You already know how to make people stop scrolling. What happens next — the offer, the media it runs in, the page that has to close it — is a different craft, and it decides what the attention was worth. That part is ours.
A narrow list, and the brief is done by the same hands that hear it — nothing gets relayed, nothing gets inherited. If a job sits outside the four, we say so on the first call instead of learning on your budget.
We plan the campaign, build it, run it and read it: budget split, audiences, what gets tested first, what gets cut in week two. Where a placement has to be negotiated with a person rather than won at auction, we negotiate it in your name and you see the price we pay. On anything we buy, the media cost and our fee are two separate lines on the invoice.
Ad creative, campaign visuals, written content, and the applied version of your visual identity — the version that has to survive being cropped to a square on a small screen. When something has to be filmed, or you have to be somewhere on a specific day, we organise the production around it. You get the working files, not flattened exports.
The site is where the money is actually made, so we build it and we keep it running. Pages that load, one clear action per page, forms that reach an inbox you control. Alongside it, the reporting: what came in, what it cost, which campaign it came from, in a document you can read without a login.
Before the budget: who the audience actually is, what they already buy, and what they would pay you for. Then the words — what you sell, to whom, in what order, said the same way on every surface. Then the structure around it, so the business stops depending on you answering messages at midnight. You end up with a written position, the sentences to say it in, and a map of how an enquiry becomes a payment.
Five phases. Each one ends with something you can hold, and nothing moves on until the one before it is on paper and you have read it.
When: Week one.
You give us access rather than summaries: the ad accounts, the analytics, the site, the sales numbers if there are any. We read them ourselves. We also read what your audience is saying and what the people selling to them are charging.
a written read, four pages at most, saying where the money currently comes from, where it is leaking, and the three things we would change first. Plain sentences. It is yours whether or not you continue.
When: Week two.
We turn the read into one page. Not a deck — a deck is built to be presented, and this has to be built to be argued with. It is written by the person who did the reading, and where the reading changed our mind about what you asked for, the page says so instead of quietly routing around it.
a single page naming the platforms, the budget split between them, the two things we are testing first, what result would prove each test right, the date we review it, and what we are deliberately not doing yet. Disagreeing with a plan is cheap. Disagreeing with a campaign that has already run is not.
When: Weeks three to five, depending on how much has to be made.
Accounts, tracking, audiences, creative, landing pages. If something does not exist yet — a page, a set of ads, a way to take payment — it gets built before anything is spent. And tracking is tested end to end before a single campaign goes live.
the accounts and the tracking built and tested, the creative in editable files, and a one-page map of what was set up and where it lives.
When: Ongoing, weekly.
Campaigns start small and staged, so a result can be traced back to a cause. Changes are made during the week, one at a time — a week that changes five things is a week that teaches nothing.
a weekly note — what we spent, what came back, what we changed and why, what happens next week, and anything we need from you. A longer monthly review that reaches a decision rather than a summary: continue, adjust, or stop. One call a month, or more when there is a reason for one.
When: Whenever you ask, including the day you leave.
You do not have to be leaving to ask for this. Any part of it can be handed over at any point in the work, and the last month of a good arrangement should look exactly like the first.
everything. Ad accounts, domain, site files, creative files, audience data, every document. Nothing is withheld over a final invoice, and nothing arrives in a format only we can open. We remove our access and confirm in writing that it is gone.
Not a client list. A description of the problem you probably arrived with. If one of these is close to yours, write to us.
Views are up, revenue is flat, and nobody can point at where the gap opens. The content is doing its job. Everything downstream of the content was built in a hurry, once, and never revisited. Most of this work is structural, and it pays for itself before any advertising starts.
Something is running. It was configured months ago by someone who is no longer around, and nobody trusts the number that comes out of it. Money went out, some of it came back, and the reason is still a mystery. That mystery is what makes spending again feel like gambling. This work starts with reading, not spending.
Something real is being sold — a service, a drop, a membership — and the only thing driving it is whichever post happened to land that week. It works until it doesn't, and it cannot be planned around, budgeted for, or repeated on purpose.
Creative, commercial and editorial decisions all go through one calendar. Everything works, and everything waits. The work here is taking specific pieces off that calendar permanently, not "supporting" you.
An agency without opinions is a pair of hands. These are the ones that change what we would do on a Monday morning, and each of them costs us work we would otherwise be paid for.
A follower count sits on a platform that can change its ranking overnight, without telling you and without owing you anything. A mailing list, a site, a record of who bought what — those survive the change. So most of the work is converting the rented thing into the owned thing while the rent is still cheap, which means budget goes to a mailing list before it goes to reach, in the months when reach would have made our reporting look better.
A spike brings people who came for the post, not for the person. They arrive with no particular reason to come back, and the platform that sent them is under no obligation to do it twice. Six months of steady spending leaves you with an audience you can advertise to again, a creative library you can reuse, and enough data to act on. It is the duller thing to sell and the only one that compounds, so we will not build a plan around a hit.
Someone watches, wants, taps — and lands somewhere that makes no reference to the thing they just watched. The interest was real for about four seconds and nothing on the other side recognised it. Repairing that handoff costs less than making more content, and it is the only repair that makes everything already published worth more. It is also the least flattering thing we can say to a person who makes the content, and the smallest invoice we could send them. We say it anyway.
Every campaign is a guess about who buys. You can test that guess by reading and asking for a week, or you can test it with money, at auction speed, in public. Both answer the same question. Only one of them sends the bill first. Research is the cheaper instrument, so it goes first — and if the brief is to skip it and put money behind a post this week, we decline the work rather than take it and hope.
A result you cannot explain cannot be repeated, scaled, or defended to whoever paid for it. So the plan states in advance what would count as this working and what would count as a reason to stop — including the number at which we would tell you to stop paying us. A prediction made afterwards is not a prediction.
Ad accounts, tracking, domain, site, creative files, audience data — created in your name, on your business account, with access granted to us rather than held by us. This is plainly against our commercial interest. An agency that holds your accounts is difficult to leave, and difficulty is not the same thing as loyalty. We would rather be kept for the work.
None of this is a philosophy. It is what we would do with your money, written down before you have given us any.
Most agencies keep this list for the second meeting. Reading it now is cheaper for both of us. There are also two things we do not sell at any price: a guaranteed number, and an agency that agrees with you.
Not once, not as a favour, not to make a launch look busier than it is. Bought numbers make every real number after them unreadable, and the first thing they cost you is the ability to tell whether anything is working.
If a figure cannot be traced back to a source, it does not go into an ad, into a report, or into a sentence about ourselves. That rule is the reason there is not a single claim about our own results anywhere on this website.
Reach is an input, not an outcome. If nobody can say what a million views would be worth, there is no way to judge the campaign and no reason to run it. We will say so, and we are not the right people to take your money for it.
Spending on a page that cannot close, a product that cannot ship, or a price nobody has tested is an expensive way to buy proof that you have a problem. We will tell you it is not ready and tell you why, which is not what anyone hires an agency to say.
Write two paragraphs: what you have built so far, and what you want to be true in six months. The reply comes within two working days, and the first exchange and the call after it cost nothing and commit you to nothing. If we are not the right agency for it, you get that answer instead — usually with what we would do in your position.