In affiliate and direct response, KPIs are not vanity metrics on a monthly report. They are operational signals. Decisions to scale, pause traffic or rebuild a funnel are made daily, sometimes hourly, based on four numbers: EPC, CPA, AOV and ROAS.
This guide explains each one with a formula and a worked example, what moves it, and, often overlooked, how technical problems in tracking, CRM logic or infrastructure quietly distort it.
Good tracking and a well-configured CRM do not magically raise EPC or lower CPA on their own. What they do is make sure the numbers reflect reality: no lost conversions, no technical leakage, no duplicated revenue. Every optimization decision after that is only as good as the data underneath it.
EPC: Earnings per click
EPC measures how much revenue is generated for every click sent to an offer.
Why EPC is critical
- It is the primary metric affiliates use to decide whether to scale traffic.
- It tells a media buyer whether paid traffic can be profitable at current click costs.
- It is the common yardstick for comparing offers, funnels and traffic sources.
"If EPC drops, I stop sending traffic, immediately." That is how most affiliates think.
What moves EPC
- Page load speed: slow pages kill EPC
- Funnel structure, including upsells and downsells
- Offer pricing and checkout UX
- Payment acceptance rate
- Tracking accuracy: every lost conversion lowers reported EPC
Technical levers
- Faster page delivery through better hosting and a CDN
- Correct CRM upsell logic, so post-purchase revenue is actually captured
- Fixing broken postbacks and eliminating tracking gaps
- Removing friction from checkout flows, without touching the copy
CPA: Cost per acquisition
CPA is how much it costs to acquire one paying customer or lead.
Why CPA is mission-critical
- It determines whether a campaign can scale.
- It ties directly to profitability.
- Ad platforms optimize toward it, so the conversion signal you send them shapes it.
Indicative ranges by vertical
| Vertical | Typical CPA range |
|---|---|
| Nutra | $40–$120 |
| Skincare | $30–$90 |
| Info products | $10–$60 |
Treat these as rough orientation only. CPA varies heavily by GEO, traffic source, offer and season.
What moves CPA
- Funnel conversion rate and page speed
- Tracking latency and pixel firing accuracy
- Whether server-side tracking is implemented
- Payment declines
Technical levers
- Server-side tracking (conversions APIs and S2S postbacks) so platforms see every conversion
- Correct conversion events, with the right values and no duplicates
- Removing funnel friction and preventing checkout errors
- Improving payment success rates through decline handling and routing
Lower CPA often comes from better tracking accuracy, not just better ads. When an ad platform only sees 70% of your purchases, its algorithm optimizes on an incomplete picture and reported CPA looks inflated.
AOV: Average order value
AOV is the average revenue generated per customer transaction.
Why AOV matters
- A higher AOV lets you tolerate a higher CPA.
- It enables more aggressive media buying.
- It increases customer lifetime value (LTV).
What moves AOV
- Upsell and downsell strategy, especially one-click upsells
- Offer sequencing, bundles and order bumps
- Subscription pricing
Common AOV mistakes
Revenue that never happens
- Broken one-click tokens
- Payment failures on upsells
- Incorrect SKU mapping
Revenue nobody sees
- Upsells not tracked in the tracker
- Upsells not firing pixels or postbacks
- Upsell revenue merged into the wrong event
Technical levers
- Correct CRM campaign configuration for every upsell and downsell
- Reliable one-click token handling
- Dedicated upsell postbacks and pixel events
- QA testing of every accept and decline path
Most revenue in DR funnels comes after the first purchase. AOV lives in CRM logic.
ROAS: Return on ad spend
ROAS measures how much revenue is generated for every dollar spent on ads.
Why ROAS is the executive metric
- Founders and finance teams use it to judge the whole operation.
- It drives scale-or-pause decisions and budget allocation.
Rough benchmarks
| ROAS | What it usually means |
|---|---|
| Below 1.0x | Losing money on the front end |
| 1.5x – 2.0x | Break-even or thin margin, depending on costs |
| 2.5x and above | Often scalable, depending on backend LTV |
Subscription offers complicate the picture: a front-end ROAS below break-even can still be profitable once rebills are counted. Always know which revenue a ROAS figure includes.
What breaks ROAS visibility
- Missing conversions and delayed postbacks
- Pixel and CRM numbers that do not match
- Inconsistent revenue reporting across systems
- Ad platform attribution windows and modeling
Technical levers
- Accurate revenue tracking from the CRM, not estimated values
- Fixing postback delays and preventing lost or duplicated events
- Aligning CRM revenue with what ad platforms report
- BI dashboards (such as Looker Studio or Power BI) that reconcile all sources
Bad tracking makes ROAS look bad even when the campaign is profitable. Before cutting budget on a "losing" campaign, confirm the revenue numbers behind it.
Diagnosing a sudden KPI change
KPIs are interdependent, and a sudden change is often a symptom of a technical problem rather than a market shift. This is the first table we reach for:
| Symptom | Likely technical cause | Check first |
|---|---|---|
| EPC drops suddenly | Tracking or payment issue | Postback logs, gateway decline rates |
| CPA spikes | Pixel or event misfire | Event counts in the ad platform vs CRM orders |
| AOV flatlines | Upsells broken | Upsell take rate, one-click token errors |
| ROAS inconsistent | Attribution mismatch | Revenue values and deduplication across systems |
Key takeaways
- EPC, CPA, AOV and ROAS are operational signals that drive daily scale-or-pause decisions.
- Each KPI is shaped by page speed, funnel structure, payments and, critically, tracking accuracy.
- Benchmarks are only orientation; they vary heavily by GEO, source and offer.
- Technology does not improve KPIs by itself, but broken tracking or CRM logic guarantees they will not reflect reality.
- When a KPI moves suddenly, rule out technical causes before changing ads or offers.
