Most of them will quote you a price to translate what you already have. That is the cheapest version of the job and the one least likely to work. Here is what the work actually is — and the question about money that tells you who is sitting across from you.
You already know the market is there. What almost nobody prices correctly is how little competition there is inside it. In the first quarter of 2025, less than 1% of US retailers' digital ad spending reached Spanish-language websites — in a country where roughly one in five people is Hispanic. That is one slice of one channel rather than all advertising, which makes it narrow; it is also the slice where attention is easiest to buy, which makes it damning.
Source: Nielsen, 2025 Hispanic consumer report. We laid out the full size of the market — GDP, consumption, the ad-spend gap — in choosing a bilingual marketing agency.
A gap that wide is not caused by ignorance. Every CMO in America has seen these numbers in a deck. It persists because the work is harder than it looks, and because the first attempt usually fails in a way that teaches exactly the wrong lesson: we tried Spanish, it didn't convert.
A funnel is not copy. It is a sequence of decisions a stranger makes, in order, about whether to trust you. Translation moves the words across and leaves every one of those decisions exactly where it was — built for somebody else.
Three quarters of US Latino adults say they can hold a conversation in Spanish pretty well or very well, and most of the US-born can too. But the same research finds fluency drops hard by the third generation, and that most Latinos say speaking Spanish is not what makes someone Latino. So «Spanish or English?» is the wrong question. It assumes one audience where there are several, separated less by language than by which country's version of a promise they grew up distrusting.
Source: Pew Research Center, Latinos' Views of and Experiences With the Spanish Language — 3,029 US Latino adults, 2022.
An agency actually doing this job rebuilds four things, and none of them is a translation task:
The first doubt is rarely price. It is more often «will this company still exist in a year», and for buyers with family abroad it can be «what happens to my money if I stop paying». Those get answered where they appear, not in a footer FAQ.
Credibility does not travel. A testimonial from someone who does not sound, look or spend like your buyer is not proof — it is decoration in a second language.
Installment plans, debit-first buyers, shared household decisions. Checkout is where translated funnels die quietly, and the dashboard records it as «drop-off».
Hispanic viewers spend 55.8% of TV time on streaming against 46% for the total US, and Spanish-speaking podcast listeners are reported 62% more likely to act on a podcast ad. Search plus a Meta retarget is fishing where this audience isn't.
If they don't have one, they are going to translate your funnel and then charge you for the rebuild.
Everything above can be performed in a pitch. This next part cannot, because it changes what the agency does on a Tuesday when nobody is watching: what happens to them if your funnel doesn't sell?
For most agencies, in most contracts, the answer is nothing. They invoice the same. That is not a scandal — it is the industry standard, and it has held for half a century. The Association of National Advertisers has tracked agency pay since the 1960s, and its 2016-17 study found the use of performance incentives had declined for the first time in fifty years. Advertisers gave two reasons: the incentives did not improve agency performance, and designing and administering them was complicated and time-consuming.
That is the strongest argument against what follows, so it goes first. But notice what was measured: a bonus of roughly 5-15% sitting on top of a fee that already covered payroll, rent and margin. Nobody restructures a company over a tip. Co-production — where the fee is small or absent and the agency's income is the share — is a different animal, and a far rarer one.
advertisers spending under $30 million a year had any part of their agency's pay tied to the result. Among large advertisers it was 70 in every 100.
That size breakout is from 2010 and has not been replaced by an openly published one with the same cut — read it as an order of magnitude, not as this quarter's photograph. The decline in incentives comes from the 2016-17 study, covering 82 member companies and more than 1,100 agreements.
The practical conclusion is not «a revenue share is better». It is simpler and less comfortable: if someone offers you one, they are doing something 92% of their industry does not do. Worth understanding why before you celebrate.
Assets shipped, meetings held, dashboards sent. Success means the work got done, the market risk is entirely yours, and they will agree with you about almost everything: arguing with the person signing the check pays nothing.
Price, offer, refunds, cost per sale. Success means money came in and stayed in, the risk is split, and they will tell you no — to your favorite idea, your date, your price. That is the uncomfortable part and also the value.
The difference is not effort or talent. Both agencies can have the same people. What changes is where their interest points when nobody is watching.
Both models are identical right up to the sale. The entire difference is one arrow: whether a path exists back from the sale to the pocket of the people who did the work.
The sale happens either way; it just never reaches the pocket that did the work
| Model | Paid up front | Who carries the risk | When it makes sense | Where it breaks |
|---|---|---|---|---|
| Flat fee | All of it, monthly, against a calendar | You, entirely | You need execution capacity and already know what to execute | You are buying activity and nobody defined what winning means |
| Fee + bonus | Nearly all; the bonus is 5-15% of the total | You, nearly entirely | You want alignment without changing the relationship | The bonus is too small to change a single decision |
| Co-production | Little or nothing; sometimes a floor covering hard costs | Split: the partner puts in time and sometimes media budget | Validated offer, activatable audience, wide margin, a partner with judgment | In the revenue base, and in who decides when you disagree |
| Commission only | Nothing | Whoever sells | Pure affiliate motion, no asset being built | Nobody invests in what they don't own — not brand, not list, not launch two |
If you go the revenue-share route, the percentage is rarely what sinks the deal. The sentence next to it is: a percentage of what? Booked or collected. Before or after media spend. Before or after refunds, chargebacks and processor fees. For the launch week only, or for everything that list buys over the next two years. Settle those four in writing and the number almost negotiates itself.
Everything above is judgment. This is arithmetic, and it is yours. Move the sliders: the calculator returns the revenue level at which a share stops being the cheap option and becomes the expensive one.
Runs entirely in your browser. Nothing is sent anywhere.
—
Read the second-to-last line before you move on. In the failure case, the fee does not come back. That is the insurance you buy when you agree to overpay in the good case — a cash-flow decision, not a matter of pride.
An agency that shares risk does not pick interesting projects. It picks projects it can get paid on. These are roughly the five questions it asks before saying yes.
1. Have you already sold this offer — or one close to it — to someone who did not know you?
2. Do you have a list, community or past customers you could activate in ninety days?
3. Would you give access to your payment processor and ad accounts so the numbers are visible without a middleman?
4. Does your margin per sale survive giving away a share and still fund delivery?
5. If a partner told you «this offer isn't ready yet», what would you do?
Answer all five to see your verdict.
Saying who this doesn't serve is the only thing that makes the rest credible. Don't hire on a revenue share if:
Ask the question anyway, in the first meeting, before the proposal: what happens to you if this doesn't sell? The answer tells you who you are talking to. The Spanish edition of this piece, qué es de verdad coproducir un lanzamiento, works the same numbers from the launch side.
EMOTIA is a bicultural performance team taking on a small number of founding clients in the US. The application takes under three minutes, and if we think it isn't your moment, we'll say so.
Apply to work together