by Trivas.aiMost ecommerce founders check five tabs before 9am: Shopify orders, Amazon Seller Central, Meta Ads Manager, Google Ads, and a GA4 report that never quite matches the others. None of them agree on revenue. None of them show margin. That's the gap an ecommerce KPI dashboard is supposed to close, and it's also where most attempts at building one fall apart.
This post covers what actually belongs on that dashboard, why the DIY versions (spreadsheets, native platform reports, cobbled-together Looker Studio pages) tend to break down, and what a setup that survives contact with a real P&L looks like.
An ecommerce KPI dashboard is one screen that pulls revenue, ad spend, and operational data into a single view. Not five logins. Not a Friday afternoon spent copy-pasting numbers into a spreadsheet before a Monday meeting. One place, updated on its own.
That's different from a generic reporting tool. A reporting tool shows you everything it can measure. A KPI dashboard shows you the 10 to 15 numbers that actually drive a decision this week, whether that's shifting ad budget or reordering inventory. Volume isn't the goal. Relevance is.
The inputs are usually the same across brands, even if the mix varies:
The hard part isn't collecting these. It's getting them to agree with each other on a timeline, a currency, and an attribution model, which is where most homegrown setups quietly fail.
Not every metric deserves screen space. Here's what does.
Revenue and net revenue. Gross revenue is a vanity number if you're not netting out returns and discounts. Track both, daily, not just as a monthly rollup. A brand running a promo needs to see the discount eating into net revenue the same day it happens, not three weeks later in a finance review.
Blended CAC and channel-level CAC. A blended CAC of $28 sounds fine until you split it out and find Meta is at $19 and Google is at $61. Blended numbers hide exactly the thing you need to see: which channel is quietly losing money while another one subsidizes it.
ROAS and MER, side by side. ROAS tells you how a single channel performed. MER (marketing efficiency ratio, total revenue divided by total ad spend) tells you how the whole business is doing. Looking at ROAS alone is how brands end up "winning" on every channel report while total revenue stalls. Run both, always paired.
If you want to sanity-check either number for a specific channel, a ROAS calculator is a fast way to catch a spend/revenue mismatch before it makes it onto the dashboard.
AOV and repeat purchase rate. These are the two levers that move lifetime value without spending another dollar on ads. A dashboard that only tracks new customer acquisition is missing half the growth story.
Contribution margin per order. Not gross revenue. Contribution margin factors in COGS, shipping, and payment processing fees. This is the number that tells you whether a "profitable" channel is actually funding the business or just moving inventory at breakeven.
Inventory turnover and sell-through rate. Essential for any brand running Amazon FBA or wholesale alongside DTC, where a stockout on one channel doesn't show up until it's already cost you a week of sales on another.
Shopify's native analytics is genuinely good at showing you Shopify data. It knows nothing about your ad spend. So ROAS and MER have to be calculated by hand, every single time, by pulling numbers from Meta and Google separately and dropping them into a formula somewhere else.
Then there's attribution. Amazon Seller Central and Shopify don't use the same attribution windows, and if you sell internationally, they're not even reporting in the same currency by default. Blend those manually and you're not just doing extra work, you're introducing errors that compound every week they go uncorrected.
Spreadsheets have their own problem: staleness. A pull from yesterday is a decision made on yesterday's numbers. If you're deciding whether to scale a Meta campaign today, working off a CSV exported 18 hours ago means you're already behind.
Add it up and most brands spend 2 to 3 hours a week just reconciling channels, before any actual analysis happens. That's not a KPI dashboard. That's an unpaid part-time job with a due date every Monday.
A dashboard that holds up needs a few things spreadsheets and native reports don't offer.
Cross-channel blending, not separate tabs. Amazon, Shopify, Meta, Google, and GA4 need to sit in one data model, joined on order-level data, not stitched together visually after the fact. If a "unified" dashboard still requires you to flip between three views to get a straight answer, it isn't unified.
Daily refresh at minimum. Ad spend and order volume should be closer to real-time. Waiting a day to know if yesterday's campaign spend paid off means you're always reacting a day late.
Drill-down without leaving the tool. A KPI dashboard should let you click from "MER dropped this week" straight down to the campaign or SKU that caused it. If the answer requires exporting to a separate spreadsheet to investigate, the dashboard only solved half the problem.
A warehouse underneath, not just a dashboard layer. This is the part most tools skip, and it's why dashboards slow to a crawl as history builds up. Trivas runs on Redshift specifically so that a year of order-level and ad-spend history doesn't choke the dashboard when you want to look at a trend line instead of just this week. The BI reporting layer sits on top of that, which is the difference between a dashboard that's fast on day one and one that's still fast a year in.
Building in-house means a data engineer's time upfront, and then their time again every time Amazon, Meta, or Shopify changes an API, which happens more often than anyone budgets for. It's not a one-time build. It's an ongoing maintenance line item, whether or not that's how it got pitched internally.
Off-the-shelf tools get you to a first dashboard faster; that's the whole pitch. But "faster" isn't the same as "correct." Before committing to one, check whether it actually blends Amazon and Shopify data natively, or whether you'll still be exporting both and joining them yourself in a spreadsheet, just with extra software licensing on top.
Three things worth checking before signing anything:
If you're selling on Shopify specifically and want to test this without a full platform commitment, Trivas AI on the Shopify App Store is a lighter way to see how the blending works before going further. For Amazon sellers, the considerations are different enough (attribution windows, FBA fees, sell-through data) that it's worth checking how Amazon-specific reporting works separately from the Shopify side, and the Shopify solution page covers what that integration looks like on its own.
Tracking too much. A dashboard with 40 metrics on it isn't more informative than one with 7. It's just noise with a login screen. Pick the numbers that actually change a decision this week and put everything else one click deeper.
Mixing attribution models without labeling them. If Meta's ROAS is on last-click and Google's is on a 7-day view-through, comparing them side by side without saying so isn't a comparison, it's a coincidence. Label the model next to the number, every time. If you're not sure which model a metric on your own dashboard is using, the data dictionary is worth a look before you trust it in a meeting.
Building it once and forgetting it. A dashboard built for a growth-at-all-costs phase looks completely wrong six months later when the priority shifts to margin. CAC-obsessed dashboards that never add contribution margin are the most common version of this. Revisit the KPI list at least once a quarter, not just when something breaks.
A KPI dashboard is the first layer, not the finish line. Once the numbers are actually trustworthy, daily-refreshed, and blended correctly across channels, that's when AI-driven insights and forecasting become worth layering on top. Forecasting on top of bad data just gives you a wrong answer faster.
If you're still at the stage of reconciling Amazon and Shopify numbers by hand, that's normal, and it's worth fixing before adding anything more advanced on top. Worth poking around a live dashboard setup to see what "solved" actually looks like, or subscribing for more on where the reporting stack goes from here.
explore more insights
3 min read
3 min read
3 min read
.d53b12e5.png&w=3840&q=75 "Trivas.ai")