Article
The dashboard nobody opens: the four levels and how to tell which one you're at
The problem with an ignored dashboard isn't the number of charts: it's that the numbers inside come from sources that disagree with each other. Four maturity levels, each with a precise limit.
The problem with a dashboard that gets ignored is almost never the number of charts. It’s that the numbers inside come from sources that disagree with each other, and the people looking know it. There are four maturity levels, each with a precise limit, and moving up a level costs very differently each time.
Category: measurement · 8 minute read · Felice · updated 18 August 2026
What this is about
You have a dashboard. Three people look at it, nobody decides anything on the basis of what they see, and every so often someone asks why the revenue in there isn’t what the accountant sees.
The question this article answers: what maturity level is your measurement at, and what it takes to move up one.
In order: the three questions a dashboard has to answer, the four levels with each one’s limit, and what it costs to move from one to the next.
Method and limits. The levels described are the ones we find on the stores we work with, and project numbers come from our own audits with the window stated. The cost figures are our own rate card, not a market price. This is not a comparison of visualisation tools.
You’ll find three labels in the text. Documentation when the source is official, measured by us when the number comes from one of our audits, recommendation when it is a professional judgement rather than a fact.
Contents
- The three questions a dashboard has to answer
- Level 1: platforms grading their own homework
- Level 2: the Monday spreadsheet
- Level 3: the connected dashboard
- Level 4: the reconciled warehouse
- What moving up costs, and when it isn’t worth it
The three questions a dashboard has to answer
Before the levels, the criterion for judging them. A dashboard someone actually opens answers three questions and no more.
Are we growing. One metric at the top, and for a store it isn’t gross revenue. It’s what’s left after cost of goods and after ad spend, because it is the only number that changes when you change the budget.
Where are we investing and what is it returning. Three visible dimensions at most, for example channel, market and product line. Everything else is a deep dive, and the deep dive lives on another page.
What changes next week. No absolute number on its own. Comparison with last year, with the target, with the previous week. A number without a comparison produces no decision.
Recommendation. Add the time of the last update to every tile. It isn’t cosmetic: someone who can’t tell whether they are looking at this morning’s data or data from ten days ago will not decide.
None of the four levels that follow answers all three well. What changes is which question stays uncovered.
Level 1: platforms grading their own homework
You open Ads Manager for spend, GA4 for traffic, the store admin for orders. No tool in between.
What works. It costs nothing, it is always up to date, and for a single campaign it is the right place to look.
The limit. Each advertising platform measures its own contribution with its own models and its own windows. None of them certifies revenue, and none has any reason to tell you that someone else brought in a sale.
The second limit is more practical. People read different dashboards. A client described it to us like this: the same end customer is six or seven different entities in six or seven systems that don’t talk to each other, marketing looks at the platform data, sales looks at the back office, management looks at the store, and the numbers never match.
How you know you’re here. In meetings people argue about which number is the right one instead of what to do.
To move up you need one person who once a week puts the figures in the same place. Nothing more.
Level 2: the Monday spreadsheet
Someone exports, pastes, calculates two ratios and sends the sheet. It is the level where most stores under a million in revenue sit, and it works better than its reputation suggests.
What works. The figures are finally next to each other. Whoever fills in the sheet is also the person who knows where each column comes from, and that knowledge is worth more than the tool.
The limit. It is all in one person’s head. When they’re on holiday the sheet doesn’t go out, and when they change jobs nobody knows what time window column C was taken over.
The second limit is that the sheet inherits level 1’s problem without solving it. If two sources disagree, the sheet puts them side by side and leaves the argument open.
How you know you’re here. There is a file with a person’s name in it.
To move up you need to connect the sources to a visualisation tool. That’s half a day’s work for standard connectors, and the tool can cost nothing.
Level 3: the connected dashboard
The platforms’ APIs feed a visualisation tool directly. The spreadsheet disappears, the charts update themselves.
What works. Nobody wastes time copying any more, and the data arrives every morning without anyone producing it.
The first limit, and the most underrated. The dashboard inherits the platforms’ definitions without questioning them, and definitions can be written badly.
Measured by us. On one GA4 property, two channel groupings defined slightly differently returned results 97.8% apart over the same period and on the same data. It was a match condition written with “exactly matches” instead of “contains”. A dashboard built on top of that definition would have shown numbers that were clean, current and wrong.
The second limit. The back office is still missing. The dashboard compares platforms with each other, not platforms with what actually happened.
The third limit. It breaks silently. When a connector stops working or a tag changes behaviour, the chart doesn’t turn red: it keeps drawing a lower line, and for weeks someone reads that drop as a market problem.
How you know you’re here. The dashboard looks good and nobody has ever reconciled it with real revenue.
To move up you need to bring orders into the system, which is integration work and not visualisation work.
Level 4: the reconciled warehouse
Raw events and the order ledger end up in the same place, usually a data warehouse, and metrics are computed there with definitions written once.
What works. There is finally a reference point that belongs to no platform. The question “how much of this difference is normal” becomes answerable, because you have the yardstick outside the system you are judging.
Measured by us. On a Shopify store, comparing the order ledger with GA4 over 90 days, 62% of online revenue was attributed to no source at all. No dashboard built on the platforms alone could have shown it, because that 62% exists in none of the sources feeding the charts.
The limit. It costs, and the cost isn’t the tool. It’s maintenance: interfaces change, definitions need updating, and without someone looking after them the warehouse becomes a more expensive level 3.
The reference threshold. On a healthy store the distance between platforms and back office sits around 10 or 15%. Above 40% something is broken and should be found before building anything on top.
How you know you’re here. You can say where every number comes from, with what window and what definition, without asking anyone.
Above level 4 there is no level 5, there is a property you can add to 3 or 4: active monitoring, that is a check that warns you when a number goes off the rails instead of waiting for you to look at it.
What moving up costs, and when it isn’t worth it
No technical recommendation without the arithmetic. It is a rule we apply internally and it applies here too.
| Situation | The level that holds | Why |
|---|---|---|
| Under €10,000 a month in ad spend | Level 2 | Maintenance time costs more than what level 3 saves |
| Between €10,000 and €25,000 a month | Level 3, with orders reconciled by hand once a month | Periodic reconciliation gives 90% of level 4’s benefit at a fraction of the cost |
| Above €25,000 a month, across several channels | Level 4 | The budget misallocated on a wrong baseline exceeds the cost of the warehouse |
| Above €50,000 a month, with twelve months of history | Level 4 plus statistical models | Below that threshold the models don’t have enough data to say anything reliable |
Recommendation, with our own numbers. Level 4 on our rate card is a setup project in the order of thousands of euros plus a monthly fee for upkeep, with a sensible minimum term of twelve months, because under a year maintenance doesn’t have time to pay for itself. Below €10,000 a month in spend we don’t propose it, and we say so upfront.
A warning worth more than the table. A dashboard can be worse than no dashboard. We have come across reports where return on spend was inflated by counting organic traffic among campaign results. They were current, tidy, and led to wrong decisions with more confidence than a hand-made spreadsheet.
In short
The four levels are not a ladder to climb all the way. They are four different trade-offs between cost and reliability, and the right level depends on how much you are spending.
The jump that changes things is the one that brings in the order ledger, because it is the first moment you have a yardstick no platform controls.
Sources
- Google Analytics Help, About attribution and attribution modeling, retrieved 18 August 2026
- Google Cloud, BigQuery Export for Google Analytics 4, retrieved 18 August 2026
- Shopify Help Center, Reports and analytics, retrieved 18 August 2026
- Tillde internal thresholds and rate card, revised 17 August 2026
- Internal measurements from audits of Shopify stores, windows stated in the text
The first step
Take the last closed month. Write down on a sheet the revenue from your back office, then what your dashboard says, then what the advertising platform says.
If the three figures are within 10 or 15% of each other, your current level is enough. If one is far off, you’ve found the work to do, and we’re happy to talk.
Content verified on 18 August 2026. Cost thresholds are updated once a year; if you are reading this much later, ask us to confirm.