The problem: channels that each look fine and a business that is not growing
Most companies that approach a performance marketing agency already run several channels. Google Ads reports a healthy ROAS, the affiliate program shows growing revenue, organic traffic is up year on year. Yet blended customer acquisition cost keeps rising and finance cannot reconcile the channel reports with what actually landed in the bank.
The cause is usually structural rather than tactical. Each channel is optimized in isolation against its own last-click metric, so three channels claim the same customer, brand search absorbs demand that content created, and affiliates are paid for orders that would have happened anyway. Nobody owns the question “where does the next dollar produce a new customer at the lowest cost?”
Performance marketing, done properly, is the discipline of answering that question continuously. It treats paid search, paid social, organic search, affiliate partnerships and conversion optimization as levers on one acquisition machine, and it measures the machine on unit economics rather than on the numbers each platform reports about itself.
What performance marketing means at Axoria
The term is often used to mean “any advertising you pay for on results.” We use a stricter definition: an acquisition program where budget allocation, creative, targeting and landing experience are all decided from one model of cost per incremental customer and the value that customer returns over time.
That definition has three consequences. First, channel mix is a decision, not an inheritance: budgets are rebalanced on evidence, including cutting a channel that looks good in its own dashboard. Second, measurement has to be built to see across channels, which means clean tracking and attribution, offline conversion data and periodic incrementality tests. Third, the funnel after the click matters as much as the click, because a 20% improvement in landing page conversion rate lowers CAC across every channel at once.
ROAS is a channel metric. MER is a business metric.
Return on ad spend tells you what one platform attributed to itself. Marketing efficiency ratio (total revenue divided by total marketing spend) tells you whether the whole program pays for itself. We report both and make budget decisions on the second.
Who this service is for
Performance marketing engagements suit companies that already have product-market fit, a working conversion path and enough monthly volume to measure. Typical fits include e-commerce brands spending across Google and Meta, SaaS companies with a trial or demo funnel, and lead-generation businesses in finance, real estate or healthcare where lead quality varies sharply by source.
It is a poor fit for pre-launch companies with no conversion data, or for businesses that want one channel managed in isolation. If you only need Google Ads run well, our Google Ads service is the better starting point.
What Axoria does
How channel mix decisions are made
The most common mistake in multi-channel acquisition is allocating budget in proportion to attributed revenue. Last-click attribution systematically over-credits channels close to purchase (brand search, retargeting, coupon affiliates) and under-credits channels that create demand (prospecting on paid social, content, upper-funnel video). Funding the former and starving the latter produces a program that looks efficient while the pool of new demand slowly drains.
We instead rank channels by estimated marginal CAC: what does the next unit of spend cost per new customer, once we adjust for cannibalization and known attribution bias? A retargeting campaign with a reported CAC well below prospecting can still be the worse investment if a holdout shows most of those buyers would have converted anyway. Marginal cost curves are steep in branded search and shallow in affiliate publisher recruitment, and the plan reflects that.
Paid search and shopping
Captures existing demand efficiently but saturates. We separate brand from non-brand, run Performance Max with tight guardrails, and treat impression share on high-intent terms as the ceiling.
Paid social
Creates demand and scales further than search, at the cost of noisier attribution. Creative testing velocity matters more than audience tinkering, and lift tests are essential before scaling.
Affiliate and partnerships
Pay-on-outcome economics with a CAC ceiling built in, but requires active fraud monitoring and commission rules that reward new-customer acquisition rather than coupon interception.
SEO and content
The slowest channel to start and the cheapest at maturity. Investment is judged on projected organic CAC over 12 to 24 months, not on next month’s numbers.
Conversion optimization
A multiplier rather than a channel. Every point of conversion rate gained reduces CAC across all traffic, which is why CRO sits inside the acquisition budget rather than beside it.
Retention and email
Not an acquisition channel, but it sets the LTV that decides how much acquisition you can afford. Repeat rate and second-order timing feed allowable CAC.
Attribution and incrementality: what we can and cannot know
No attribution model tells the truth. Data-driven attribution in GA4 is a reasonable default for comparing channels within the data Google can see, but it cannot see view-through effects, offline sales, or the customers who would have bought regardless. Platform attribution (Google Ads, Meta) uses wider windows and claims view-through conversions, so summing platform-reported revenue routinely exceeds actual revenue.
Our approach is to use attribution for relative comparison and incrementality testing for absolute decisions. Geo-split tests (spend on in some regions, off in others) work well for brands with national distribution and enough volume per region. Conversion-lift studies inside Meta or Google are useful when geo tests are impractical. For affiliates, we compare new-customer rates between referred and non-referred cohorts to spot publishers intercepting rather than creating demand. Where a clean test is impossible, we say so and decide with stated uncertainty rather than false precision.
Unit economics we manage against
| Metric | What it tells you | How we use it |
|---|---|---|
| Blended CAC | Total acquisition spend divided by new customers, all channels | The headline efficiency number; trended monthly and by cohort |
| Marginal CAC | Cost of the next customer from an extra unit of spend in a channel | Decides where incremental budget goes and where to cut |
| LTV and payback period | Gross-margin value of a customer over time; months to recover CAC | Sets the allowable CAC ceiling per segment and product |
| MER | Total revenue divided by total marketing spend | Cross-checks platform ROAS against reality |
| New vs returning revenue | Share of revenue from first-time buyers | Stops retargeting and coupon affiliates from inflating results |
| Contribution margin after CAC | Revenue minus COGS, fulfilment and acquisition cost | The number finance actually cares about; guides scaling decisions |
Which metric leads depends on the business model. A subscription SaaS company with a long payback period cares about cohort LTV and churn far more than first-month ROAS. A high-margin e-commerce brand with low repeat rates should manage almost entirely on first-order contribution margin. We agree the primary metric with you before spend starts, then hold every channel to it.
Our process
Audit and economics
We review every active channel, the tracking behind it, and the real financials. The output is a baseline: blended CAC, channel-level marginal cost estimates, payback, and the measurement gaps that would make decisions unreliable.
Measurement fixes
Before reallocating budget we fix what would corrupt the data: broken conversion events, missing offline conversion import, absent new-customer flags, duplicated purchase events. This step frequently changes the picture of which channels work.
Allocation plan
A quarterly channel mix with target CAC by channel, guardrails for scaling and cutting, and a small holdout reserve for testing new channels or tactics.
Execution and testing
Weekly management of accounts and partners against the plan, with a running experiment backlog. Creative, offer and landing page tests run with defined sample sizes and stopping rules.
Incrementality checks
Scheduled lift or geo tests on the channels carrying the most budget or the most attribution doubt. Results feed back into the marginal CAC model.
Scale or reallocate
Monthly review against unit economics. Channels that hold marginal CAC under target get more budget in controlled increments; channels that do not get restructured or cut.
Scaling rules
Scaling is where most programs break. Doubling a Meta budget overnight resets learning phases and pushes into lower-intent audiences; doubling search budget on already-saturated terms mostly buys higher CPCs. We scale in steps, typically 15 to 25 percent at a time, and hold for long enough to see whether marginal CAC stays within tolerance before the next step.
We also define the stop conditions up front. If marginal CAC exceeds allowable CAC for a defined window, or if new-customer share falls below a threshold, spend is pulled back rather than defended. Written rules take the emotion out of budget conversations and give finance a clear answer to “why did spend go up?”
Tools we typically work with
Google Ads, Meta Ads Manager, Microsoft Advertising and the major affiliate platforms (Impact, PartnerStack, Awin, CJ) for execution. GA4 and Google Tag Manager for measurement, with server-side tagging where consent or ad blockers degrade browser-side data. Looker Studio or your BI tool for reporting, and Search Console, Semrush or Ahrefs on the organic side. We work inside your accounts; you keep ownership.
How results are measured and reported
Reporting is monthly, with a weekly summary during active scaling. The monthly report reconciles platform-attributed revenue to actual revenue or closed pipeline, shows blended and channel CAC against target, tracks payback and new-customer share, and lists the experiments run, what they showed and what changed as a result. If a number is an estimate, it is labelled as one.
We do not report vanity metrics as outcomes. Impressions, clicks and platform ROAS appear as diagnostics, not headlines. The headline is whether the program acquired customers at an acceptable cost, and what changes next month.
What to expect and common challenges
The first quarter often looks worse before it looks better
Fixing tracking removes double-counted conversions and inflated ROAS, so reported numbers may fall while real performance improves. We prepare stakeholders for this in advance.
Attribution debates need a referee
When sales, brand and performance teams each claim the same revenue, the acquisition model becomes the shared source of truth. Finance sign-off early avoids months of argument.
Incrementality tests cost something
A geo holdout means deliberately not spending in some markets for a few weeks. The short-term revenue cost is real; the payoff is confidence in a far larger budget decision.
What we will not do
We will not scale a channel on platform-reported ROAS alone, pay affiliates for intercepted brand traffic, or promise a CAC or revenue figure before we have seen the data. Forecasts are ranges with assumptions attached.