Cost per Acquisition by Channel: Where Is Your CPA Gap Coming From?
Your blended cost per acquisition is above target. You know the number is bad. What you do not know is where the problem is coming from.
Related articles
TL;DR
- Cost per acquisition should be compared on the same customer definition, time window, order filters, and attribution framework before you compare channels.
- For acquisition analysis, New Customer CPA is more useful than total CPA because it focuses on first-time customers rather than mixing acquisition and repeat purchases.
- Google Ads CPA, Meta CPA, TikTok CPA, and Facebook Ads CPA can look very different inside their own platforms because the platforms do not necessarily assign conversion credit in the same way.
- A channel CPA benchmark adds a second question: is the channel unusually expensive compared with the same channel for stores in a similar AOV range?
- The biggest channel gap is a place to investigate first, not an automatic instruction to pause or move budget.
Is Meta CPA dragging the account? Is Google Ads CPA unusually high? Is TikTok CPA expensive because the channel is inefficient, or because it plays a different role in the customer journey?
This is where a single blended CPA stops being useful. It tells you that acquisition is expensive, but not which channel is creating the gap.
A better approach is to calculate New Customer CPA consistently across channels, compare each channel under the same measurement framework, and then add a peer benchmark. Attribuly's NC-CPA Benchmark is built around that sequence: overall New Customer CPA first, then channel-level comparison against stores in a similar AOV range.
Why Blended Cost per Acquisition Hides the Real Problem
A blended CPA averages the output of multiple channels into one number.
That is useful for a high-level view of paid acquisition, but it can hide the fact that each channel is performing very differently.
Imagine three channels. Google is close to a reasonable peer reference. Bing is slightly better than peers. Meta is dramatically more expensive. The blended CPA will tell you the account is costly, but only channel-level analysis will tell you where the relative difference is concentrated.
This matters because the corrective action depends on where the problem sits. A broad account-level response to a channel-specific problem can create more noise than clarity.
What Is CPA by Channel?
CPA by channel means calculating cost per acquisition separately for each advertising platform or paid channel instead of relying only on a blended account-level number.
For ecommerce, the more precise version is New Customer CPA by channel: attributed paid-media spend from a channel divided by attributed first-purchase conversions associated with that channel under a consistent attribution framework.
This keeps the business question focused on acquisition. Returning-customer purchases should not make a new-customer metric look more efficient.
Once that definition is consistent, you can compare Meta, Google, TikTok, Bing, and other supported channels more meaningfully.
Why Meta CPA and Google Ads CPA Often Do Not Match
A common search is “how to compare Meta and Google CPA.” The difficulty is that platform-reported numbers are not guaranteed to be calculated under one shared rulebook.
A shopper can see a Meta ad, later search on Google, return directly, and then purchase. Multiple platforms may have evidence that they influenced the same order. Depending on attribution windows, click versus view treatment, and conversion rules, the reported CPA can differ materially.
That does not make platform reporting useless. Meta Ads data is valuable for optimizing Meta campaigns, and Google Ads CPA is valuable inside Google. The problem appears when you treat self-reported platform CPA as if the numbers were directly comparable across ecosystems.
For cross-channel allocation, use a common conversion goal and a consistent attribution method.
New Customer CPA by Channel: Use One Definition
A fair channel comparison starts with a consistent definition of a new customer.
If Meta includes returning-customer purchases while Google is evaluated on first purchases, the two CPAs answer different questions. If one channel excludes refunds and another does not, the comparison becomes even weaker.
Attribuly's current NC-CPA methodology uses attributed paid-media spend and attributed first-purchase conversions, with canceled and fully refunded orders excluded from the new-customer benchmark calculation.
The benchmark uses a rolling 28-day period and applies the same attribution framework across the store result and peer comparison. The point is comparability, not simply producing another dashboard number.
What a Channel CPA Benchmark Adds
Even after you normalize channel CPA, you still only know how your own channels compare with each other.
Suppose your Meta CPA is $100 and your Google Ads CPA is $70. It is tempting to conclude that Google is better because the number is lower.
But a lower absolute CPA does not tell you whether either channel is strong or weak relative to comparable businesses.
A channel CPA benchmark adds peer context. Instead of asking only “which channel is cheaper for us?”, you can ask “which channel is furthest from the relevant peer reference for stores with similar AOV?”
That shift is important because different channels can have different normal cost levels and different roles in the buying journey.
A Practical Example: Find the Gap, Not Just the Cheapest Channel
Attribuly's current public NC-CPA Benchmark page uses an illustrative channel example that shows exactly why relative gaps matter.
In the example, Google Ads is close to its peer median, Meta Ads is far above its peer median, and Bing Ads is below its peer median. The overall acquisition result is expensive, but the largest relative gap is concentrated in Meta.
The useful conclusion is not “Google is cheaper, so move all spend to Google.” The useful conclusion is “Meta is the first place to investigate because its gap versus comparable stores is much larger.”
This is the difference between a raw channel ranking and a peer benchmark. One tells you which number is lower. The other tells you which difference is unusual.
How to Compare Meta and Google CPA Fairly
Start with the same conversion goal. If the business is trying to acquire new customers, evaluate both channels on first-time purchases.
Use the same measurement period. A 7-day Meta snapshot should not be compared with a 30-day Google result if you are trying to make a fair channel decision.
Use the same order filters. Canceled and fully refunded orders should be treated consistently.
Use a consistent attribution framework. This reduces the distortion that comes from comparing two platforms that each assign credit using their own logic.
Then compare the resulting New Customer CPA by channel with the relevant peer benchmark. That is where a Meta vs Google CPA comparison becomes a true diagnostic rather than a dashboard contest.
Compare Meta, Google, and other channels against peer benchmarks →
What If Meta CPA Is the Biggest Gap?
If Meta CPA is materially above the comparable-store reference while other channels are close to peers, Meta becomes the logical first place to investigate.
Possible investigation paths include campaign structure, prospecting versus retargeting mix, audience composition, creative fatigue, offer quality, landing-page conversion, and the role Meta plays earlier in the customer journey.
These are hypotheses, not conclusions. A high Meta CPA vs benchmark does not prove which one is responsible.
The benchmark narrows the search. Your campaign and customer-journey analysis must explain the cause.
What If Google Ads CPA Is the Biggest Gap?
The same logic applies when Google Ads CPA is the outlier.
You may want to inspect which campaigns or query groups are driving first-time customer cost, whether branded and non-branded demand are being mixed, whether landing pages are converting new visitors efficiently, and whether the channel looks different under a consistent attribution view than it does inside Google Ads.
Again, the benchmark tells you that Google differs from comparable stores. It does not automatically tell you why.
Where TikTok CPA and Other Channels Fit
Channel diagnosis should not stop at Meta vs Google CPA.
If your acquisition mix includes TikTok, Bing, or other supported channels, the same framework applies: define new customers consistently, calculate New Customer CPA, and compare the channel with an eligible peer reference.
A TikTok CPA that looks expensive in isolation may be normal for comparable stores, while a seemingly acceptable Google number may still sit well above its channel benchmark. Absolute cost and relative performance are different questions.
This is why a channel-level view becomes more useful as the media mix becomes more complex.
Do Not Turn the Benchmark Into a Budget Rule
A peer benchmark should guide investigation, not automate budget allocation.
Being above a peer median does not mean a channel should be paused. Being below the peer median does not mean the channel should automatically receive more money.
Budget decisions still depend on your own margins, scale, incrementality, customer quality, repeat purchase behavior, cash flow, and the way channels work together across the customer journey.
Attribuly's NC-CPA Benchmark is best used as a diagnostic layer: it helps answer where the meaningful gap is, so the team can investigate with more focus.
A Simple Workflow for Finding Your CPA Gap
- Calculate your overall New Customer CPA using a consistent first-purchase definition.
- Compare the overall result with stores in a similar AOV range.
- Identify whether your cost is near the peer median or materially above it.
- Break the result down into Meta CPA, Google Ads CPA, TikTok CPA, Bing CPA, and other eligible channels.
- Compare each channel with its relevant peer median or distribution.
- Find the biggest channel gap.
- Investigate that channel's campaigns, audience, creative, landing pages, and customer journey before changing budget.
This sequence turns “our CPA is bad” into a more useful statement: “our NC-CPA is high relative to comparable stores, and the biggest difference is concentrated in this channel.”
Conclusion
When blended CPA is high, the most useful question is not “which channel has the lowest number?”
It is “which channel is unusually expensive under a consistent new-customer measurement framework?”
Calculate New Customer CPA by channel, compare Meta, Google, TikTok, Bing, and other supported channels on the same rulebook, then add a peer benchmark based on similar AOV stores.
That turns a vague cost problem into a focused diagnostic: where is the gap, how large is it, and what should we investigate first?
Log in to see your biggest New Customer CPA channel gap → Explore the NC-CPA Benchmark →