The Full-Funnel KPI Library

A KPI is meaningless without its stage label. Cost per acquisition (CPA) on prospecting (advertising aimed at people who have not yet engaged with the brand) is a cost of discovery. CPA on branded search (searches that include the brand name) is a cost of harvesting demand that already existed. Same formula, different jobs, different verdicts. When the reporting layer collapses both into one CPA, the budget conversation collapses with it.

This library walks the four funnel stages in order: awareness, consideration, conversion, loyalty. Each stage lists the KPIs that belong to it, what each metric actually tells you at that stage, and what it hides. The principle that ties the whole thing together is the same one most reporting gets wrong: a funnel stage (where the customer is in the journey from awareness to loyalty) changes what a metric means, and dropping the label is how teams quietly optimize into a smaller business with prettier dashboards.

A KPI is meaningless without its stage label

Take ROAS. On a retargeting line item (advertising aimed at people who have already visited the site), a 6x ROAS often reflects demand the brand would have captured anyway. On a prospecting line item, a 1.5x ROAS may be the single most valuable spend in the account because it is recruiting new customers who would not have arrived otherwise. The dashboards will rank these in the wrong order every time. The number tells you nothing without the stage.

The same trap shows up with click-through rate (CTR), conversion rate (CVR), and average order value (AOV). High CTR on a cold-audience awareness ad is a signal that the creative earns attention. High CTR on a branded search ad mostly tells you the searcher already wanted the brand. Same metric, different reads. A KPI without its stage label is a number with no unit.

A KPI without its stage label is a number with no unit.

— QRY Customer Stage Framework

Awareness: what to measure and what each metric actually tells you

Awareness is the stage where the brand is being introduced to people who do not know it yet, or being reinforced with people who know it but have not bought. The job is mental availability and future demand, not this week's revenue. Judging an awareness campaign by last-click ROAS is the most common reporting error in paid media, and it is the one that quietly drains long-run growth.

The headline KPIs here are reach, frequency, share of voice (SOV), aided and unaided awareness, brand lift, and branded search lift.

Reach and frequency describe how many people saw the campaign and how many times. Reach without frequency is a count of impressions that may have hit the same small audience repeatedly. Frequency without reach is a number that says nothing about audience size. Effective reach (reach at a frequency high enough to register) is the version that maps to whether the campaign actually built memory.

Share of voice (SOV) is the brand's share of total category advertising weight over a window. It tells you how loud the brand is relative to competitors. It does not tell you whether the message landed. SOV is a positioning input, not a creative diagnostic.

Aided and unaided awareness come from survey work. Unaided awareness asks people to name brands in a category without prompting; aided awareness asks whether they recognize the brand when shown the name. Movement in unaided awareness is the harder read and the better signal that the brand is occupying real estate in the category.

Brand lift studies measure shifts in awareness, recall, or favorability between exposed and unexposed audiences during a campaign. They answer whether the spend moved perception, which is the actual job of an awareness buy.

Branded search lift is the cheapest, fastest awareness diagnostic most brands have access to. Branded query volume in geos that received the awareness campaign should rise against geos that did not. If it does not, the campaign reached people but did not register with them.

The failure pattern at this stage is judging awareness on conversion metrics. An awareness campaign optimized to last-click ROAS will collapse into a retargeting campaign within a week, because retargeting always wins last-click. The team then concludes awareness does not work, when what actually happened is they stopped measuring it.

Consideration: what to measure and what each metric actually tells you

Consideration is the stage where someone knows the brand exists and is evaluating whether to engage. They click an ad, land on the site, browse a category, add something to cart, abandon, come back. The job at this stage is to make the next step easier and more obvious, not to drive a same-session purchase.

The headline KPIs here are click-through rate (CTR), cost per click (CPC), engagement rate, add-to-cart rate, quality visit rate, and cost per session.

Click-through rate (CTR) and cost per click (CPC) describe whether an ad earns the click and what it costs when it does. The trap is treating cheap clicks as good clicks. A low CPC from a broad, mistargeted audience is the most expensive media in the account, because the clicks do not turn into anything downstream. Read CPC in pairs with what happens after the click.

Engagement rate (likes, saves, comments, video completions, scroll depth, depending on the surface) is a creative diagnostic. It tells you whether the ad held attention long enough to register. Engagement does not equal intent; it tells you the creative is doing its job at the consideration stage. Whether that engagement converts is a different question handled downstream.

Add-to-cart rate and quality visit rate are the on-site signals that map most cleanly to consideration. Quality visit rate (sessions that exceed a depth, duration, or engagement threshold the brand defines) filters out the bounce traffic that high CTR can mask. Add-to-cart rate is the closest leading indicator of conversion that still belongs to consideration, because the decision to buy has not been made yet.

Cost per session (paid media spend divided by sessions delivered) is the consideration-stage cost metric that survives audience changes better than CPC does. It rolls platform-level click-to-session leakage into the number, which keeps the comparison honest across channels.

The failure pattern at this stage is optimizing the cheapest click rather than the most useful one. Auction algorithms will always find a cheaper, lower-intent audience if the only signal they get is CTR or CPC. Pair the click-side metrics with on-site quality metrics from the first day of the campaign, or the channel will quietly walk itself down-funnel.

Conversion: what to measure and what each metric actually tells you

Conversion is the stage where the customer decides whether to buy. The job is to remove friction, present the offer cleanly, and close the loop on demand that the upper funnel created. This is where most paid media reporting lives, and it is where the most common metric, ROAS, does the most damage when read without context.

The headline KPIs here are average order value (AOV), customer acquisition cost (CAC), cost per acquisition (CPA), conversion rate (CVR), ROAS, incremental ROAS (iROAS), revenue, and new-customer rate.

ROAS is the platform-attributed efficiency signal, and it is the metric most likely to overstate paid media's actual contribution. Platform ROAS counts conversions the campaign was credited for, not conversions that would not have happened without the spend. On retargeting and branded search, that gap can be a factor of two or three.

Incremental ROAS (iROAS) is ROAS calculated only on revenue the spend actually caused, measured against a control. It is the version of ROAS that maps to a budget decision. The mechanics of measuring it (geo-lift, holdout tests, calibrated marketing mix modeling) are covered in the geo-lift testing explainer and the incrementality formulas reference.

Customer acquisition cost (CAC) and cost per acquisition (CPA) sit next to each other and get used interchangeably; they should not be. CPA is usually paid-channel and platform-attributed: campaign spend divided by attributed conversions. CAC is a business-wide number: total acquisition spend divided by new customers acquired, regardless of channel. CAC is the one finance cares about. CPA is the one the platform reports. They diverge most when paid spend is harvesting demand the brand would have captured organically.

Conversion rate (CVR) and average order value (AOV) are the on-site mechanics of the conversion stage. CVR tells you whether the offer, the page, and the audience match. AOV tells you what the average buyer is spending in a single transaction. Read together they answer whether the conversion stage is working; read separately, either one can be gamed (a CVR spike from running only on the highest-intent retargeting audience, an AOV spike from discounting that hurts contribution margin).

New-customer rate (share of orders that come from first-time buyers) is the conversion-stage metric that anchors paid media to growth rather than to harvesting. A channel with high ROAS and low new-customer rate is a finance optimization, not a growth one. The unit-economics math behind this is in the CAC and unit economics reference.

The failure pattern at this stage is steering the whole account by platform ROAS. Every channel will report a healthy ROAS because every channel is graded by the platform serving it. iROAS, contribution margin, and new-customer share are the three numbers that turn the conversion stage into a budget decision instead of a press release.

Loyalty: what to measure and what each metric actually tells you

Loyalty is the stage where existing customers either come back or do not. The job here is not new-customer acquisition; it is retention, repeat purchase, and the slow build of cohort value over time. Most brands underinvest in loyalty measurement because the wins are months away from the spend, and the dashboards reward whatever happened yesterday.

The headline KPIs here are cohort retention, lifetime value (LTV) and its variants, net promoter score (NPS), repeat purchase rate, repeat revenue share, recency frequency monetary (RFM) scoring, and subscriber rate.

Cohort retention is the most honest loyalty metric a brand has. It tracks a cohort (a group of customers tracked together over time, usually grouped by acquisition month) and reports what share of that cohort is still buying at month 3, 6, 12. Cohort curves do not lie the way blended retention numbers do, because they cannot be inflated by new customers showing up.

Lifetime value (LTV) is a family of metrics, not one number. Revenue LTV is total revenue per customer over a window. Gross-margin LTV strips out cost of goods. Contribution-margin LTV strips out variable costs (shipping, payment processing, returns) on top of that and is the one that matches the CAC the business pays. Quoting LTV without specifying which version is the single most common reporting sleight of hand in DTC.

Repeat purchase rate and repeat revenue share describe whether the brand is building a base or churning through one-time buyers. Repeat purchase rate is the share of customers who buy more than once. Repeat revenue share is the share of period revenue that comes from existing customers. The two move together but tell different stories: the first is about base size, the second is about base value.

Recency frequency monetary (RFM) scoring segments the customer base by how recently they bought, how often, and how much. It is a targeting and CRM tool more than a reporting one, but the segmentation it produces is the right lens for reading every other loyalty metric. Aggregate LTV across an unsegmented base hides which cohorts are actually carrying the business.

Net promoter score (NPS) and subscriber rate sit on either side of declared intent. NPS asks customers how likely they are to recommend the brand; it is a leading indicator of word-of-mouth and a lagging indicator of product and service quality. Subscriber rate (the share of customers on a recurring purchase) is a structural loyalty lever where category and product allow it. Both are useful, neither is a substitute for cohort retention.

The failure pattern at this stage is treating loyalty as a marketing report rather than a business outcome. If the brand's LTV and CAC do not show up in the same conversation, the paid media plan is running without a constraint that matters.

Cross-stage metrics that hold the funnel together

A small set of metrics belongs to no single stage and tells you whether the whole funnel is working. These are the ones to put on the executive dashboard.

Branded search growth is the single best lagging indicator of awareness work paying off. Branded query volume rising over a multi-quarter window is the funnel's tell that the upper stages are doing their job. Branded search flat or declining while paid spend rises is a sign the brand is harvesting more aggressively without creating more demand.

Contribution margin (revenue minus cost of goods minus variable costs) is the dollar figure paid media is actually buying. Reporting on contribution margin instead of revenue forces the conversation past top-line numbers that AOV mix or discounting can flatter.

Marketing efficiency ratio (MER) is total revenue divided by total marketing spend across all channels. It is the blended number that does not care which platform claims credit. MER is a coarser signal than iROAS but a more honest one across a portfolio, because it cannot be gamed by reallocating spend between channels that double-count each other's conversions.

New-customer share (the share of orders or revenue coming from first-time buyers) is the cross-stage check that the funnel is still recruiting, not just harvesting. A rising MER with a falling new-customer share is the classic late-cycle shape of a brand consuming its own demand base.

How to use these metrics together

Full-funnel measurement is an operating model, not a channel checklist. The practical move is to label every KPI with its stage before it ever reaches a dashboard, then read each stage on its own terms. Awareness gets brand lift and branded search; consideration gets quality visits and cost per session; conversion gets iROAS and new-customer rate; loyalty gets cohort retention and contribution-margin LTV. The cross-stage trio of iROAS, contribution margin, and new-customer share is the steering view that the executive team reads weekly.

ROAS without incrementality stays in the report as a diagnostic, not a steering metric. Awareness CPA and conversion CPA stay in different rows of the dashboard, not the same one. A cheap cost per thousand impressions (CPM) against the wrong audience is treated as the most expensive media in the account, because it is.

Where this most often breaks is at the seam between awareness and conversion. Teams that judge awareness on last-click ROAS will starve it, and the brand will run out of upper-funnel demand 6 to 12 months later. Teams that judge conversion on platform ROAS without iROAS will overstate the contribution of their lowest-funnel channels and underfund the work that grows the business. Stage labels are the cheapest discipline that prevents both.

If your reporting still rolls awareness and conversion into one CPA, or treats platform ROAS as the steering metric, the media plan is making unapproved tradeoffs between short-term harvesting and long-term growth. The fix starts with the labels.