The problem
A user clicks your Meta ad. The URL contains fbclid=abc123 — the only string connecting that click to a future conversion. By the time your pixel fires, the browser has already started destroying it. Safari strips the parameter via ITP. Cookies expire in 7 days. Consent banners delay script execution. Ad blockers prevent the pixel from loading entirely. 20-40% of click identities are lost before your tracking code runs a single line.
Why this works
- Timing is everything. The raw HTTP request still contains
fbclid,gclid, andttclidbefore the browser processes the page. An edge worker intercepts this request at the CDN layer — before HTML renders, before JavaScript loads, before any consent banner appears. - Server-set cookies survive ITP. Cookies set via
Set-CookieHTTP header from your own domain bypass Safari’s 7-day limit on JavaScript-set cookies. The identity persists for 90+ days, not 7. - First-click filtering fixes prospecting learning. PPC strategist John Moran’s thesis: when the algorithm learns from retargeting clicks, it gets good at finding people already in your funnel. Filtering to first clicks only forces it to learn from genuine cold-audience conversions.
Results
| Brand | What changed | Result |
|---|---|---|
| Smars | Edge-level click capture deployed | EMQ went from 3.2 to 8.7. Attributed conversions up 28%, same actual volume. |
| MNMLST | Ditched Shopify CAPI for server-side with edge capture | EMQ jumped immediately. Algorithm training on deterministic matches, not guesses. |
| Typical brand | Edge capture live for 2+ weeks | 20-40% improvement in attribution accuracy. More matched conversions, fewer modeled. |
Smars: “The transparency around channel and event level success rates is a huge plus. After deploying edge-level click capture, our Event Match Quality jumped from 3.2 to 8.7.” — Sahil Kanojiya, Head of Mediabuying
John Moran’s first-click thesis: Sending only first clicks to the algorithm is the single most impactful change you can make to new customer acquisition. When the algorithm receives retargeting conversions, it learns to find people already in your funnel. Filter to first clicks and it learns what a real new customer looks like.
How to set up your campaigns
On Meta
- Campaign type: Any campaign using CAPI (Advantage+, manual prospecting, retargeting).
- Conversion event: Use CAPI-only events from CustomerLabs. The edge-captured click ID is included in every server event, giving Meta deterministic matches instead of probabilistic modeling.
- Settings: Check Event Match Quality in Events Manager. Target EMQ of 8+. Edge capture typically gets you there in the first week.
On Google
- Campaign type: Performance Max, Search, or Shopping with enhanced conversions enabled.
- Bidding strategy: Any value-based or conversion-based bidding. Better attribution data improves all strategies.
- Conversion action: Use server-side conversion tags with the preserved
gclidfrom edge capture. Enable enhanced conversions for redundancy.
Setup with CustomerLabs
- Week 1: Edge worker deploys on your CDN (Cloudflare Workers, CloudFront Functions, or equivalent). No code changes to your website. Click capture starts immediately.
- Week 2: Identity matching goes live. Every visitor with a click ID has their identity preserved server-side and matched to CRM records and order history.
- Week 3: First-click filtering activates. Prospecting campaigns receive only first-touch attribution data. Retargeting events route to retention campaigns.
The bottom line
Your click identity starts dying the moment the browser takes over. Edge capture grabs it at the infrastructure level — before ITP, before consent banners, before ad blockers — and stores it where the browser cannot touch it. Combined with first-click filtering, you give the algorithm the cleanest acquisition signal possible.
“Helped take our Event Match Quality from 4.9 to over 9. The transparency of the platform lets you see channel and event-level success rates.”
Frequently asked questions
Does edge capture work with consent management platforms like OneTrust or Cookiebot?
Yes. Edge capture stores the click ID server-side before consent is collected. No cookies are placed on the user's browser until consent is granted. Once consent is given, the stored click ID is matched to the user's session. If consent is denied, the click ID is discarded. This is compliant with GDPR and ePrivacy — you're storing a URL parameter on your own server, not placing anything on the user's device.
Which CDN or hosting providers support this?
CustomerLabs deploys edge workers on Cloudflare, AWS CloudFront, Vercel, and Fastly. If your site runs behind any of these CDNs, edge capture works out of the box. For custom hosting setups, CustomerLabs provides a lightweight middleware that runs at the server level before page rendering.
How does this affect my attribution reporting?
Attribution accuracy improves because fewer click IDs are lost. Platforms receive more deterministic matches (click ID to conversion) instead of relying on probabilistic modeling. Your reported conversion counts in Meta and Google may increase by 10–30% — not because you have more conversions, but because more of them are now matchable.
Does this replace the need for server-side tracking?
No. Edge capture handles one specific problem: preserving click identity at the moment of arrival. You still need server-side tracking (Meta CAPI, Google Ads API) to send conversion events. Edge capture feeds into your server-side pipeline — it's the first step, not the whole system.
Ready to improve your signals?
Book a 30-minute walkthrough. We'll audit your current setup, show you what's missing, and map out a 3-week implementation plan.