Case study - ICP Segmentation improves ad performance

Segmenting customers to make every ad dollar work harder.

An ad strategy for a B2B distributor (anonymized as "Store 1"). Store 1 had never segmented its customer base by Ideal-Customer-Profile (ICP) business type, so the first step was building that segmentation from scratch, then feeding it into Google and Meta to build lookalike audiences. Generic acquisition spend became precision-aimed campaigns, and the performance followed.

69%
meta roas, yoy
+29%
google roas, yoy
-13%
GOOGLE Cost/conversion, YOY
01
analyze
Channel performance
Benchmarked Google and Meta results YTD to find where ROAS and CAC were trending and why.
02
segment
Create ICP segments
Classified the existing customer base into distinct business-type segments matching the ideal customer profile — segmentation that didn't exist before
03
target
Test alongside existing campaigns
Ran the new ICP and lookalike audiences incrementally alongside the existing campaigns to see how they performed.
04
reallocate
Double down on what worked
Measured campaign-level lift and reallocated budget toward the segments and audiences outperforming the account average.

Modest budget increase. Sharper targeting.

Store 1's digital ads were already profitable across both Google and Meta. The opportunity wasn't just spending more — it was pointing spend at the customers most likely to convert.

That took building an ICP segmentation that didn't exist yet: going through the full existing customer base and classifying accounts into a handful of distinct business-type segments that matched the ideal customer profile.

Those segment lists became the seed audiences for lookalike targeting on Google and Meta , and the Conquest and ICP lookalike campaigns both outperformed the account average as a result. Store 1 is also currently dark on Microsoft Ads, where B2B desktop-oriented competitors are actively investing, a clear expansion opportunity for I recommended.

Quick Facts

• Client: B2B distributor
• Initiative: Improve ad performance
• Current ad channels: Google Ads, Meta (Facebook & Instagram)
‍• Segmentation: ICP business segments

ytd results - google

ROAS growth outpaced spend growth.

+29%
ROAS, YoY

Sales growth outpaced ad spend across the account.

-13%
CAC, YoY

Lower acquisition costs alongside ROAS gains. No tradeoff.

9x
Overall ROAS

A 9x return on ad spend, achieved while expanding campaign scale.

By clicks, CTR and impressions
clicks
▲ 2.7%
CTR
▲ 14.8%
impressions
▼ 10.5%
By conversion rate, ROAS and customer acquisition cost
Conversions
▲ 28.2%
Conv. Rate
▲ 13.4%
Ad Sales
▲ 43.6%
roas
▲ 29.2%
Cost/Conv
▼ 13.3%
By ad spend, avg. CPC and avg. CPM
Ad Spend
▲ 11.2%
avg. cpc
▲ 8.2%
avg. cpm
▲ 24.2%
ROAS
Existing (account avg)
0%
ICP + lookalike
+17%
Cost/Conversion
Existing (account avg)
0%
ICP + lookalike
-60%
ICP + lookalike targeting
+17%
ROAS VS AVG
-60%
CAC VS AVG
A curated list of Ideal-Customer-Profile accounts and their lookalikes outperformed the account average. The audience will keep growing as the customer list expands.
ytd results - meta

Shifting spend to ICP lookalikes paid off.

6x
Account ROAS on reallocated spend

By shifting budget away from cold subscriber growth toward the ideal-customer-profile customers and lookalikes.

+69%
ROAS, YoY

Sales growth outpaced ad spend across the account with more spend allocated to focus on the ICP + lookalikes.

9x
Overall ROAS

A 9x return on ad spend, achieved while expanding campaign scale.

Meta tracking wasn't fully consistent YOY, so this breakout is limited to the metrics that were reliably measured in both periods. It doesn't map one-to-one to the Google breakout above.
By clicks, CTR and impressions
clicks
▲ 10.9%
CTR
▲ 107%
impressions
▼ 46.4%
By conversion rate, ROAS and customer acquisition cost
purchases
▲ 123%
ad sales
▲ 118%
roas
▲ 69.3%
By ad spend, avg. CPC and cost/purchase
Ad Spend
▲ 28.9%
avg. cpc
▲ 16.3%
cost/purchase
▼ 42.2%  
ROAS
Existing (broad targeting)
-13%
ICP + lookalike
+74%
Below Average
Existing broad targeting campaign
The existing lookalike targeting campaign was continued as the control in A/B testing, and tracked slightly below the YTD average in the last month measured.
Well Above Average
ICP + lookalike targeting
The ICP-informed audience significantly outperformed the YTD average in the same period.
The segmentation work

Segmentation that didn't exist, built from the ground up.

Store 1 had no ICP segmentation in place. The base was classified into a handful of distinct business-type segments (specific verticals withheld here for confidentiality), and each segment's customer list became the seed for a Google and Meta lookalike audience.

Step 1
Audit the base

Went through Store 1's complete customer and prospect list and identified accounts matching the ideal customer profile by customer name or email domain.

Step 2
Define & tag segments

Grouped customers into distinct business-type segments matching the ideal customer profile.

Step 3
Build lookalikes

Fed each segment's customer list into Google and Meta to build lookalike audiences for prospecting.

The takeaway

Segmentation isn't just a retention tool, pointed at acquisition, it's a direct lever on ROAS and CAC.

Knowing who your best customers look like make both Google and Meta targeting sharper...without asking for a bigger budget.

I build ICP and lifecycle segmentation and turn it into media targeting, not just retention email. Want to improve ROAS and CAC on your campaigns?