The E-commerce Paid Ads Playbook (With the Real Math)

Paid social and search will happily spend your money at any scale you allow. Whether that spend builds a business depends almost entirely on decisions made before the campaign goes live: the offer, the margin math and the measurement. This playbook covers all three, plus the day-to-day operating rhythm.

1. Start with the math, not the platform

Before touching an ad account, calculate what you can afford to pay for a customer. You need four numbers:

• AOV — average order value
• COGS % — cost of goods as a share of revenue
• Variable costs — payment fees, shipping, packaging, returns
• Repeat rate / 12-month value — how much a customer is worth beyond the first order

Worked example. AOV $80, COGS 38%, variable costs 12%, so contribution margin per order is $40. If you want to keep half of that as profit on first purchase, your allowable first-order CAC is $20 — a break-even ROAS of 4.0. If 30% of customers buy again within a year at the same margin, blended twelve-month value rises to about $52 and allowable CAC moves to roughly $32, or a ROAS target near 2.5.

2. Fix the offer before you fix the campaign

Advertising amplifies whatever exists. If the product page is unclear, the shipping is expensive and the reviews are thin, more traffic simply means faster money loss. Audit these before scaling spend:

• Is the value proposition legible in five seconds on a phone?
• Is total cost, including shipping, visible before checkout?
• Do you have at least a dozen recent reviews on the hero product?
• Is there a reason to buy today — bundle, threshold, guarantee — that is not just a discount?
• Does the landing page continue the ad's promise, or dump traffic on a generic homepage?

3. Account structure: simple, on purpose

Modern bidding systems learn faster with more data in fewer buckets. Fragmenting spend across dozens of ad sets starves each one and delays learning.

Meta

• One broad prospecting campaign, one to three ad sets, minimal targeting, six to eight creatives.
• One retargeting campaign with a genuinely different message — objections, proof, guarantees.
• Advantage+ style catalogue campaign for the product range, if the catalogue is clean.
• Exclusions kept minimal: over-exclusion is a more common problem than overlap.

Google

• Brand search, isolated, so it never contaminates non-brand performance reporting.
• Non-brand search on tightly themed groups with negative lists reviewed weekly.
• Shopping / Performance Max with a well-structured feed — the feed is the targeting.
• Do not let Performance Max cannibalise brand: exclude brand terms where the account allows it.

4. Creative is the targeting

Once audiences are broad, creative decides who sees the ad. Build a testing cadence rather than an occasional "creative refresh".

A workable monthly cadence

1. Week 1: ship four new concepts — different angles, not different colours.
2. Week 2: kill anything below half the account's average click-through with statistically meaningful impressions.
3. Week 3: produce three variations of the leading concept — new hook, new opening frame, new length.
4. Week 4: refresh the winners into the evergreen set and write next month's brief from what you learned.

Angles worth testing on almost any product: problem/solution, before-after, founder explanation, customer review read aloud, comparison with the obvious alternative, "how it is made", and unboxing. Each maps to a different buyer objection, which is why they behave differently.

5. Measurement that survives modern privacy

Platform-reported ROAS is a marketing claim made by the vendor being evaluated. Build a measurement stack that does not depend on it:

• Server-side conversions API with deduplication against the browser pixel.
• Consent mode implemented properly, so modelled conversions have a defensible basis.
• Blended MER — total revenue divided by total ad spend — as the top-line number.
• New-customer CAC tracked separately from blended CAC; growth comes from new customers.
• Post-purchase survey — "How did you hear about us?" — as a cheap, surprisingly stable signal.

When platform ROAS and blended MER disagree, trust MER. The bank balance does.

6. Scaling rules that avoid whiplash

Aggressive budget changes reset learning and produce the sawtooth performance pattern most accounts suffer from. Our rules:

• Increase budget by 20–30% at a time, no more than every third day, while MER holds.
• Decrease in similar increments; do not slam a campaign to zero because of one bad day.
• Judge changes on rolling seven-day windows, never on daily numbers.
• Scale horizontally — new creative concepts, new geographies, new product angles — before scaling vertically.
• Hold at least 15% of spend in a testing budget permanently, or you will run out of winners in a quarter.

7. Retention makes acquisition affordable

Every point of repeat rate raises the price you can afford to pay for a customer. The cheapest wins usually sit in email and SMS flows rather than in the ad account: welcome, browse abandonment, cart abandonment, post-purchase education, replenishment reminder, win-back. Get those running before you argue about bid strategies.

A useful framing: acquisition buys you the first order, retention decides whether that purchase was profitable.

8. The weekly scorecard

One page, same format every week, no screenshots:

9. Nine ways accounts quietly lose money

• Judging performance on daily data and reacting to noise.
• Over-segmenting audiences until every ad set is in permanent learning.
• Counting brand-search conversions as acquisition.
• Letting Performance Max eat branded traffic and calling it efficiency.
• Running discounts continuously until the full price becomes unbelievable.
• Testing landing pages and creatives simultaneously, so neither result is interpretable.
• Ignoring returns in the margin calculation, especially in apparel.
• Leaving broken UTM parameters in place for months.
• No creative pipeline — the single most common reason accounts plateau.

10. A ninety-day plan

1. Weeks 1–2: margin math, measurement rebuild, offer and landing page audit.
2. Weeks 3–4: rebuild account structure, launch six to eight creatives across three angles.
3. Weeks 5–8: weekly creative cadence, hold budgets steady, establish baseline MER.
4. Weeks 9–12: scale in increments, add retargeting with differentiated messaging, launch retention flows.

By day ninety you should be able to answer one question with evidence: for every dollar we put in, how many come back, and how quickly? Everything else in paid media is detail.