Abandoned Cart Recovery for WooCommerce: What Actually Works (With Numbers)

One Bit  ·  April 2026  ·  8 min read

Most abandoned cart guides tell you to send three emails with increasing discount levels. That advice is 10 years old and it does not work as well as it used to. Browser privacy changes have reduced email open rates, and customers have learned to abandon carts specifically to wait for the discount email.

We took a different approach. Instead of leading with discounts, we lead with information. And instead of only emailing, we use WhatsApp for high-value carts. Here is what our numbers look like after 12 months of running this system on zamanistore.com.

The Sequence That Works

We use a 3-step sequence. No discount in steps 1 or 2. Discount only in step 3, and only for carts above a threshold value.

Total: 14.1% across all three steps. After step 3, we stop. Sending more messages hurts your sender reputation more than it helps.

Step 1: The 4-Hour Email

The biggest mistake in abandoned cart emails is sending too early or too late. Under one hour feels pushy and catches people who were still shopping. Over 24 hours and they have forgotten about it. Four hours is the sweet spot — long enough that the purchase feels considered, short enough that the product is still in mind.

The first email has one job: remind them what they left behind. No discount language. No “you forgot something!” subject line. The subject line that works best for us is simply the product name.

Content structure:

1. Product name and image
2. One sentence about the product (not marketing copy — something specific)
3. Cart total and a single clear CTA: “Complete your order”
4. Optional: 1-2 recent reviews for social proof

Step 2: The 24-Hour Social Proof Email

By the 24-hour mark, a customer who has not purchased has likely seen the product, thought about it, and moved on. You need a reason for them to reconsider that is not a discount.

Two angles work well:

Low stock warning: If the product genuinely has under 10 units, include this. “Only 4 left in stock” is not a dark pattern if it is true. We only show stock counts when they are real and below 10.

Social proof: Include 1-2 reviews from customers in the same country as the abandoner, if you have them. A review from someone in Riyadh lands differently than a generic five-star rating when you are targeting Saudi Arabia.

Step 3: WhatsApp for High-Value Carts

This is the part most WooCommerce stores skip, and it is where we get a disproportionate amount of our recovery.

For carts above 150 AED (about $41), after 48 hours of no purchase, we send a WhatsApp message. We use Twilio for delivery. The message is short, personal in tone, and includes a one-time 10% discount code.

Why WhatsApp works better than a third email: In the UAE, Saudi Arabia, and most of the Middle East, WhatsApp is the primary communication channel. Open rates are near 100%. A WhatsApp message from a brand feels more personal and less filterable than an email.

The Technical Setup (n8n)

We build this in n8n rather than using a plugin. Here is the architecture:

1. WooCommerce webhook on woocommerce_checkout_order_created fires when a checkout is initiated but not completed.
2. n8n stores the cart in Postgres with a timestamp and cart value.
3. Scheduled triggers at T+4h and T+24h check Postgres for carts that have not converted and send the respective emails via Gmail API.
4. T+48h trigger checks if cart value exceeds threshold and sends WhatsApp via Twilio if so.
5. When an order completes, a woocommerce_order_status_completed webhook marks the Postgres record as converted so no further messages are sent.

Total workflow: 4 n8n workflows (WF49a-d) handling approximately 200-400 abandoned carts per month across both stores. Zero manual intervention required after initial setup.

What Does Not Work

• Leading with a discount: Trains customers to abandon deliberately to get 10% off. We tested this — conversion rate goes up, average order value goes down, and repeat discount-hunters increase.
• Four or five emails: After step 3, you are just burning your unsubscribe budget. Open rates on emails 4-5 drop below 5% and unsubscribe rates spike.
• Generic subject lines: “You left something behind!” and “Your cart misses you” both underperform product-specific subject lines in every A/B test we have run.
• Sending to all abandoned carts equally: A cart abandoned at step 1 of checkout (viewing cart) is different from one abandoned at the payment page. We only trigger the sequence for carts that reached the checkout form — intent is higher.

How to Set Up Facebook CAPI for WooCommerce and Shopify (2026)

One Bit  ·  April 2026  ·  10 min read

iOS 14.5 shipped in April 2021 and changed e-commerce advertising permanently. Before it, the Meta Pixel tracked almost every purchase. After it, browser-side tracking dropped 30-40% on iOS devices — and that number has only grown as Safari, Firefox, and ad blockers have tightened further.

The Meta Conversions API solves this by sending events from your server instead of the visitor’s browser. The server does not care about iOS privacy settings or ad blockers. It sends the Purchase event directly to Meta with hashed customer data, and Meta matches it to an ad click.

We run CAPI across both zamani and Coloring Palestine. Here is exactly how we set it up — and the deduplication mistake most guides do not mention.

Step 1: Generate a Conversions API Access Token

Go to Meta Business Suite, then Events Manager. Select the pixel attached to your store. Click the Settings tab. Scroll down to the “Conversions API” section and click Generate Access Token.

Copy this token and store it somewhere secure — you will need it in the next step. Do not put it in your theme’s functions.php or anywhere that gets committed to version control.

Step 2: Install Server-Side Tracking on Your Store

WooCommerce

You have two paths:

Option A — Plugin (easiest): PixelYourSite Pro has native CAPI support. Install it, paste your access token, and it handles Purchase, AddToCart, and InitiateCheckout server-side events automatically. It also handles event deduplication using its own event ID system.

Option B — Custom n8n workflow (full control): This is what we use at One Bit. A WooCommerce webhook fires on woocommerce_payment_complete, sends order data to n8n, and n8n forwards a hashed CAPI Purchase event to Meta’s Graph API. More setup work, but you control exactly what data is sent and can add custom parameters.

The Graph API endpoint for CAPI events is:

POST https://graph.facebook.com/v21.0/{PIXEL_ID}/events
Content-Type: application/json

{
  "data": [{
    "event_name": "Purchase",
    "event_time": 1711929600,
    "event_id": "unique-uuid-here",
    "action_source": "website",
    "user_data": {
      "em": ["sha256_hashed_email"],
      "ph": ["sha256_hashed_phone"],
      "fn": ["sha256_hashed_first_name"],
      "ln": ["sha256_hashed_last_name"],
      "country": ["ae"],
      "zp": ["00000"]
    },
    "custom_data": {
      "currency": "AED",
      "value": 149.00,
      "content_ids": ["product-sku-123"],
      "content_type": "product"
    }
  }],
  "access_token": "YOUR_TOKEN_HERE"
}

Shopify

Shopify has a native CAPI integration. Go to Settings > Customer events, click Add connection, and select Facebook. Choose your pixel. Enable “Send server events” and paste your access token.

Shopify’s native integration covers the main purchase funnel: PageView, ViewContent, AddToCart, InitiateCheckout, Purchase. It handles hashing automatically.

Step 3: Implement Event Deduplication

This is the step most tutorials skip, and it causes real problems if you miss it.

When you have both a browser pixel and CAPI running, every Purchase event fires twice — once in the browser, once from the server. Meta sees two Purchase events and counts them as two conversions. Your reported ROAS doubles. Your campaign data becomes useless.

The fix is a unique event_id. Here is how it works:

1. When a customer lands on your order confirmation page, generate a UUID: crypto.randomUUID() in JavaScript works fine.
2. Fire the browser pixel with that UUID: fbq('track', 'Purchase', purchaseData, {eventID: 'your-uuid-here'});
3. Send the same UUID in your server-side CAPI call in the event_id field.
4. Meta receives both events with the same event_id and counts them as one.

For WooCommerce with PixelYourSite: it handles this automatically. For custom n8n setups: generate the UUID on the thank-you page (JavaScript), store it in a session variable or pass it via a WooCommerce order meta field, then include it in the n8n CAPI call when the woocommerce_payment_complete hook fires.

Step 4: Hash Customer Data Correctly

CAPI matches events to Meta users using hashed PII. The matching happens on Meta’s side — you never send plain-text personal data. The fields Meta uses:

• em — email (SHA-256, lowercase, trimmed)
• ph — phone (SHA-256, E.164 format: +971501234567)
• fn — first name (SHA-256, lowercase)
• ln — last name (SHA-256, lowercase)
• country — 2-letter ISO code (SHA-256, lowercase)
• ct — city (SHA-256, lowercase, no spaces)
• zp — zip/postal code (SHA-256)

In PHP:

$hashed_email = hash('sha256', strtolower(trim($customer_email)));
$hashed_phone = hash('sha256', preg_replace('/[^0-9+]/', '', $phone));

Step 5: Test and Verify in Events Manager

Before going live, use Meta’s test tool:

1. In Events Manager, open your pixel and click Test Events.
2. Copy the test event code (looks like TEST12345).
3. Pass this code as test_event_code in your CAPI API call (remove it before going to production).
4. Run a test purchase on your store.
5. Confirm the event appears in Events Manager with source “Server” alongside the browser pixel event with source “Browser”.
6. Verify both show the same event_id — this confirms deduplication is working.

After you go live, check your Event Match Quality score (shown in Events Manager for each event type). A score below 6 means Meta cannot reliably match the event to a user — your CAPI data is underperforming. Common causes: missing phone field, not sending country, or email not trimmed/lowercased before hashing.

Common Mistakes We See

• No deduplication: Running browser pixel and CAPI without event IDs — ROAS appears doubled.
• Wrong pixel ID: CAPI events land in a different dataset from browser pixel events — no deduplication possible.
• Phone number format: Sending 0501234567 instead of +971501234567 — E.164 format is required for phone matching.
• Test event code left in production: All production events land in test mode and are not counted in campaigns.
• Missing currency: Always include currency and value in custom_data for Purchase events — without them, Meta cannot optimize for purchase value.