Most pre-send QA goes into the part of the report that is least likely to be wrong. You reread the commentary, you sanity-check the CPA against last month, you second-guess whether the dip in week three deserves an explanation. Meanwhile the thing the client will actually email about is sitting in the header: a date range that still says September. A useful client report checklist is built around that asymmetry, not around the analysis.
This matters most on the Friday where five reports are queued and the person sending them has no capacity left for a real review. You are not going to re-derive the numbers at 16:00. You can, in five minutes per report, catch the handful of errors that cost credibility out of proportion to how trivial they are.
The mistakes clients notice are rarely the ones you spent the most time on
A client reading a monthly report is not auditing your attribution logic. They are scanning for the three or four things they already have an opinion about: the spend figure, whether the period matches the invoice, whether the headline number moved, and whether it looks like it was made for them.
That is why the errors that get caught are almost always surface-level. Last month’s reach number left in the summary paragraph while the chart below it shows the new one. A logo that is the client’s old mark from before the rebrand. A link that opens fine on your laptop and asks for a password on their phone. A table header still labelled “Q2” in an October report. None of these change a single conclusion, and all of them make the client wonder what else was copied forward without looking.
The analytical mistakes do happen, but they tend to be caught internally, or they survive for months because nobody outside the team can see them. Pre-send review time is better spent where the detection rate is highest.
A client report checklist with only four items on it
Run these in order. Each one takes under a minute on a report you built yourself.
- Date range. Check it in three places: the report header, the chart axes, and anywhere the period is written in prose. These do not always update together.
- The summary against the data. Take every number written in the commentary or executive summary and find it in the report below. Not “does it look right”, but literally locate it.
- The link, from a phone, logged out. Not the preview in your builder. The actual thing you are about to paste into the email.
- The branding and the client name. Logo version, client name spelling, and whether any placeholder text survived from the template.
That is the whole report QA checklist. If you have time for a fifth, reconcile delivered spend against the media plan, because budget is the one number clients check without being asked.
The date range is the first thing to break and the last thing anyone checks
Date ranges break because they live in more places than anyone remembers. The filter on the data, the title of the report, the subtitle on a chart someone duplicated in March, the sentence in the commentary that says “over the past four weeks”. Change the period in one place and the others quietly disagree.
The failure mode worth watching is the partial month. If the export was pulled on the 29th and the report says “1 to 31 October”, the spend figure is short and the client’s finance team will find the gap before you do. Either pull the full period or say explicitly what the report covers. “Data through 29 October” in the header costs nothing and prevents a conversation.
Reports built on a frozen export inherit this problem permanently, because the file and the label drift apart the moment anything is regenerated. When the report is a live page fed from the campaign file you uploaded, the period shown and the data shown come from the same place, which removes the most common source of client report errors without anyone having to check it. That is also why the handoff from export to report deserves as much attention as the layout does.
Open the link the way the client will, not the way you built it
You have the report open in a tab where you are already authenticated. The client does not. They are on a phone, on mobile data, possibly on a corporate network that strips unfamiliar domains, and increasingly they are forwarding your link to someone you have never met.
So open it cold. New browser profile or private window, phone if you can, and click through to the sections that matter. Check that charts render at narrow widths, that long table rows are still readable, and that the access key or link you are about to send actually grants the access you think it does. Sharing setups fail in ways that are invisible from inside the account, which is the main argument for understanding how your access keys and share links behave before a client’s IT team asks.
If you send through an automated schedule, test the delivery path too. A report that is correct but arrives from an address the client’s filter distrusts is still a report that did not arrive, and keeping scheduled delivery from sounding automated is a separate problem from making the file right.
The executive summary is where last month’s number hides longest
The summary is written in prose, which means it is never regenerated and never validated against anything. Every other part of the report is tied to data. The summary is tied to whatever you typed last time.
The check is mechanical: read the summary with a highlighter mindset, mark every figure, and confirm each one appears somewhere below. Also check the direction words. “Up”, “improved”, “ahead of plan” are claims, and they survive copy-paste just as easily as numbers do. A summary saying CPM came down while the chart shows it climbing is worse than no summary, because it reads as though nobody looked.
Summaries that are built to be read rather than skimmed tend to be shorter, which also makes them faster to verify. Three sentences with two numbers in them is a thirty-second check. Four paragraphs is not.
What to check after the client has already opened it
Sending is not the end of the review. Two things are worth a glance in the days after.
First, which questions come back. A question that repeats across three clients is usually a structure problem, not a comprehension problem, and it often means something is in the wrong place rather than missing. Section order shapes what clients ask more than most people expect.
Second, whether you had to send a correction. Keep a running note of what you fixed after the fact, even informally. After a quarter, that note is a better pre-send report review list than any generic fifty-point audit, because it is made of mistakes your team actually makes. Most agencies find the same four or five items keep reappearing.
If you want the stale-date and mismatched-summary problems to stop being checklist items at all, that is largely a tooling question: see the plans for how live reports change the pre-send step.
Questions that come up around pre-send checks
How long should a pre-send check take? Five minutes for a report built from an existing template, closer to fifteen for a first report on a new account or the first run after a template change. If it routinely takes longer, the problem is upstream: too much is being retyped by hand.
Should someone else proofread the report? A second pair of eyes helps most on new accounts and on reports with bad news in them. For recurring monthly reports on stable accounts, peer review usually costs more than it catches, and the same time spent on a structured self-check finds more. Report proofreading in an agency works better as a rotating spot-check than as a rule applied to everything.
What is the most common error clients actually report? Mismatches between a number in the written summary and the same number in a chart or table, followed by date ranges that do not match the billing period. Both are copy-forward errors rather than analysis errors.
Do we need a formal sign-off process? Only if reports go to a regulated client or a legal department. For most retainer work, a documented checklist that the sender runs themselves gets you most of the benefit without adding a dependency on someone else’s calendar on the busiest afternoon of the week.
