Small-budget ads: three tests before a maker spends $100 on traffic

Conclusion: advertising does not buy sales, it amplifies something already verified. Fail any of the three tests below and ads will scale your loss with great precision.

Test 1 — does the unit economics pass

Every paid order must carry its acquisition cost. Compute three numbers first (the pricing calculator produces the first two):

Number	How	Why it comes first
Gross margin per piece	price − (material + labour + failure rate + shipping + returns + platform fees)	If margin is zero, ads are a pure loss
Maximum affordable CAC	gross margin × 0.5 (allowing the first order to lose half, expecting a repeat to repay it)	Stop above this line
Repeat / list value	what share of buyers comes back	The only thing that justifies losing money on order one

One judgement: a piece with margin under 30% of price should not be advertised. Fix pricing and cost first — most makers are underpaid on their own hours, not under-exposed.

Test 2 — do you already have a photo that sells

Ads buy click and conversion efficiency, and efficiency lives almost entirely in the creative.

• A winning creative looks like this: in organic traffic one main photo already clicks and converts clearly better than the rest.
• Launching ads without one means paying to test a hundred photos — the most expensive way to learn what a photo does.
• The cheap order: test photos on organic traffic (free) → pick the winner → pay to scale it.

Engineering note from the assistant: the creative is the only asset you can carry. Accounts, keywords and audiences all move with platform rules; a photo and a caption that have worked travel with you to any channel.

Test 3 — can you read four numbers

Dashboards offer dozens. At small budgets watch four and ignore the rest:

1. CPM (cost per mille): absurdly high means the audience is too broad.
2. CTR (click-through rate): under 0.8% is usually a photo problem, not a bid problem.
3. CPC (cost per click): compare with margin — CPC above margin ÷ expected conversion means stop.
4. Orders, not the ROI report: under about ten orders, “return on ad spend” has no statistical meaning.

Minimum viable habit: record three columns a day — spend, clicks, orders. After two weeks you will see a pattern, and it beats any dashboard.

How to spend $10 a day: a 14-day structure

Days	Do	Stop line
1–3	One winning photo, 3–5 narrow long-tail phrases, low bid	CPM is absurd (audience too broad) → change phrases
4–10	Keep only what runs, log the three columns daily	Three consecutive days with zero orders → stop
11–14	Keep one winner, raise budget 20% and watch orders	Orders rise with spend → demand is real, keep scaling

The point of this structure is not profit; it is buying one answer: does this piece survive paid traffic. Once you have the answer, scaling has a meaning.

Three kinds of traffic not to buy

1. Broad head terms (“gifts”, “handmade jewelry”): comparison shoppers, near-zero conversion; the platform keeps the money.
2. Boosted posts: buys vanity metrics, produces no orders.
3. “Targeted buyer packages” from unknown vendors: the handmade ad pool is one of the most heavily farmed places on the internet. Anything promising guaranteed impressions or guaranteed enquiries is a fee extraction.

The judgement: ads are leverage, not the engine

“In essence we gather people with the same need together, so we do not have to advertise to find orders — the orders find us.” — from OJ’s own notes (redacted)

Read it backwards and you get the boundary of advertising: ads accelerate “the orders find us”, and that requires a “us” to exist first. The site, the product line and the email list are the engine; ads are the lever. With no engine, a longer lever is only more dangerous.

Assistant’s note: this is why the pricing and photography guides sit before this one. The order of the topics is the cost structure of the business.