The assumption behind giving a client live dashboard access is that transparency buys you patience. Show the numbers as they move and the client will stop chasing you for updates. For the first few weeks it tends to work in reverse. The client opens the link on a Tuesday morning, sees conversions down 34% week on week, and messages you before you have looked at the account at all. Dashboard annotations are the cheapest fix for that, and most agencies either skip them or write them like internal notes nobody outside the team can parse.
What changed with live reporting is not the client’s temperament. It is the order of arrival. In the PDF era the number and the explanation were the same artefact: you exported on Monday, saw the dip yourself, wrote two sentences about the creative rotation that ended on the 7th, and sent both together. The client never met the number alone. Now they do, every time they open the link, and whatever story they build in those five minutes is the one you have to argue with.
Annotating is not about covering yourself. It is about making sure the first explanation the client reads is yours rather than the one they invent on the way to their own management meeting.
The dip is visible for hours before anyone on your side notices
Account managers do not watch dashboards. They work in the platforms, in the media plan, in their inbox. The client, on the other hand, has one dashboard and sometimes checks it with coffee in hand. That asymmetry is the whole problem. A YouTube campaign whose VTR drops because a 6-second bumper started serving alongside the 30-second asset is visible to the client at 08:40 and visible to you when they ask about it.
There is a second version of this that is worse, where the number moved for reasons that have nothing to do with performance. Conversions that backfill over 72 hours, a platform export that arrived short because a market was still processing, spend that looks like it collapsed because yesterday is only half a day. Those are data behaviour, not campaign behaviour, and clients reasonably cannot tell the difference. We wrote about that pattern separately in why dashboard numbers change, and it is the single biggest source of unnecessary Slack messages.
Why “we’ll cover it on the call” stopped being good enough
It was a defensible answer when the dashboard was a monthly deliverable. Nothing in the client’s hands contradicted it. Now the deferral costs you something specific: the client holds an unexplained number for nine days, mentions it to their director, and by the time you get to the review call you are not explaining a dip, you are relitigating a conclusion that has already been reported upward.
Review calls are still where interpretation happens. A live dashboard changes what those meetings are for, but it does not make them a place to deliver news the client found themselves. Annotations are what keeps the call about the decision instead of the archaeology.
A dashboard annotation needs three things, not a paragraph
The failure mode is writing too much. An annotation is not a mini executive summary, and nobody reads 80 words pinned to a KPI tile. What it needs:
A date or range, stated explicitly. “Oct 6 to Oct 9”, not “last week”. The client is looking at a chart and needs to match your note to a shape on it.
The cause, in plain terms, with the mechanism visible. “Creative A exhausted frequency cap, so delivery shifted to Creative B which has a lower CTR” tells the client why the line moved. “Performance fluctuation due to optimisation” tells them nothing and reads like a hedge.
Whether it is handled or being watched. One clause. “Rotation rebalanced on the 10th, CTR recovering since” or “monitoring, no action yet, this is within normal weekly variance”. Clients forgive dips. They do not forgive not knowing whether anyone is on it.
Three lines, written in the same register you would use in an email to that client. If you cannot explain a performance dip to a client in three lines, the problem is probably that you do not understand it yet, and the honest annotation is the one that says you are investigating.
The dips worth annotating, and the ones worth leaving alone
Annotate everything and the notes become wallpaper. The filter I would use:
- Annotate anything that broke a trend the client has been watching, anything caused by a decision you made (budget shift, pacing correction, new market, creative swap), anything caused by the data rather than the campaign, and anything that touches a metric named in the media plan.
- Leave alone normal day-of-week movement, single-day noise on a small-volume metric, and anything that will resolve before the client next looks. A note on every wobble teaches the client that every wobble is noteworthy, which is the opposite of what you want.
Dashboards with fewer tiles make this easier, because there is less surface area generating questions in the first place. That is part of the argument in why six tiles beat fourteen.
Where the note lives decides whether the client reads it
An explanation sent in a separate email is an explanation the client has to go find, which means it does not exist at the moment they need it. Context has to sit next to the number it explains: pinned to the tile, or anchored to the date range on the chart. Clients who have to cross-reference two places will reliably use the one that is already open.
This is why live dashboard context notes work better than a commentary block at the bottom of the report. A commentary section gets read by the person preparing for the review call. A note attached to the conversions tile gets read by the person who just saw conversions drop. Those are often different people, and the second one is the one who messages you.
Screebl reports let you pin a short note to a tile or a date range so the explanation travels with the number instead of living in your sent folder. Where the agency has written a campaign brief, the optional report assistant can answer follow-up questions from that same report, which handles the next layer of “but why” without a thread. Neither removes the need to write the note. They just mean the note is where the client is looking.
Annotation discipline builds client dashboard trust faster than any design change. A client who has seen you flag three dips before they noticed them stops treating the dashboard as a surveillance tool and starts treating it as reporting, which is a different relationship and a much easier renewal conversation. It also reduces the question volume that arrives even on a well-built report, though it never gets it to zero.
If you want to see how annotated live reports look from the client’s side before committing a team to the habit, the plans are a reasonable place to start.
