by Om RathodAsk three people on a marketing team what "ROAS" means and you'll get three different answers, all confidently stated, all technically defensible. That's the problem with DTC analytics vocabulary right now. It gets borrowed from Shopify, bent by Meta's ad reporting, redefined again in GA4, and by the time it lands in a board deck, nobody's actually comparing the same thing anymore.
This is exactly why brands need a real DTC ecommerce analytics terminology guide, not just a glossary tucked into a platform's help docs.
Here's a concrete case. Meta Ads Manager reports ROAS based on its own attribution window, often 7-day click or 1-day view, counting only what it thinks it caused. A blended MER calculation, by contrast, divides total revenue by total spend across every channel, no attribution model involved. Someone pulls a "3.5 ROAS" from Meta, someone else pulls a "1.8 MER" from a spreadsheet, and suddenly the team is arguing about budget with two numbers that were never meant to be compared.
This guide groups terms into five buckets: attribution and measurement, revenue and profitability, ad performance, customer and retention metrics, and data infrastructure. Jump to whichever section solves today's argument.
Last-click attribution gives 100% of the credit for a sale to the final touchpoint before checkout. It's simple and it's wrong more often than it's right, especially for brands running multiple channels at once.
Multi-touch attribution (MTA) spreads credit across several touchpoints in the customer journey, using weighted models instead of a single "last click wins" rule. It's more honest, but it depends on tracking data that's gotten a lot leakier since 2021.
Media mix modeling (MMM) skips individual user tracking entirely. It uses aggregate spend and sales data over time to estimate each channel's contribution statistically. Slower to build, but it doesn't break when a platform changes its tracking policy.
Incrementality measures whether an ad actually caused a sale that wouldn't have happened anyway. Incrementality testing (holdout groups, geo tests) is how you find out. This matters more than platform-reported ROAS because platforms are graded on their own homework: they'll always claim credit for revenue that might've happened organically.
First-party data is information you collect directly, your own site events, your own email list, your own purchase records, rather than data bought or tracked through someone else's pixel. iOS 14.5's tracking prompt and the slow death of third-party cookies pushed DTC brands hard toward first-party data and server-side tracking. This is a big reason GA4 became less optional and more infrastructure.
Blended vs. platform-attributed metrics: platform-attributed numbers come from a single ad platform's own model. Blended numbers combine total spend and total revenue across everything. Example: $50,000 total ad spend across Meta, Google, and TikTok, $150,000 total revenue. Blended MER is 3.0. Meta alone might report a 4.2 ROAS on its slice of that spend. Both are "true." Neither tells the full story alone.
MER (Marketing Efficiency Ratio) is total revenue divided by total ad spend. No attribution model, no per-channel breakdown, just a blunt efficiency check. Formula: Total Revenue / Total Ad Spend.
Contribution margin is revenue minus variable costs, including shipping, payment processing fees, and cost of goods, but before fixed overhead. It's a sharper number than gross margin because gross margin usually ignores the shipping and transaction costs that quietly eat DTC margins alive.
AOV (Average Order Value) is total revenue divided by number of orders. Bundling, free-shipping thresholds, and post-purchase upsells all move this number, sometimes a lot, which is why a rising AOV alone doesn't mean your funnel is healthier. It might just mean your bundle page is working.
New customer revenue vs. returning customer revenue should never get blended into one line. A brand doing $2M a month could be growing purely on repeat purchases while new customer acquisition quietly stalls, and a single blended revenue chart would hide that completely. Separate these two lines and you separate two very different business problems.
ROAS (Return on Ad Spend) is ad revenue divided by ad spend. Formula: Ad Revenue / Ad Spend. As covered above, always check whether you're looking at platform-reported or blended ROAS before reacting to it. If you're doing this math manually across channels, a ROAS calculator saves you from spreadsheet errors that quietly skew budget decisions.
CAC (Customer Acquisition Cost) is total spend divided by new customers acquired. The common mistake: calculating it off ad spend alone, ignoring agency fees, tools, and creative production costs. A "fully loaded" CAC includes all of that. It's almost always higher than the number teams present in meetings.
Quick paid-media basics, since these get thrown around loosely too:
CPC (Cost Per Click)
CPM (Cost Per Mille)
CTR (Click-Through Rate)
Frequency is the average number of times a single person sees your ad within a set period. Saturation curve describes the point where increasing frequency stops driving incremental sales and starts driving ad fatigue instead. On Meta and TikTok specifically, watch frequency climb alongside a falling CTR: that combination is the early warning sign, not a drop in ROAS itself.
LTV (Lifetime Value) estimates the total revenue (or profit) a customer generates over their relationship with your brand. Common pitfalls: measuring LTV over a 90-day window when your actual repurchase cycle is 6 months, or averaging LTV across your whole customer base instead of by acquisition cohort. A cohort acquired through a discount promo behaves nothing like one acquired through organic search.
LTV:CAC ratio compares lifetime value to acquisition cost. The commonly cited benchmark is 3:1, meaning a customer should be worth roughly three times what it cost to acquire them. Treat that as a general rule of thumb, not gospel [VERIFY industry-specific benchmarks before treating 3:1 as a target]. A subscription skincare brand and a one-time-purchase furniture brand shouldn't be held to the same ratio.
Repeat purchase rate is the percentage of customers who buy more than once. Cohort retention curve plots how a group of customers acquired in the same period keeps buying (or stops) over time. It's the difference between knowing "we have decent retention" and knowing exactly which month customers tend to drop off.
Churn rate, for subscription DTC models specifically, is the percentage of active subscribers who cancel in a given period. Get this wrong and your LTV projections are built on sand, since LTV for subscription businesses depends heavily on how long the average subscriber sticks around before churning.
Data warehouse is a centralized database built to store and query large volumes of data from multiple sources. Trivas dashboards, for example, run on Amazon Redshift as the underlying warehouse, which is what lets us pull Shopify, Amazon, and ad platform data into a single queryable layer instead of leaving it scattered across five logins.
ETL/ELT pipeline is the process that moves data from source systems (Shopify, Meta, Amazon Ads) into that warehouse. ETL transforms the data before loading it; ELT loads it raw and transforms it after. The difference matters mostly to engineers, but the outcome matters to everyone: it's why your dashboard numbers update automatically instead of needing a manual CSV export every Monday.
Single source of truth is the idea that everyone in the company looks at the same number for the same metric, pulled from the same place. This matters because Shopify, Meta, and Amazon will each report a different revenue figure for the exact same day, and none of them are lying, they're just measuring different things. A BI reporting layer that unifies these sources is usually the only real fix.
Funnel visualization, in a GA4 context, maps the drop-off between events: add to cart, checkout initiated, purchase. A 60% add-to-cart-to-checkout rate paired with a 20% checkout-to-purchase rate tells you exactly where to focus, which is a lot more useful than a single blended conversion rate.
Forecasting / predictive analytics projects what's likely to happen next, based on historical patterns, as opposed to reporting on what already happened. It's the difference between a rearview mirror and a windshield, and most DTC dashboards only give you the mirror. For a deeper breakdown of formulas and edge cases across all of these terms, the data dictionary is the fuller reference.
None of this is academic. A marketing team that agrees on what "MER" and "ROAS" actually mean, and when to use which one, stops having budget arguments rooted in a vocabulary mismatch instead of a real strategy disagreement.
These definitions get a lot more concrete once they're tied to a real dashboard, one that's actually pulling live Shopify, Amazon, and ad platform data into a single view instead of living in separate tabs. That's usually when a term like "blended MER" stops being abstract and starts being the number someone's staring at every Monday morning.
If you want the formulas, edge cases, and metric-by-metric breakdowns in one expandable reference, the data dictionary covers the details this guide only had room to summarize.
explore more insights
3 min read
3 min read
3 min read
.d53b12e5.png&w=3840&q=75 "Trivas.ai")