Playbooks
How to Split a $1,000 Monthly Ad Budget
Published September 20, 2026
A small ad budget fails by dilution more often than by competition: $1,000 spread across six campaigns buys six unreadable experiments, and the month ends with opinions instead of data. The fix is a split decided before the month starts, with each slice assigned a job and a signal that moves money.
This playbook lays out one workable split — 70% proven, 20% testing, 10% in reserve — then covers the four numbers that decide where money moves, a week-by-week first-month calendar, and the margin math that must check out before anything scales. Every metric named here has a matching calculator linked below; the framework is the part worth remembering.
The short answer
Split a small ad budget into three jobs: proven campaigns get about 70 percent, one or two controlled tests get 20, and 10 stays reserved for what the first two weeks reveal. Judge tests on CTR and conversion rate first; scale only when ROAS covers costs with margin to spare.
The 70/20/10 split, and why the reserve exists
On a $1,000 month, the split assigns every dollar a job before the first click arrives. The proven slice keeps what already converts fed; the test slice buys exactly one or two experiments — not six; the reserve exists so that mid-month reallocation happens on schedule, from data, rather than on mood whenever a campaign has a loud week.
| Slice | Budget | Job | Signal that moves money in |
|---|---|---|---|
| Proven | $700 | The one or two campaigns already returning conversions | ROAS at or above target — keep it fed, but do not grow past stable delivery |
| Testing | $200 | One new keyword theme or one new audience at a time | CTR in week one, conversion rate by week two |
| Reserve | $100 | Unspent until the first two weeks report | The month's best performer receives it — not the loudest one |
The four numbers that decide scaling
Four metrics cover nearly every scaling decision at this budget size, and the order of reading them matters as much as the numbers themselves: CTR diagnoses the ad, conversion rate diagnoses the page, and ROAS is the verdict that arrives last.
| Metric | Sanity range to start | What it tells you |
|---|---|---|
| CTR | Search 3-6%, display under 1% | Does the ad match the query? Low CTR is a message problem before it is a bidding problem |
| Conversion rate | 2-5% on e-commerce landing pages | Does the page keep the promise the ad made? |
| CPM | Varies by market — compare only within your own campaign, month over month | How expensive attention is; rising CPM with flat CTR means auction pressure, not failure |
| ROAS | Breakeven = 1 ÷ gross margin; scale above roughly 1.3× that | Whether a dollar comes back as more than a dollar |
A first-month calendar
- Week 1: launch the proven slice plus one test cell. Do not touch bids. Install UTM tags so traffic sources stay separable in analytics from day one.
- Week 2: first read of CTR only. Kill obviously dead keywords from the search-terms report — never whole campaigns, which lose their learning and their auction history when paused.
- Week 3: read conversion rate. If the clicks convert, move the reserve into the winner; if they do not, rewrite the landing page before touching the ads again.
- Week 4: full ROAS read. Set next month's split from this month's data, and promote the test cell into the proven slice only if it earned it.
Cutting without killing the learning
Pause keywords and ad groups, not campaigns. A paused campaign keeps its history but stops bidding; a deleted one starts from zero next month, which on a small budget usually means the learning you paid for evaporates. The search-terms report is where the waste hides: exact-match negative the junk queries weekly, and the proven slice gets cheaper without losing volume.
Tests need enough budget to become readable. A $200 cell over two weeks buys roughly enough clicks for a first CTR and conversion-rate read on focused keywords. If the test slice cannot reach that on the audience you wanted, narrow the audience — splitting a thin test thinner just produces quieter noise.
The math check before you scale
Breakeven ROAS = 1 ÷ gross margin. At a 50% margin, breakeven is 2.0 — every advertising dollar must return two in revenue just to stand still. At a 30% margin it is 3.3. A workable scaling target is roughly 1.3 times breakeven, which leaves room for refunds, returns and your own time.
Worked through: a $60 product at 50% margin has breakeven ROAS of 2.0, so a $200 test cell needs about $400 of attributed revenue before it deserves a raise. The target ROAS calculator turns your margin into that number in a few seconds — run the check before moving money, not after.
Tools for this lesson
Split a monthly budget across campaign slices and read the daily spend for each.
Convert impressions and spend into CPM to compare auction costs month over month.
Quick weekly CTR reads from clicks and impressions.
Turn your gross margin into a breakeven and a scaling target before you move money.
Tag every campaign and ad so week-one traffic stays separable in analytics.
Frequently asked questions
- Is $1,000 a month enough to run Google Ads?
- Enough to learn, if concentrated: one proven campaign plus one controlled test produces readable data. Spread across six campaigns, the same budget produces noise. Most failures at this level are dilution, not auction prices.
- Should I use Smart Bidding with a small budget?
- Use it on the proven campaign, where real conversions feed the algorithm, and manual or maximise-clicks bidding on tests, where you need clean reads. Smart bidding on a cell with almost no conversions optimizes toward whatever noise it saw first.
- What ROAS should I aim for?
- Compute your own: breakeven is 1 divided by your gross margin, and a workable target is roughly 1.3 times that. A 'good' ROAS borrowed from another business — with their margins and their returns rate — is how profitable-looking campaigns quietly lose money.
- How long should I wait before judging a test?
- Judge CTR after about a week, conversion rate after two, and ROAS only at the end of a full month. Deciding at day three is reading noise; the calendar above exists because timing discipline beats tool tweaks.
Part of a public learning journal — general educational content, not professional advice. See our disclaimer.