Facebook Ads for a shoe workshop: 3,267 purchases, 464 million VND in conversion value

Background & challenge

The client is a shoe workshop selling online. The workshop has an edge in sourcing and cost of goods, but lacked a dedicated ad operator. LDH Media took over management of its Facebook Ads accounts and campaigns. The brand name is hidden by agreement.

The workshop’s challenge is a familiar one for footwear business owners:

• Many models, limited budget. It’s impossible to run ads evenly for every model.
• No clear view of where the ad money goes. Some campaigns bring orders; others just burn money.
• Optimizing for the wrong goal. Running for messages or clicks easily brings many inquiries but few buyers.

The goal: switch to optimizing for purchases, measure conversion value, and concentrate budget on campaigns with a low, stable cost per purchase.

Diagnosis from the data

The Ads Manager table (maximum time range, to 22/06/2025) lists 9 purchase-optimized campaigns. Reading column by column:

• Total purchases: 3,267. Total conversion value: 463,947,000 VND.
• Estimated spend about 114.3 million VND. This is the sum of purchases × cost per purchase for each campaign (calculated from dashboard data).
• Estimated account-wide ROAS about 4.1 (463.9 million ÷ 114.3 million).
• Average cost per purchase about 35,000 VND.
• Average value per purchase about 142,000 VND (463.9 million ÷ 3,267, calculated from the dashboard).

The table also has an impressions column. The 9 campaigns total about 2.34 million impressions. Campaign 1 achieved about 1.67 purchases per 1,000 impressions, campaign 2 about 1.43. Campaign 3 only about 0.47 (calculated from the dashboard).

Good metrics: the two core campaigns have a cost per purchase around 31,000–33,000 VND, with estimated ROAS of 4 or higher.

The bottleneck: the remaining 7 campaigns have higher costs per purchase, from about 38,000 to 113,000 VND. This group uses budget but contributes few purchases.

Growth strategy

LDH Media’s framework for workshops and manufacturers selling online through Facebook:

Phase 1 – Foundation: measure the purchase event correctly

1. Set up Pixel and the Conversions API, and check whether the purchase event records order value.
2. Standardize account structure: one objective per campaign, with clear names so reports are easy to read.
3. Pick a core group of shoe models to test first, instead of spreading thin.

Phase 2 – Activation: test widely, cut quickly

1. Run several purchase-optimized campaigns in parallel with different ad creatives.
2. Set an acceptable cost-per-purchase threshold based on profit margin.
3. Campaigns that exceed the threshold after a test period are cut or get refreshed creatives.

Phase 3 – Scale: concentrate budget on the winning campaigns

1. Raise budget step by step for campaigns with a low, stable cost per purchase.
2. Add new videos and images to the core campaign so the ads don’t “fatigue”.
3. Track cost per purchase daily and stop increasing budget when the metric keeps rising.

Results

Spend and ROAS are estimates: purchases × cost per purchase (≈ 114 million VND spend); conversion value ÷ spend ≈ 4.1. The main campaign achieved ROAS ≈ 4.5.

The Ads Manager dashboard for the 9 campaigns is shown at the top of the page. The table has no “vs. previous period” column because these are lifetime figures for the account.

Analysis: where the growth came from

For purchase-optimized ads, results can be split into: impressions × purchase rate per impression × value per purchase.

1. Impression volume is concentrated in campaign 2. This campaign had about 1.39 million impressions, nearly 60% of total impressions (calculated from the dashboard). It brought in 1,998 purchases, about 61% of total purchases.

2. Purchase rate determines cost. Campaigns 1 and 2 have a purchase rate per 1,000 impressions about 3x that of campaign 3. That’s why their cost per purchase is much lower.

3. Value per purchase is fairly even. Most campaigns have a value per purchase around 128,000–156,000 VND. Campaign 2 is about 149,000 VND (calculated from the dashboard). Differences in performance come mainly from cost, not order value.

Concentration is the key. Campaigns 1 and 2 generated 2,646 / 3,267 purchases, about 81%. They used about 76% of estimated spend (calculated from the dashboard). That means the rest of the budget (about 24%) brought in only about 19% of purchases.

The weak spots and how to fix them:

• Cost per purchase varies about 3.6x between the lowest campaign (31,789 VND) and the highest (113,266 VND). Campaign 3 has an estimated ROAS of only about 1.4. It should be cut early or have its creative replaced, not left to “learn” too long.
• Campaigns 5 and 8 have a cost per purchase above 54,000 VND and estimated ROAS below 3. Reduce their budget and shift the savings to the core campaigns.
• Dependence on one large campaign. When campaign 2 “fatigues”, the whole account’s results will drop. Always keep 1–2 backup campaigns testing new creatives.

Lessons for store owners

• Optimize for purchases, not clicks. Without conversion data (Pixel + Conversions API), Meta can only optimize for clicks.
• Calculate break-even ROAS before running. Know your profit margin to set a cost-per-purchase threshold.
• Read reports campaign by campaign. The account-wide average can hide a campaign that’s losing money.
• Cut weak campaigns fast, put the money into winners. Most results usually come from a small number of campaigns.
• Always have new creatives in testing. So that when the core campaign loses steam, a replacement is ready.

Actual results depend on the product, pricing, inventory, budget and platform policies. This is not a guarantee of results for every client.