Six tiles. That is roughly what a client sees on a shared dashboard before the first scroll: a KPI row, a chart, and the top edge of whatever comes next, on a 13-inch laptop with a browser toolbar and a bookmark bar in the way. Everything past that is below the fold, and below the fold is a decision the client makes about whether your report deserves more of their morning.
Most teams treat the number of client dashboard metrics as a running total. A client asks for cost per completed view in week three, so it gets added. Someone in procurement wants spend split by publisher, so that gets added. By month four the home screen carries fourteen tiles, nobody remembers who asked for the eleventh, and the account manager is getting the same feedback the old deck used to get: it looks busy, and the client still asks “so how did we do?” on the call.
The tile count is not a total. It is a design limit, and it should be set before the first client ever asks for anything.
Six tiles fit, and then the scrolling starts
The arithmetic is unforgiving. A KPI tile needs a label, a value, and usually a comparison against the plan or the previous period. Make that readable at the distance someone holds a laptop and you get four across on a wide screen, two or three on a smaller one. Add a chart with a legend and you have used the visible area.
So a dashboard with twelve KPI tiles does not show twelve metrics. It shows six and hides six, while telling the client, through sheer volume, that all twelve are equally important. That is the worst of both outcomes: no hierarchy and no brevity.
Mobile makes it sharper. Tiles stack. Twelve tiles become twelve screens of thumb-scrolling before the first chart, which is exactly how a client reads your report on the train before a 9:30 internal meeting.
More client dashboard metrics read as homework, not transparency
The usual defence of a crowded home screen is transparency. We show everything, nothing is hidden, the client can see the full picture. In practice the client reads a wall of numbers as a request to do work.
Think about what you are actually handing over. Fourteen tiles with no ranking means the client has to decide which ones matter, compare them against a media plan they do not have open, and form a view on the month without any indication of what you think. That is your job, delegated upward, in a layout that looks thorough.
The agencies whose reports get read tend to show less and say more. A tight KPI row plus one sentence of interpretation beats a complete metric inventory, because the client’s real question is whether to keep spending, not whether you captured every column from the export. There is also a quieter cost to reporting dashboard overload: when everything is on the home screen, nothing on the home screen is a signal. A rising CPA no longer stands out, because it is sitting next to eleven numbers that are behaving normally.
The home screen answers one question
A useful constraint for dashboard home screen design: the first screen exists to answer one question, and that question depends on what the campaign was bought to do.
For a lead generation account it is cost per qualified lead against target, with volume next to it. For an always-on brand campaign it is completed views or VTR against the plan. For retail media it is attributed sales against spend. For a pacing-sensitive programmatic flight it may simply be delivery: spend to date, impressions to date, days remaining. Different channels have genuinely different headline questions, which is why one dashboard rarely survives contact with every campaign type.
Once that question is named, the home screen fills itself. One tile is the answer. Two or three qualify it. One chart shows the trend behind it. Everything else belongs further down the page, in the tables and breakdowns a client reaches when they have a specific follow-up. That is not hiding data. It is putting the second question on the second screen.
What earns a permanent tile
A metric earns a permanent place on the home screen if it can change the next decision. Pacing qualifies, because being 18% under plan with nine days left triggers an action this week. CPA against target qualifies. Frequency on a reach campaign qualifies, because runaway frequency is both a problem and a fixable one.
A metric belongs further down if it is diagnostic rather than directional. CTR by creative, CPM by publisher, clicks by device, impressions by day of week: these explain the headline once someone has asked why, and they are close to meaningless as standalone tiles. If you want a working split between the two, the twelve metrics worth showing and how to read them is the shorter version of this argument.
Tiles that fail both tests usually survive for political reasons. Someone senior asked for the number once, in 2024, during a specific crisis.
“Just one more number” comes from the same few places
Requests to add a tile are rarely about the tile. Three patterns cover most of them.
The client needs the number for their own internal reporting. They have a monthly slide for their marketing director and your dashboard is the only source. A tile is one answer, but an export of the underlying table is usually the better one, because they need it in their format anyway.
The client does not trust a number and wants to triangulate it. Adding a second metric next to a suspect one does not build trust. Explaining the definition does. A glossary line or a note on how the rate is calculated settles it permanently.
The client saw something different in a platform UI. This is a reconciliation conversation wearing the costume of a feature request. Adding the platform’s version of the metric alongside yours creates two sources of truth on the home screen and guarantees the question comes back next month.
None of those three is solved by a new tile. All three get worse when the home screen grows.
Put the limit in the template, not in the conversation
The reason the dashboard tile count creeps up is that it gets renegotiated per account, per month, by whoever is least able to say no at the time. Nobody wants to tell a client their request is not important enough for the first screen.
Fix it once, structurally. Decide the home screen layout per campaign type, write down what the headline question is, and make the second screen the default destination for anything diagnostic. Then the answer to a request stops being a judgment call and becomes a description of how your reporting works: that number lives in the breakdown below, here is where to find it.
This is where a reusable template does more than save setup time. In Screebl the layout is built once per campaign type and the template locks as soon as reports use it, so the tile count is a property of the template rather than a monthly argument. Each row carries its own group-by and metrics, which means the diagnostic detail still exists in the report, just not in the KPI row. The broader case for working this way is in building the layout once and reusing it; if you want to see which plan fits the number of accounts you run, the pricing page is the short version.
The test for any dashboard is not whether the client can find a metric. It is whether someone who opens the link for forty seconds leaves with the right impression of the month. Six tiles can do that. Fourteen almost never do.
