Digital advertising
If your agency earns more when you spend more, you already know what it's going to recommend.
A large part of the market charges a percentage of your ad spend; some agencies publish it on their own sites, between 15% and 20%. We charge a flat fee. Recommending you cut the budget on a channel that isn't performing costs us nothing, which is why we'll tell you.
- Flat fee, not a percentage
- Your accounts, your billing
- Nothing goes live without measurement
Control KPIs · not promises
This is what we optimize against and what you evaluate us on every month. They're set from your starting point at kick-off, not from an industry average that doesn't know your margin.
| Month 2 | Month 4 | |
|---|---|---|
| CPA | Documented | −15 to −25% |
| ROAS (e-commerce) | 2.5× min. | 3.5×+ |
| Ad CTR | +0.5 pp | Top 20% of sector |
| Landing page conversion | Documented | +15% |
Working targets, not contract clauses. Google controls the auction, not us — and anyone who says otherwise is offering something they can't deliver.
What we find almost every time
Five things repeat across the accounts that come in for review. None of them gets fixed by raising the budget.
- 01
The report goes up and revenue doesn't
Impressions, clicks and reach improve month after month. Sales stay the same. What's being measured is the campaign's activity, not its effect on the business.
- 02
Conversion tracking is set up wrong
The event fires twice, or counts the thank-you page load, or was never tested. The account has spent months optimizing toward a number that isn't real.
- 03
You're paying for searches that were never going to buy
Without a maintained negative keyword list, part of the budget goes to neighboring terms: people looking for jobs, for free options, for your competitors.
- 04
WhatsApp contacts aren't measured
In Mexico and much of Latin America, that's where a good share of sales close. If that click doesn't count as a conversion, the channel that sells the most is the one that looks worst in the report.
- 05
Every month's recommendation is to raise the budget
Sometimes it's the right one. But when the person signing it earns a percentage of that figure, there's no way to know whether it comes from the data or from the invoice.
We start where demand already exists.
We don't switch on every channel in month one. First the one that captures people already looking for you; the rest come in when there's data to justify opening them.
-
Google Search
In front of people who already typed what you sell. The highest intent you'll find, which is why it's the starting point for almost any account.
-
Shopping & PMax
Product with photo and price inside the result. For online stores it's the difference between showing up and being compared.
-
Meta Ads
Facebook and Instagram to reach people before the search exists. With the Conversions API in place, not just the browser pixel.
-
TikTok Ads
Short video, where the creative matters more than the targeting. Recycling the Meta piece here doesn't work, so we don't do it.
-
Remarketing
People who already came in and didn't buy are still your cheapest audience. It's usually the most profitable part of the account and the most forgotten.
-
Creative production
Copy, static pieces and video generated with AI and reviewed by a person before publishing. Monthly refresh included.
Fourteen business days before the first ad.
That time isn't slowness. It's what keeps month one from being spent generating data that's later useless for deciding anything.
- Day 1–2
Kick-off
Ninety minutes to understand the business, the average order, the margin and the buying cycle. We leave with your starting numbers documented.
- Day 3–5
Measurement verification
GA4, Tag Manager, Google Ads conversions, Meta pixel and API. Tested event by event, including WhatsApp clicks and calls.
- Day 4–6
Marketing context
Who buys, why, who you compete against and which objections come up in the sale. You validate it before a single ad is written.
- Day 7–10
Campaign plan
Structure, audiences, bidding strategy and budget split by channel. It goes in writing and you approve it.
- Day 11–14
Launch
Approved copy, campaigns live and the first week of data running.
- Every month
Optimization
Negatives, bids, creative refresh and tests with a written hypothesis. Weekly update, monthly report and a thirty-minute call.
Four rules you can verify.
These aren't corporate-page values. They're concrete conditions, and if any of them breaks you can check it yourself.
- 01
Flat fee, separate from ad spend
You see two separate figures: what goes to the platforms and what you pay us. If your investment goes up, our invoice doesn't move. If it should come down, we lose nothing by saying so.
- 02
Nothing goes live without verified measurement
If tracking doesn't pass the test, the launch date moves. We'd rather start late than start blind. It's the rule that's most uncomfortable at the start and the one that prevents the most problems later.
- 03
The accounts are yours
They're created or kept in your name and under your billing. The history and accumulated learning stay with you the day you decide to leave, which is what forces the relationship to hold on results.
- 04
No automation publishes on its own
We use AI to produce and analyze, and that makes the work cheaper. Everything it generates stays paused until a person reviews it. The machine sets the speed; it doesn't make the decision.
When we're not the right fit
- You're looking for someone to guarantee a return or a position by contract.
- You need results this month and there's no room for two weeks of preparation.
- Your product doesn't have demand yet: ads amplify what exists, they don't invent it.
- You want a vendor who executes without asking. We're going to discuss the goal before the tactic.
What people ask before signing
-
Why a flat fee when most of the market charges a percentage?
Because the percentage puts the incentives on the wrong side. If your agency's income is a fraction of what you invest, every recommendation to raise the budget benefits them even when the data doesn't support it. And the recommendation to lower it costs them money, so it rarely arrives.
This isn't a theory about the industry. Some agencies publish their model openly, with rates between 15% and 20% of ad spend. It's a legitimate and very widespread model. It's just worth understanding what it produces.
What changes in practice
- The end-of-month conversation. With a flat fee, "we should cut Meta and concentrate on Search" is a normal recommendation. With a percentage, that same sentence reduces the invoice of the person saying it.
- The ceiling. A percentage grows with you without the work growing in the same proportion. Running an account twice the size doesn't cost twice the hours.
- What you're buying. A flat fee is quoted against concrete scope: which channels, how many tests, how often. It's much easier to discuss whether it's worth it.
Where a flat fee doesn't suit you
It's worth saying in full. With small ad budgets, a percentage can come out cheaper than any flat fee, because a percentage of a small figure is a small figure. The flat model starts paying for itself as the investment grows.
If you're in that range we'll tell you on the call, instead of selling you a retainer that doesn't serve you yet. Sometimes the right move is to wait, or to start by fixing measurement and the landing page, which costs less and pays off sooner.
How we quote it
One fixed monthly figure, separated from your ad spend in the same document. No commission on spend, no separate charge for creatives, no fee for reporting. If the scope changes — a new channel comes in, the number of tests goes up — it's re-quoted and you approve it before work starts.
-
I've worked with an agency before and it didn't work. Why would this be different?
It's the right question and it deserves something better than "because we actually know what we're doing". Most of the accounts that come in for review didn't fail for lack of talent. They failed because of three things that can be checked in an afternoon.
The three causes that repeat the most
- Measurement was never right. The event counted thank-you page visits, or fired twice, or didn't exist for the channel that sold the most. Everything else — bids, audiences, creatives — got optimized toward that number. Months of work on a wrong foundation.
- Nobody defined what a result was. Without a target cost per lead and without knowing what a customer is worth, there's no way to say whether a campaign is going well. You end up reporting what can be measured instead of what matters.
- The problem wasn't in the campaign. The ads brought qualified traffic and the page lost it, or the lead arrived and nobody replied the same day. Adjusting ads there is pulling the wrong lever for months.
What changes here
The measurement review comes before launch, not when something smells wrong in month three. If it doesn't pass, the date moves. It's the part of the process that creates the most friction at the start, and it's non-negotiable precisely for that reason.
Then, targets are documented on your numbers and not on industry benchmarks. If today you don't know what one of your customers is worth, that gets calculated at kick-off: without that figure nobody can tell you how much you can pay to acquire one, and any budget proposed to you is a guess formatted as a proposal.
What you can demand from anyone
Even if you don't work with us, there are three things worth asking for before signing: that they show you the conversion event working live, that they tell you whose name the accounts will be in, and that they explain which number they'll optimize against. If any of the three gets a vague answer, there's your answer.
-
How much do I need to invest in ad spend?
It isn't decided from how much you want to spend. It's decided from how much a customer is worth to you. That's the number that orders everything else and the one that's almost never on the table when someone asks for a proposal.
The math, in order
- How much a sale leaves. Margin, not revenue. And if your customer buys again, how much they leave over a year.
- How many leads you need to close one. If you close two out of ten, each customer costs five leads. Without that figure the budget is a gamble.
- What you can pay per lead. It comes from dividing the first by the second. That's your ceiling, and with it you know whether the channel is viable before spending a cent.
- The learning period. In the first weeks the account generates data, not performance. It's budgeted as part of the start-up, not as a result.
Concentrating pays off more than spreading
The most expensive mistake we see is opening five channels with a budget meant for one. None gathers enough conversions for the algorithm to learn, all of them deliver noisy data, and by month three there's nothing that can be decided with confidence. It's better to master one channel, confirm it's profitable and grow from there.
What no budget fixes
Ads multiply what already exists. If the landing page is slow, if the ad's message doesn't match what the person finds on arrival, or if the lead waits a day for a reply, raising the investment just makes the same waste bigger. That's why those three points are checked before switching on, and when one fails we say so even if it delays the start.
Scaling on evidence
When a campaign shows measured profit, the goal stops being containing spend and becomes growing while the return holds. With a flat fee that conversation is clean: our invoice doesn't move, so the recommendation to scale only shows up when the numbers support it.
-
What will you deliver and how often?
"Campaign management" is a black box in most proposals. This is what concretely happens, so you can judge whether what you pay matches what gets done.
The rhythm
- Every week. A short message: what happened, what was adjusted and what's next. It isn't a formal document; it exists so you never have to ask how your account is doing.
- Every month. A report with cost per lead, cost of acquisition, return and comparison against your starting point. It includes what didn't work: a report where everything goes well every month is a report nobody is reading.
- Every month. A half-hour call to decide, not to present slides.
- Every quarter. A longer review: which channels to keep, which to close and where to move the budget next quarter.
The work you don't see
Within each month there are tasks that don't shine in a presentation and are what sustains performance: reviewing search terms and adding negatives, adjusting bids where the data calls for it, refreshing creatives before frequency wears them out, running tests with a written hypothesis and a decision criterion set in advance, and re-checking that measurement is still alive.
That last one matters more than it seems. Conversions break on their own: someone touches the site, the platform updates, the consent banner changes. If nobody checks, the account keeps optimizing for weeks toward a data point that no longer exists.
The data is yours, always
Everything comes from your own accounts and you have permanent access. You can log in any day and see exactly the same numbers we see. The report exists to interpret them and decide, not to be your only window into what's happening.
-
Can you guarantee results?
No. And if someone guarantees it to you by contract, either they don't understand how an auction works or they know they won't be able to deliver.
Your ads compete in real time against competitors whose budgets and strategies change daily, on platforms that modify their rules without notice. Nobody controls that system. What can be controlled is the method, and that we do answer for.
A target isn't a guarantee
We publish control KPIs: a reduction in acquisition cost by month four, a minimum operating return for online stores, CTR improvement over your own baseline. They're criteria we optimize against and that you evaluate us on, not contract clauses.
The difference matters. A target says where we're working toward and gives you something to judge us on every month. A guarantee claims control over variables nobody has. We'd rather have the former in writing than the latter hidden in an appendix.
What we do commit to
- Documenting your starting numbers at kick-off — the real ones, not the brochure ones.
- Reporting every month, including what went wrong.
- Optimizing against cost per lead, acquisition cost and return. Not against clicks or impressions.
- Telling you when something isn't performing, even if the conclusion is to cut investment or close a channel.
If by month three we're not doing well
The minimum commitment is three months and there's a technical reason behind it: less than that isn't enough to get out of the learning phase and have data to decide anything with. Once that period is up, if the direction isn't right we review it with the numbers in front of us and adjust, or we tell you this channel isn't right for your business today.
We'd rather lose a retainer than keep an account that gives the client nothing back. It isn't generosity: an account that doesn't perform ends up getting cancelled anyway, and with a worse memory.
arborix Almost no one knows where to start.
That's why we begin with a diagnostic: we review what you already have running, tell you where the most expensive leak is, and give you a 90-day plan. With that, you decide what to hire — or whether to hire anything at all.
Reply within 24 business hoursFree callNDA signed by Arborix