Traffic source performance: Core Web Vitals by marketing channel

Acquisition shows how first visits land, by where people came from: Core Web Vitals, bounce and Gemini Score for each channel, campaign, landing page and ad click ID, and how much of a change came from the traffic mix rather than the site itself.

Acquisition how first visits land, by where people came from

“Which traffic source lands slowest, and did the site change or the mix?”

Sample data
Last 24 hrs
All platforms
shop.example.com
All visits
Compare: previous period
Metric · LCP

Landings: 412,860; ▲ 4% vs comparison
Paid share: 38.6 %; ▲ 2.8 pts · social, search, display
First-time visitors: 56 %; ▲ 0.8 pts
LCP p75 · all landings: 2.86 s; ▼ 108 ms vs comparison
Paid clicks, poor LCP: 6.9 %; ▼ 0.4 pts · LCP over 4 s
Explained by mix: −71 ms; 66% of the −108 ms change

Direct jumped from 24% to 52% of landings at 02:00 and stayed up: looks like a campaign start
LCP p75 fell 108 ms: −71 ms is traffic mix, −37 ms is the channels themselves
Paid search itself got faster: −24 ms of the blended change
First visits are 44% slower on LCP than returning visitors (3.38 s vs 2.35 s)

Traffic mix

landings per hour, by channel · dashed line = a share jumped and stayed up Landings · Share %

Paid social
Organic social
Paid search
Organic search
Other
Direct

20:0000:0004:0008:0012:0016:00

Why it changed

LCP p75 of all landings · previous period → now 66% from the mix

Previous period2.97 s
Traffic mix−71 ms

who arrived: Direct 31.0% → 28.6%, Paid search 31.2% → 33.0%
Within channels−37 ms

Paid search −24 ms · Organic search −8 ms · Direct +7 ms
Now2.86 s

Channel scorecard

p75 per channel · first visits only · channel recognised for 99.9% of landings

Channel	Landings	Share	vs comparison	First visit	LCP	INP	CLS	TTFB	Pass CWV	Bounce
● Paid search	136,240	33.0%	+1.7 pts	60%	● 2.08 s	● 211 ms	● 0.00	● 630 ms	76%	81%
● Direct	117,910	28.6%	−2.5 pts	55%	● 6.32 s	● 238 ms	● 0.29	● 960 ms	21%	58%
● Organic search	69,420	16.8%	−1.2 pts	54%	● 2.44 s	● 266 ms	● 0.07	● 752 ms	55%	63%
● Paid social	23,940	5.8%	+1.2 pts	59%	● 1.72 s	● 188 ms	● 0.00	● 578 ms	84%	95%
● Organic social	5,360	1.3%	−0.1 pts	57%	● 1.98 s	● 199 ms	● 0.00	● 560 ms	79%	89%
● Email & messaging	4,250	1.0%	+0.8 pts	68%	● 3.18 s	● 196 ms	● 0.00	● 780 ms	61%	76%
● AI assistants	3,960	1.0%	−0.1 pts	70%	● 2.28 s	● 183 ms	● 0.01	● 702 ms	72%	80%
● Other	51,780	12.5%	+0.2 pts	52%	● 1.05 s	● 420 ms	● 0.00	● 290 ms	82%	62%
All landings	412,860	100%		56%	● 2.86 s	● 244 ms	● 0.05	● 742 ms	57%	70%

LCP p75 by channel

if every channel holds steady while the dark line rises, the mix changed, not the site poor > 4 s · good ≤ 2.5 s

All landings
Direct
Organic search
Paid search
Paid social

20:0000:0004:0008:0012:0016:00

Top campaigns

from utm_campaign · top 8 of 450 campaigns

Campaign	Main channel	Landings	LCP p75	Poor	Pass CWV
brand-search-always-on	● Paid search	158,400	● 2.14 s	7%	76%
retargeting-dpa	● Paid social	26,300	● 1.63 s	4%	88%
shopping-feed-pla	● Paid search	8,540	● 1.38 s	7%	83%
generic-search-q4	● Paid search	6,810	● 1.71 s	7%	80%
share_product	● Organic social	6,290	● 2.10 s	7%	77%
newsletter-weekly	● Email & messaging	4,020	● 3.05 s	12%	58%
partner-offer	● Affiliate	3,260	● 2.66 s	11%	26%
app-install-push	● In-app	2,740	● 1.84 s	7%	73%

Top landing pages

where each channel sends people · top 7 of 46 pages

Page	Main channel	Landings	LCP p75	Poor	Pass CWV
shop.example.com · /product/:id	● Paid search 54%	302,100	● 2.10 s	8%	73%
shop.example.com · Home	● Direct 57%	107,800	● 3.66 s	22%	17%
shop.example.com · other pages	● In-app 87%	66,300	● 1.14 s	4%	80%
shop.example.com · /search	● Direct 95%	39,700	● 12.90 s	84%	6%
shop.example.com · /category/:slug	● Paid search 79%	29,900	● 1.88 s	9%	76%
m.example.com · /product/:id	● Direct 85%	14,600	● 3.18 s	15%	35%
m.example.com · Home	● Direct 92%	4,120	● 5.02 s	34%	2%

Click IDs and TTFB

time to first byte of landings, by the ad click ID in their URL

Click ID	Landings	TTFB p75	vs no click ID	LCP p75
Google Ads · gclid	131,200	● 588 ms	0.7×	2.18 s
Meta · fbclid	18,900	● 611 ms	0.7×	1.81 s
Google Ads · gad_source	1,120	● 630 ms	0.8×	3.40 s
Google Ads (iOS app) · gbraid	1,050	● 566 ms	0.7×	1.73 s
Microsoft Ads · msclkid	820	● 640 ms	0.8×	2.05 s
No click ID	365,400	● 820 ms	baseline	3.21 s

Landing bounce vs speed

paid landings that never reached a second page Paid · All

LCP ≤ 2.5 s85%

117,600 landings
LCP 2.5 – 4 s85%

14,100 landings
LCP > 4 s84%

10,200 landings · speed is not what sends these visitors away

Cold-start tax

first visit vs returning visitor · p75 · all visitors

Metric	First visit	Returning	Difference
LCP	3.38 s	2.35 s	+44% worse
FCP	2.34 s	1.66 s	+41% worse
CLS	0.04	0.08	48% better
INP	248 ms	241 ms	+3% worse
TBT	708 ms	395 ms	+79% worse
TTFB	742 ms	742 ms	same

Gemini score of landings

mean score 76.0 ▲ 0.9 · the points each input costs 76.0

TBT−8.8 pts
LCP−6.3 pts
Speed Index−3.2 pts
CLS−2.9 pts
FCP−2.4 pts

Sample data on example domains (shop.example.com). Your dashboard shows your own sites and apps.

What you can see on Acquisition

LCP, INP, CLS and TTFB for each channel, from first visits only
Traffic mix by hour, with campaign starts marked
How much of a change is traffic mix and how much is the channels themselves
Top campaigns (utm_campaign), landing pages and ad click IDs
First visit vs returning visitor speed: the cold-start tax

Questions about Acquisition

Which visits does Acquisition count?

Landings only: the first page of each visit, with tagged bots excluded. That is why its numbers differ from SEO perf, which covers every page view.

How are channels recognised?

From the referrer, UTM tags and ad click IDs such as gclid and fbclid. [Confirm the channel rules.]

Do ad click IDs slow down the first byte?

They can, when a CDN treats every click ID as a new cache key. The Click IDs table compares TTFB with and without one.

Last updated 6 Nov 2026 · Reviewed by [Name], [Role]

Related screens

Book a demo Start a free trial