Blog · November 3, 2026 · 6 min read

What our first $30 of Meta ads taught us

By Pavan, Founder, FirstReaction · Updated

In short

Our first Meta campaign spent ₹2,534 (about $30) for 2 landing-page views and no sign-ups, because we optimized for an event we'd never had, launched six ads on a tiny budget, and let Meta pick the winner. Switching to landing-page views, broader placements and four ads got 20 landing-page views for ₹744 on the first day.

We build a tool for deciding which ads deserve your budget before you spend it. Then, in late October 2026, we launched our own first Meta campaign and let Meta make that decision for us. Here's what happened, what it cost, and what we changed.

What we launched

On 27 October we published one campaign with one ad set and six ads: two videos, two carousels and two single images. It ran on Instagram only, to a US audience aged 25 and up, with Meta's Advantage+ audience and a few interest suggestions. The goal was sign-ups, and the budget was about ₹1,000 (roughly $11) a day.

It looked reasonable. It wasn't.

What happened in the first two days

By the end of 29 October:

Metric28–29 October
Spent₹2,534 (about $30)
Impressions318
Link clicks8
Landing-page views2
Sign-ups0
Cost per 1,000 impressionsabout ₹8,000 (around $90)

And the money didn't spread across the six ads. Meta chose for us:

AdSpendLink clicks
Single image₹1,5074
Video₹7401
Carousel₹1832
Single image₹720
Carousel₹251
Creator-style video₹60

One carousel earned 2 clicks from ₹183, a far better click-through rate than the image that got ₹1,507. It never got a real chance. On 318 impressions, nobody, Meta included, could tell which ad was actually best.

The four mistakes

  1. We optimized for an event we'd never had. Meta's delivery learns from results. With zero sign-ups, it had nothing to learn from, so it kept spending on guesses.
  2. We launched too many ads. Six ads on $11 a day means most of them get almost nothing. Meta doesn't run a fair test; it backs an early favorite.
  3. We narrowed placements to Instagram only, which made each impression expensive.
  4. We let delivery pick our winner. The decision about which ads deserved money got made by an algorithm looking at a few hundred impressions, instead of by us, before launch.

What we changed

On 30 October we turned the first campaign off and launched a new one:

  • The goal became landing-page views, an event that happens dozens of times a day, so delivery had something to learn from.
  • Facebook and Instagram, instead of Instagram alone.
  • Four ads instead of six, including a reworked version of the main image ad with a clearer line about what FirstReaction does.

First day of the new campaign:

First campaign (2 days)New campaign (first day)
Spent₹2,534₹744 (about $9)
Landing-page views220, at ₹37 each
Cost per 1,000 impressionsabout ₹8,000about ₹1,080

The reworked image ad got 13 link clicks from 289 impressions (about a 4.5% click-through rate) and 10 landing-page views. The carousel got 9 clicks and 10 landing-page views. The creator-style video got ₹67 and no clicks, and one video got no delivery at all. Even with fewer ads, Meta put almost all the money behind two of them.

What we'd do from day one

  • Decide which ads go in before launch. Simulate every version against the same audience, and launch only the best two. Since Meta won't split the budget fairly, that decision has to happen before launch.
  • Optimize for an event you'll get every day until real conversions start coming in.
  • Start with broad placements and let delivery find cheaper impressions.
  • Give each ad enough impressions to judge, or use a proper A/B test, before calling a winner.

None of this is new advice. It still took ₹2,534 to make it stick, which is the whole reason FirstReaction exists: the cheapest lessons are the ones you learn before the money goes out.

For the step-by-step version of this, read our guide to validating ad creatives on a small budget.

Sources

  1. Meta Business Help Center: About the learning phase