Weekly Review: Week 2 vs Week 1

Decision for you: Hold the new paid-social spend at its current level for one more week. Don't scale it and don't cut it. Traffic rose 20% but orders were flat, and the data can't yet tell you whether the new paid visitors buy at all.

Headline

Net revenue fell 4.6% (15,200 → 14,500) even though sessions rose 20%. Ad spend rose 39%, so blended ROAS dropped from 4.90 to 3.37.

Key metrics

Decomposing revenue as sessions × conversion × AOV: sessions rose 20%, conversion fell 16% and AOV fell 5%. More traffic was cancelled out by weaker conversion and a lower AOV.

What changed and why

1. Paid social drove all the traffic growth (likely).

   • Paid social sessions went from 3,400 to 6,300 (+2,900), while every other source combined went from 9,000 to 8,600 (−4%).
   • Paid social is now 42% of sessions, up from 27%.
   • This is a mix effect. If paid-social visitors convert worse than search, email and direct, overall conversion falls even if no source changed.
   • I can't confirm this because I don't have orders by source.

2. The extra spend bought almost no extra orders (confirmed, with a caveat).

   • Spend rose by 1,200 and orders rose by 2.
   • Not all of the extra spend went to paid social, and the orders may have come from other sources, so treat the incremental CPA as a rough signal.
   • It points to the new campaign's visitors converting poorly or arriving too early in their buying journey.
   • One week is too short to judge, since some of these buyers may convert later.

3. Week 1 is not a clean baseline (confirmed).

   • The 10% email promo on Wednesday of week 1 probably lifted that week's orders.
   • Email sessions were also higher in week 1 (2,100 vs 1,900).
   • Part of the conversion drop may be the promo ending, not only the new traffic.
   • A discount would normally lower AOV, but AOV was higher in week 1, so I can't tie the AOV drop to the promo. Product mix or the new traffic are other possibilities (hypothesis).

Search (−8%) and email (−10%) sessions also dipped. That's modest and could be normal variation, but it's worth watching. Direct rose 12%, which may be a small halo from the paid campaign (hypothesis).

Problems and risks

• Ad efficiency: CPA is up 38% and ROAS is down 31% after one week of a new campaign. That's a flag, not yet a verdict.
• Returns: 21 vs 14 returned orders. These are small counts, so some of the rise could be chance. Two other explanations are worth checking:
  • Week 1 promo orders may be coming back as returns in week 2.
  • Returns may be concentrated in paid-social buyers or in a few products.
• Payment option: It launched Friday of week 2, so it covers only about 3 days. Don't credit or blame it yet. Check payment failures and checkout completion before and after.

Actions

Data notes

• Missing data: orders by source, funnel steps (add-to-cart, checkout start, completion), new vs returning customers, product mix, spend by channel, and whether net revenue already excludes returns. Without these, the explanations above stay at "likely" or "hypothesis".
• Sample size: the whole comparison rests on about 310 orders per week, so small differences can be noise. One week is not a trend, so I'd also look at the previous 3–4 weeks if you have them.
• Definitions: I used net revenue as you gave it, and I computed return rate against same-week orders, not against the original order cohort.
• Attribution: use one analytics source for session counts by channel, and check that its numbers match your ad platform's.

Suggested experiment

The open question is whether paid social brings real incremental buyers. Run the campaign for two more weeks while tracking orders by source. Then, if its conversion is still low, cut its budget by about half for one week and watch total orders. If total orders don't drop, the campaign isn't adding much. Decide the threshold beforehand, for example "a drop of fewer than 10 orders means not incremental", so you don't read the result after the fact.