by Trivas.aiHere's the pattern: Meta ROAS looks fine. Google ROAS looks fine. TikTok is even up a little. And yet blended profitability is quietly sliding, and nobody can explain why.
That's what happens when a dashboard is built channel-first instead of business-first. Each platform grades its own homework, reports the numbers that make it look good, and none of them are forced to reconcile against what actually landed in the bank account.
A real marketing efficiency dashboard doesn't start with channels. It starts with one question: how much did we spend across everything, and how much revenue and margin did that spend produce? That's the whole point of learning how to build a marketing efficiency dashboard for DTC instead of stitching together five platform exports every Monday.
The anchor metric for this is MER, marketing efficiency ratio: total revenue divided by total ad spend. No attribution model, no per-channel claims, just the top-line math.
This guide is for DTC founders and growth leads running enough spend across Shopify, Meta, Google, and often Amazon that the platform-native numbers have stopped adding up to reality.
Start with the core layer, the numbers that describe overall business health:
Underneath that sits a diagnostic layer. This is where channel-level ROAS for Meta, Google, and TikTok actually earns its keep, not as the headline number, but as the thing you check once MER moves and you need to know why. A falling MER with flat channel ROAS usually points to a mix or margin problem. A falling MER with one channel's ROAS cratering points somewhere much more specific.
Add a new vs. returning customer split too. This one gets skipped a lot, and it shouldn't. MER can hold steady for months while new customer acquisition efficiency quietly rots underneath it, propped up by returning-customer revenue that has nothing to do with this month's ad spend.
One more thing worth saying plainly: platform-reported ROAS from Meta Ads Manager or Google Ads shouldn't sit next to your blended metrics as if it's equally trustworthy. It isn't. Attribution inflation is real, platforms over-credit themselves for the same conversion, and treating that number as ground truth is how teams end up confused about why "great ROAS" isn't showing up in the bank account. If you want a gut check on what a channel's numbers should look like before you compare them to platform claims, the ROAS calculator is a useful sanity check.
None of the metrics above mean anything if the underlying data is messy. At minimum, you need Shopify orders, ad spend from Meta, Google, and TikTok, and GA4 sessions and conversions, all landing in the same place.
The most common data quality problem here is double-counted revenue. Meta claims a conversion. Google claims the same one. Both platforms report healthy ROAS off a single sale. MER doesn't inflate the same way, because it's anchored to actual total revenue, not the sum of what every platform claims. But if you're pulling numbers manually and not watching for this, you'll end up trusting channel numbers that are structurally overstated.
This is really an argument for a warehouse-first setup over spreadsheet pulls. Something like Amazon Redshift, where spend and revenue data lands on a consistent daily grain, in a consistent currency, from a single pipeline, rather than five CSV exports that each define "today" slightly differently. This is the foundation of BI reporting done right, and it's the part most teams underinvest in before jumping straight to dashboard design.
If you sell on both Shopify and Amazon, decide up front whether MER is calculated per-channel or blended across both. Neither answer is wrong, but flipping between the two without noticing is how a dashboard quietly stops meaning anything.
Weekly rollups feel efficient. They also hide the exact thing you're trying to catch. A bad week doesn't announce itself until it's already a bad month. Daily grain is what lets you see an efficiency drop while it's still small enough to act on.
Attribution window is the other decision to make deliberately, not by default. Platforms will hand you 7-day click / 1-day view as the standard setting. Some teams prefer last-click. Fine for channel diagnostics. Wrong for the MER layer specifically.
MER should stay attribution-model-free. Pure spend against pure revenue, no window, no click model, no view-through assumption. That's not a simplification, it's the entire reason MER works as a check against inflated platform ROAS in the first place. The moment you apply an attribution model to it, you've reintroduced the same bias you built the dashboard to catch.
Last thing, and it sounds small until it bites you: lock one time zone and one reporting calendar across every data source. Ad platforms, Shopify, and GA4 don't all default to the same day boundary. A one-day mismatch here is enough to make a perfectly fine Tuesday look like a crisis.
Layout matters more than people expect. A dashboard someone actually opens every morning needs to answer "is something wrong" in about three seconds.
Top row: MER trendline, 30/60/90-day, with a target line drawn across it. A drop should be visible without reading a single number.
Second row: blended CAC and contribution margin, side by side. This pairing catches the case that MER alone misses: a ratio that looks healthy while margin is shrinking underneath it, usually from discounting or rising COGS.
Third row: a channel breakdown table, spend, revenue, ROAS, and percent of total spend per channel. This is the click-down layer, the place someone lands after spotting an anomaly up top and needing to know which channel or campaign is behind it.
Build this once, as a template, not as a Monday-morning ritual. Teams pulling from four or more platforms manually are routinely losing several hours a week just assembling numbers before any analysis even starts. That's not an efficiency dashboard, that's a part-time job. This is the exact gap marketing leaders run into once spend crosses a few channels: the reporting overhead grows faster than the team does.
A dashboard someone checks when they remember to isn't a monitoring system, it's a folder. The better setup is threshold-based: flag MER when it drops more than 10% week over week, and let the alert come to you.
The useful next layer is having something surface the likely cause automatically, rather than leaving that diagnosis to whoever opens the dashboard that day. An AI insights layer can flag that the drop traces to a specific channel, a specific campaign, or a shift in new-versus-returning mix, instead of just telling you the number moved. That's the difference between a report and something that actually saves you the investigation time. This is what a layer like Insights is built to do: catch the anomaly and point at the reason, not just the symptom.
Once you've got a stable efficiency baseline, forecasting is the natural next step. Spend planning based on a trend line beats spend planning based on last week's snapshot, every time.
A few ways this setup quietly goes wrong, even for teams doing everything else right:
Mixing attributed and total revenue. Plugging platform-attributed revenue into the MER formula instead of true store revenue understates efficiency in some places and overstates it in others. Pick total revenue and don't deviate.
Refreshing on mismatched schedules. Ad spend updates daily, order data updates on a different cadence, and now yesterday's MER is comparing today's spend to last week's revenue. Same-schedule refresh isn't optional.
Ignoring discounts, refunds, and COGS in the margin layer. MER can look great while contribution margin is bleeding out from a discount code nobody re-evaluated in six months.
Never revisiting the target MER threshold. The target that made sense at $500K in monthly spend usually doesn't hold at $2M. A dashboard that never gets recalibrated slowly turns into background noise.
Trivas unifies Shopify, Amazon, and ad platform data on Redshift, so MER and blended CAC get built on one consistent daily source instead of five spreadsheets stitched together under deadline pressure. The Wingman AI layer sits on top of that, flagging MER drops and pointing to the channel or campaign most likely behind them, so the diagnosis isn't something you're doing manually at 8am on a Monday.
If you're currently rebuilding this dashboard by hand every week, it might be worth seeing what it looks like set up properly, on your own data, once. No pressure to switch anything, just worth a look.
explore more insights
3 min read
3 min read
3 min read
.d53b12e5.png&w=3840&q=75 "Trivas.ai")