Scale your buying or open a new vertical: a decision based on numbers

How an affiliate team decides where the next $100,000 goes: into the current campaign or into a new vertical

In media buying this decision comes up constantly. The campaign works, ROI holds, the team spends $80,000 a month and wonders: pour more into the same thing, or open a new geo, a new vertical, a new traffic source.

The problem is that in paid traffic both options mislead you — just in different ways. Scaling looks safe (the campaign is already proven), but ROI starts dropping exactly when you spend more. A new direction looks promising because it has no bans, no burned-out creatives and no cut caps yet — but the test can eat three months of budget and return nothing.

Let's look at how to compare the two scenarios in numbers.

Why the mistake costs more in affiliate marketing

In a classic business, a wrong growth decision shows up a year later. In media buying it shows up in six weeks — and along with the budget you lose two more things.

The team. A buyer pulled off a working campaign to test a new vertical loses momentum on the old direction within two months. When you move them back, the audience has changed, competitors have taken the space, creatives are burned. Recovery takes longer than the test itself.

The advertiser relationship. If you shifted volume from a stable offer to a new one and missed the plan, your cap can be cut or moved to worse terms. Getting the previous terms back is harder than winning them the first time.

That's why a growth decision here needs a more precise calculation than in a business with a longer cycle.

Step 1. Calculate how much is left in the current campaign

The main illusion of scaling paid traffic: if ROI is 45% at $20,000 a month, it will be the same at $60,000. It almost never is.

ROI as spend grows

Build a table by month or by week: how much you spent, what ROI you got.

Period	Spend	Revenue	ROI	Profit
Month 1	$18,000	$27,900	55%	$9,900
Month 2	$26,000	$38,400	48%	$12,400
Month 3	$34,000	$47,600	40%	$13,600
Month 4	$44,000	$57,200	30%	$13,200

Profit grew through month three, then fell in month four despite $10,000 more spend. That is the point where scaling within the current campaign stops making money.

Calculate the marginal profit on every extra dollar of spend:

• Month 2 → 3: extra spend $8,000, extra profit $1,200 → 15% marginal ROI
• Month 3 → 4: extra spend $10,000, extra profit –$400 → negative

If marginal ROI has fallen below 15–20%, you are already paying more for volume than it is worth.

What exactly is the constraint

When ROI drops as spend grows, the cause is usually one of four:

Audience. The core is exhausted and the algorithm moves into lower-converting segments. You see it as rising CPM and CPC on the same creative plus falling landing-page CR.

Offer cap. The advertiser won't take more leads, or takes them at a reduced payout. This ceiling isn't yours — it's external.

Traffic quality. FTDs are there but retention is falling. For rev-share deals this is critical: volume grows while cohort payouts get worse than before.

Operational capacity. Number of accounts, payment methods, moderation, number of buyers. Unlike the first three, this constraint can be removed with money.

The distinction matters: if capacity is the constraint, an investment fixes it and scaling still has room. If audience or cap is the constraint, the room is gone and no amount of money changes that.

Step 2. Calculate the full cost of entering a new direction

The main mistake here is counting only the test budget. The real cost of entry has four parts.

Test budget until the first working campaign. How much you need to spend to find a working setup in a new geo or vertical. For a new geo in a familiar vertical that is usually $8,000–15,000. For a new vertical with different mechanics — $20,000–40,000.

Buyer time. A buyer on a test brings no profit. If they were pulled off a working direction, you lose their current profit. If it's a new hire — salary plus 1–2 months to ramp up.

Infrastructure. New accounts, new payment methods, sometimes new anti-detect setups and proxies per geo, landers and translations, an affiliate network with the right offers. For a new vertical — often new analytics and a tracker for a different funnel.

Time to stable profit. Not to the first conversion, but to the moment the direction is profitable three months in a row. For a new geo that's 2–4 months; for a new vertical, 4–8.

A worked example

A gambling team on Tier-2 geos is considering moving into nutra.

Item	Amount
Test budget to find a working campaign	$28,000
Buyer pulled off a working direction (3 months)	$19,500 of forgone profit
Infrastructure: accounts, landers, translations	$6,500
New tracking and analytics for a different funnel	$3,000
Buyer and media-buyer salary during the test	$16,500
Full cost of entry	$73,500

Against the "let's allocate $28,000 for a test" figure usually quoted at the start, that's a 2.6x difference.

Step 3. Compare both scenarios over the same horizon

Use the same budget and the same period. Below is an example for a team currently spending $44,000 a month with $120,000 of available resource over 12 months.

Scenario A: expanding the current direction

The capacity constraint is removed — more accounts, two more buyers, expansion into two adjacent geos in the same vertical.

	Mo 1–4	Mo 5–8	Mo 9–12
Investment	$55,000	$40,000	$25,000
Spend	$190,000	$260,000	$310,000
Average ROI	34%	30%	27%
Profit	$64,600	$78,000	$83,700
Net effect	+$9,600	+$38,000	+$58,700

Cumulative: +$106,300. ROI slides gradually, but volume compensates.

Scenario B: entering a new vertical

	Mo 1–4	Mo 5–8	Mo 9–12
Investment	$73,500	$30,000	$16,500
Spend in the new vertical	$28,000	$95,000	$180,000
Average ROI	–40%	18%	42%
Profit of the new direction	–$11,200	$17,100	$75,600
Loss in the core direction	–$19,500	–$8,000	$0
Net effect	–$104,200	–$20,900	+$59,100

Cumulative: –$66,000 for the year — but in the final period the new direction reached the level of the core one and keeps growing, while in scenario A ROI keeps sliding.

How to read this

Over 12 months, scaling wins by a wide margin. But the trajectories differ: in scenario A ROI falls from 34% to 27% and will keep falling; in scenario B it rises.

So the question isn't "which is better", but "over what horizon are you deciding, and can the team survive a year of running the new direction at a loss".

Step 4. Three filters that override the calculation

The liquidity filter

Find the worst month in the new-direction scenario — the point of maximum accumulated loss. In the example above that's the end of month four: –$104,200.

If that amount exceeds your free reserve minus three months of operating expenses, the scenario is impossible, however attractive it looks in month 12.

In media buying this filter is stricter than elsewhere, for two reasons: money is locked in spend until the network pays out (often net-15 or net-30), and some networks may delay a payout or shave it during a traffic-quality dispute. The reserve has to cover not only the test but also the cash gap on payouts.

The team filter

Who specifically runs the test, and what happens to their current direction.

If the answer is "we'll take the best buyer, they'll figure it out" — calculate how much profit they generate now and add that amount to the cost of entry. The best buyer is the most expensive precisely because they bring in the most.

If the answer is "we'll hire someone new for the new direction" — budget 2–3 months for them to reach results, and factor in the chance that they never do.

The transferable-experience filter

How much the new direction leans on what you already have.

A new geo in the same vertical on the same source — creative approaches, funnel, audience understanding and the network all transfer. Cost of entry is low, risk moderate. In practice this isn't a new direction, it's an expansion.

The same vertical, but a new source — offers and funnel transfer; everything about running traffic doesn't: moderation approaches, campaign structure, account management. Medium risk.

A new vertical on a familiar source — the technical side transfers; audience understanding, lead economics and advertiser requirements don't. High risk.

A new vertical on a new source — nothing transfers except money. This is launching a team from scratch inside an existing company, and it should be modelled as a separate startup.

When the answer is obvious before the math

Scale if marginal ROI on extra spend is still above 20%, the constraint is operational capacity (accounts, buyers, payment methods) rather than audience, and the advertiser is willing to raise the cap.

Open something new if ROI has fallen three months in a row at the same spend, the cap was cut or the offer closed, the source started banning your approach at scale, or the geo burned out — competition has pushed CPM so high the economics don't work at any volume.

Do neither if your reserve is less than two months of operating expenses plus current spend. In media buying, running out of liquidity isn't a risk — it's a stop: there's nothing left to fund even the working campaign.

Look at a third option if both scenarios are weak. Often the biggest gain comes not from new volume but from working with what you already have: moving from CPA to rev-share where player LTV justifies it, renegotiating network terms on volume, lowering cost per lead through funnel optimisation, or dropping the worst-ROI geos to concentrate the budget.

How to make the test cheaper if you decide to run it

Test one variable. A new geo on a familiar source and vertical — that's one test. A new vertical on a familiar source — that's another. Changing two parameters at once produces a result you can't draw conclusions from: you won't know what actually failed.

Set the stop-loss before you start. Agree in advance: if after $X of spend there is no campaign with ROI above Y — we close it. Without that, the test turns into "just a bit more, we're almost there" for six months.

Separate analytics from day one. The new direction runs its own P&L: its own spend, its own payouts, its own team salaries, its own share of infrastructure. If costs blend into the core direction, you won't know whether it turned profitable — and you'll keep funding a loss believing it's a profit.

Network terms agreed before the test. Terms, cap, payout schedule and traffic-quality rules — before the first dollar is spent. In a new vertical the traffic requirements are often different, and learning them after the first shave is expensive.

Summary

The decision comes down to four numbers.

1. Marginal ROI on extra spend in the current direction — it shows how much is left there, more precisely than any reasoning about market potential.
2. The full cost of entry into the new direction — including buyer time, infrastructure and forgone profit, not just the test budget.
3. The maximum accumulated loss in the new-direction scenario — and whether your reserve covers it given network payout terms.
4. How much existing experience transfers — the less it does, the higher the expected return has to be to justify the risk.