Pulsar · Metrics playbook · No. 05 · Attribution. Real source data from the test pixel, 30-day window (currency EGP).
Who Really Made the Sale?
The busiest ad is rarely the one that earns — and most of your sales may be hiding in untagged traffic.
Companion to No. 01 — Efficiency & Fatigue, No. 02 — Quantity vs. Quality, No. 03 — Garbage In, Lookalike Out and No. 04 — The Invisible Sales.
Contents
01 The budget question · 02 Volume is not value · 03 The dark sales · 04 Why sales go dark · 05 The ladder · 06 Last-click lies · 07 The fix
The question that decides your budget
Every scaling decision rests on one question: which ad actually MADE the sale? Get it wrong and you pour money into a channel that only looked busy, and starve the one quietly earning. Clicks and sessions are easy to see; the ad that earned the money is often hidden.
The ad with the most clicks and the ad that earns the money are almost never the same ad.
Volume is not value — real data
Raw sessions flatter the wrong campaign. Compare the test pixel’s two real campaigns over the same 30 days — then flip the metric and watch the winner change.
Two campaigns, two stories — pick your metric
Same two campaigns, same 30 days. “By sessions” crowns the big one; “by CVR / value-per-session” crowns the small one. The metric you pick decides who gets your budget.
| Campaign | Sessions | Buyers | CVR | Value | Verdict |
|---|---|---|---|---|---|
| Campaign …850455 | 2,864 | 2 | 0.1% | 1,410 EGP | busy, not efficient |
| Campaign …610455 | 256 | 1 | 0.4% | 490 EGP | efficiency winner |
The rule: judge a source by CVR and value-per-session, never by raw clicks or sessions. Volume tells you a campaign is loud; only conversion tells you it earns.
The chart is interactive in the HTML version: a “by sessions” state shows campaign …850455 (2,864 sessions) dwarfing …610455 (256 sessions); flipping to “by CVR / value-per-session” reverses the picture — …610455 converts at 0.4% and earns about 1.91 EGP per session, against 0.1% and about 0.49 EGP per session for the big campaign.
The dark-sales problem — real data
Of 11 total buyers this month, 7 came from UNTAGGED traffic — no campaign tag, no way to say which ad earned them. That’s 64% of your sales you cannot attribute.
Buyers by source — 30 days, 11 buyers total
The red bar is the problem: sales with no campaign tag at all. Hover a bar for that source’s sessions and CVR.
The chart is interactive in the HTML version: hovering each bar shows the source’s sessions and conversion rate — untagged 1,253 sessions / 0.6% CVR; campaign …850455 2,864 sessions / 0.1% CVR; campaign …610455 256 sessions / 0.4% CVR; other tags 1 buyer.
Why sales go dark
Four reasons credit lands in the wrong bucket — or no bucket at all. All four are live on real accounts; one of them is measurably live on this pixel.
Reason 1
Not tagged
no UTMs, no click ID
URLs shipped without UTM tags or click IDs collapse into Direct / untagged. The ad did its job; the link never carried the receipt.
Reason 2
Broken macros
real example from this pixel
13 sessions arrived tagged with the literal text “{{campaign.id}}” — the URL macro never expanded. The tracking template is misconfigured, so the data is garbage.
Reason 3
Lost click IDs
stripped in transit
Redirect chains, link shorteners and in-app browsers strip the fbclid — so a paid click lands on your site looking organic.
Reason 4
Cross-device / slow buyers
the purchase changes machines
Clicked on a phone, bought on a laptop days later — the purchase session has no memory of the ad, so it books as Direct.
Dark sales aren’t mysterious — they’re plumbing failures: a missing tag, a macro that never expanded, a stripped click ID, a changed device.
How the source is REALLY decided — the ladder
UTM tags alone miss a lot. Pulsar classifies every session by a priority ladder: a paid click ID (fbclid / gclid / ttclid) wins first → then UTM tags → then the referring domain → then Direct as the last resort. So a genuine Meta click is still caught by its fbclid even when the UTMs are missing or broken.
The priority ladder — click a rung
Each rung claims what it can prove and passes the rest down. Only what nothing can claim ends up Direct.
Where the ladder puts your actual customers
Among actual customers on this pixel, the ladder attributes 57% to Meta, 29% to Referral and 14% to Meta (organic) — attribution the UTM-only view would have lost.
The ladder is interactive in the HTML version: clicking a rung (Click ID → UTM → Referrer → Direct) shows what that rung catches and what falls through to the next. In print, every rung is shown open above, and the customer-source split — Meta 57% / Referral 29% / Meta (organic) 14% — is rendered as a static bar.
Last-click lies
Last-click attribution hands 100% of the credit to the FINAL touch and erases the ad that INTRODUCED the customer.
The money leak: defunding the top-of-funnel discovery that feeds every later sale. Kill the introducer and the “Direct” sales it was quietly seeding dry up weeks later — with nothing in the report to explain why.
The closer gets the credit; the ad that opened the door gets cut.
The fix — five steps, in order
Tag it, protect it, judge it by value, read it by the ladder — and never let the last click write the whole story.
TAG every ad URL with consistent UTMs
…and VERIFY the macro actually expands — no literal “{{campaign.id}}” in your data. A tracking template you never tested is a tracking template that’s lying.
PROTECT the click IDs
Avoid redirect chains and shorteners that strip the fbclid. The click ID is the strongest evidence a session came from a paid ad — don’t launder it away in transit.
JUDGE by CVR and value-per-session
Not raw clicks or sessions. Section 02 is the proof: the campaign with 11× the traffic converted at a fifth of the rate. Volume flatters; value decides.
READ the source by the ladder
Click ID → UTM → referrer — not UTM alone. A genuine paid click with broken UTMs is still a paid click; the fbclid proves it when the tags can’t.
DON’T trust last-click for discovery channels
Give upper-funnel ads credit for the sales they start. The introducer’s value shows up days later under someone else’s name — budget accordingly.
Where Pulsar fits: it classifies every session by the click-ID → UTM → referrer ladder, counts BUYERS and VALUE per source (not just clicks), surfaces untagged and broken-macro traffic, and counts real people — so you can see who actually earned the money.
Cross-reference — see No. 02 — Quantity vs. Quality for the funnel behind each source, and No. 01 — Efficiency & Fatigue for CPA per real source.