on the market fits without changing how you work, and somebody in your business will own the tool for years. If any one of those three is missing, buy. The two-year cost of a custom build is roughly twice its build price.
What a custom AI app for a brand really costs
Why the same brief gets quoted at five thousand and at five hundred thousand, what actually moves the number, and the running costs nobody puts in the proposal.
The short versionCost is driven by how wrong the tool is allowed to be, how much of your existing mess it has to touch, and who maintains it afterwards. A narrow tool doing one job for one team is a few weeks of work. Anything with accounts, permissions and other people’s data is a different category of project entirely.
Your cost per thousand went up. Check these, in order.
A diagnostic order for a rising cost per thousand impressions, ending with the case where the right move is to do nothing.
The short versionCheck in this order: is the whole market up, have you run out of new people, is the creative tired, did something change in the account. Only the last two are yours to fix. And a rising cost per thousand impressions with rising sales is not a problem — it is you buying better people.
Why Meta and Shopify never agree
How to read two contradictory sales numbers without believing either one, and the monthly reconciliation that tells you which to trust.
The short versionTwo systems count different things. The ad platform counts people it showed an ad to who later bought. Your store counts the last link clicked before checkout. Neither is the truth. Bank deposits are the truth, so measure at that level and use the platform for direction, not for arithmetic.
Hiring an offshore agency, objections answered
The four real risks of hiring an agency in another country, how to test for each one before you sign, and when the answer is simply not to.
The short versionFour risks are real: overlap hours, turnover, cultural fluency in the copy, and accountability across borders. Each one is testable before you sign. Test them on us as hard as on anyone. If your work needs same-hour presence all day, hire locally.
Marketing contract red flags
Six clauses that decide what happens when the relationship ends, read one at a time, with the ones we refuse to use named.
The short versionSix clauses decide everything: notice period, ad account and page ownership, intellectual property assignment, auto-renewal, exclusivity, minimum term. Read those six and you have read the contract. The rest is administration.
Signs your agency is coasting
Five tests that tell you whether the agency on your retainer is still working on your account or merely still invoicing for it.
The short versionCount new creative concepts shipped in the last sixty days. Ask what was killed and why. Ask what the agency would do with half the budget. If reporting grows while creative output shrinks, you are paying for maintenance and being invoiced for growth.
Thirteen questions to ask before you sign
The thirteen questions that would hurt us most to answer badly, each one answered for Drifted, including the two answers that should end the call.
The short versionAsk who opens the file, who owns the ad account, what happens if the number is missed, and how fast you can leave. Vague answers on ownership and exit are the two that should end the call. Everything else is negotiable.
The minimum ad spend worth managing
Why agencies have spend floors, the platform arithmetic underneath them, and what to do with your money on either side of the line.
The short versionFloors exist because Meta needs about fifty conversions a week per ad set to optimise, and because a fee has to be a small share of media to pay for itself. Below roughly $5,000 a month, spend on creative and your product page instead.
Agency, freelancer, or in-house
The fully-loaded monthly cost of all three, including the management time nobody puts in the spreadsheet, and a rule for choosing.
The short versionA freelancer is cheapest and thinnest. In-house is most expensive and most controllable, and only pays back above serious spend. An agency sits between them and buys creative volume. Add your own management hours to all three before comparing.
Is an ads agency worth it under ten thousand a month
The arithmetic that tells you whether a management fee can pay for itself yet — and the three cases where the honest answer is no.
The short versionBelow roughly $5,000 a month in ad spend, a full agency fee usually costs more than it can recover. Work out the extra profit a fee has to produce to break even. If that number needs an unlikely improvement, hire a freelancer instead, or nobody yet.
What a Meta ads agency actually costs
The three ways agencies charge for Meta ads, what each model quietly hides, and the market bands to measure any quote against.
The short versionAgencies charge one of three ways: a flat monthly fee, a percentage of your ad spend, or both. The market sits around 10 to 20 per cent of spend, or roughly $1,500 to $10,000 a month flat. We charge one flat monthly fee, priced on the first call.
DM-first selling, explained
Some products do not get bought. They get discussed into being bought — and selling those through a checkout is measuring the wrong door.
The short versionIf the sale closes in a conversation, the conversation is the funnel. Point the creative at a message rather than a click, optimise toward the inbox, and measure where the sale actually happens.
How to brief an agency without getting robbed
Five questions that separate an agency that will move your number from one that will bill you for workshops. Written by an agency.
The short versionBring a number and a date. Ask who opens the file. Refuse discovery theatre. Demand receipts that name their client. Get weekly reporting against the target you agreed.
Reach beats followers
The Poster Project had 811 followers at the time and did 2.2M views in ninety days. Those two numbers in one sentence are the whole lesson.
The short versionFollower count is a subscription-era metric. Every platform now shows people what they will probably watch, whoever made it — so a small account competes on the same field as a big one. Make things strangers finish.
What AI-native delivery actually means
Not a chatbot on your website. A different shape of team, and a different set of things that are cheap.
The short versionAI-native does not mean the work is automated. It means the boring middle is, so the same people can hold the whole project — and a small team stops being a constraint.
How to price a creator product line
Margin first, design second. The table that decides whether a merch drop is a business or an expensive hobby.
The short versionBuild the margin table before anything is designed. If the product does not clear roughly half its retail price after everything, change the product, the supplier or the price — do not launch and hope volume fixes it.
How to read an ad account in ten minutes
The four screens that tell you whether money is working, and the vanity numbers that will waste your afternoon.
The short versionStart at the money and walk backwards. Revenue, then cost to get it, then which creative earned it, then what the landing page did with it. Anything that does not sit on that path is context, not a decision.
Tell us the
number. We’ll tell
you if we can.
We’ll tell you if we can hit it. If we can’t, we’ll say that too — and point you at whoever can.