Which ad actually drove the sale?
Which ad brought in this sale? It sounds simple, but the answer is never a single name — because customers see you many times before they buy. 🔗
A search result, social media, retargeting and a friend’s recommendation. Where they clicked last may not be what persuaded them.
Short answer: attribution is a point of view, not a certainty. The right approach is not trusting one number but looking from several angles. 🔍
Why the numbers never match
BU BÖLÜMÜN ÖZETİ
- Multiple touchpoints
- Device switching
- Measurement limits
- Time windows
Four reasons. 🧭
Multiple touchpoints
The customer meets several channels; each can claim the sale for itself. 🔁
Device switching
Someone who sees it on a phone and buys on a laptop can look like two different people. 📱
Measurement limits
As cookie and tracking limits grow, channels rely on estimates. 🔐
Time windows
Each channel decides how many days back it counts; the windows overlap. ⏳
Three points of view
BU BÖLÜMÜN ÖZETİ
- Last click
- First touch
- Distributed
Together they get closer to the truth. 👓
Last click
The touchpoint just before the sale. Simple, but it ignores what prepared it. 👆
First touch
The channel that found the customer first. It sees demand creation but hides what closed it. 🌱
Distributed
Credit is split across several touchpoints. More realistic, but it needs interpretation. ⚖️
Practical ways to verify
BU BÖLÜMÜN ÖZETİ
- “How did you hear about us?”
- A holdout test
- A total profit check
- Call tracking
Four methods, cheap and useful. 🔧
“How did you hear about us?”
One question in the order or form. Imperfect but surprisingly informative. ❓
A holdout test
Pause one channel for a while and watch total sales; the real contribution shows here. ⏸️
A total profit check
Not channel figures — is the business’s total profit rising? 💰
Call tracking
If phone sales are not counted, that channel always looks small. 📞
Four common mistakes
BU BÖLÜMÜN ÖZETİ
- Adding up channel figures
- Watching last click only
- Cutting the preparing channel
- Changing the measure often
All four move budget to the wrong place. 🚧
Adding up channel figures
Summing what each channel claims gives far more than reality. ➕
Watching last click only
Brand search always wins; but another channel created that search. 🏷️
Cutting the preparing channel
Switching off a channel that does not close sales directly can pull down the whole line the next month. ✂️
Changing the measure often
A different view every month blurs whether things are improving or declining. 🔄
A healthy measurement setup
BU BÖLÜMÜN ÖZETİ
- One source of truth
- Channel tagging
- Monthly reading
- Check against the whole
Three principles. 🛡️
One source of truth
Decisions use your own sales data; the platform is a direction indicator. 📊
Channel tagging
Links are tagged consistently; untagged traffic counts as lost. 🏷️
Monthly reading
Daily attribution is noise; decisions are monthly; reading sits in the success metric guide. 📅
Check against the whole
If a channel grows but the total does not, there is cannibalisation. ⚖️
What should I do today?
BU BÖLÜMÜN ÖZETİ
- Step 1: add the single question
- Step 2: fix your point of view
- Step 3: check the total
- If you want help
Three steps, one hour. 🪜
Step 1: add the single question
“How did you hear about us?” — in forms and checkout. Today. ❓
Step 2: fix your point of view
Which attribution view will you decide by? Write it down. 📐
Step 3: check the total
Not the channel — is the business total growing? 💰
If you want help
Let us set up your measurement and tagging: use the consult your expert form. For your current setup see the advertising digital audit; the whole sits on the advertising consulting page. 🎯
Related reading from the archive: unmeasured channels · conversion tracking.
📝 Notes From the Field
At one business, adding up channel reports produced a sales count far above the real number of orders. A channel that brought no direct sales was paused and total sales fell. That channel did not close sales, it started them. The holdout test showed in one go what no report could.
📖 Short Glossary
Attribution: the approach to which touchpoint gets credit for a sale. Last click: crediting the touchpoint just before the sale. Holdout test: pausing a channel to measure its total effect. Cannibalisation: one channel taking credit for another’s sales.
⚡ Quick Summary
Attribution is a point of view, not a certainty. 🔗 Numbers diverge because of multiple touchpoints, device switching, measurement limits and time windows. Last click, first touch and distributed views are read together. Channel figures are never summed; decisions rest on your own sales data and total profit.
🎯 Next Step
Let us set up your measurement and tagging: use the consult your expert form. What happens when ads stop sits in the dependency guide; for your setup see the advertising digital audit.
Frequently Asked Questions
Sık Sorulan Sorular
Use the same view consistently when deciding; switching it breaks comparison. 📐
There is no single right model; each shows a different side. What matters is using the same one consistently when deciding and checking it periodically with a holdout test.
It can be, which is why it is short, run on one channel and avoided in peak season. In return, it provides information no report can give.
Not on their own, but valuable as a direction indicator. They are especially good at revealing channels that are otherwise unmeasured, like phone and word of mouth.