Clients read one number in your white label report

Watch a client open a monthly report. They scroll to find one thing, look at it for a few seconds, and close the file. If what they were looking for is not there, or is on page four next to a chart of impressions, they close it anyway and the report has cost you an hour for nothing.

This is not a criticism of clients. It is a description of what a report is for. They are not auditing your work, they are answering one question about whether the money is doing something, and the report either answers it in the first five seconds or it does not.

White labelling is the cheapest part and it is sold as the product

Reporting tools compete on it. Your logo, your colours, your domain, your sending address. All useful, all worth having, and none of it changes whether the document gets read.

A rebranded dashboard export is still a dashboard export. It contains what the tool measures, laid out the way the tool's designer thought about the problem, in the vocabulary of the tool's industry. Putting your logo on it makes you the author of a document written by somebody who has never met your client.

The part that matters is the translation, and it cannot be automated because it depends on knowing what this particular business is trying to do. A dental practice cares about booked appointments. A trade supplier cares about quote requests from named companies. A restaurant cares about covers on Friday. None of those is a metric any tool exports, and all of them are the only number the client will look for.

Choosing the one number

Ask, in the first month, what the client would look at if they were only allowed one figure. The answer is usually not traffic, and when it is traffic that is worth a conversation, because a client who cares about visits has usually been trained to by a previous agency.

Three properties make a good headline number.

It maps to money without arithmetic. Enquiries, bookings, orders, calls. Sessions do not. "Impressions" actively does not, and putting a large impressions figure at the top of a report is the fastest way to be caught inflating, because the client eventually learns what it means.

It moves slowly enough to be readable monthly. A number that swings sixty per cent either way on normal variation produces a conversation about noise every second month.

You can influence it. Reporting a number you do not affect invites the obvious question. This is the one that rules out uptime as a headline for most clients, and the reason is worth being clear about: you did not cause the site to be up, the host did.

The four blocks that get read

One page. Everything else is an appendix, and appendices are fine as long as nothing important is in them.

What changed. The headline number, this month against last, in one sentence with a direction. Not a chart. A sentence, because a sentence cannot be misread and a chart can.

What we did. Three to five items, in plain language, each with the reason. "Rewrote the pricing page because it was the most visited page with the lowest enquiry rate" is a sentence a client can repeat to their partner. "Optimised on-page SEO" is not, and they cannot defend your invoice with it.

What did not happen, and what it would have cost. This is the block almost nobody includes and it is the one that renews contracts. The certificate that renewed on schedule, the backup that ran every night, the plugin update that broke a form in staging and never reached production, the outage detected at 04:12 and resolved before the office opened. Absence of disaster is invisible by construction, and if you do not name it, nobody knows you were there.

What is next, and what you need from them. One or two items, with a date. The second half matters: most stalled work is stalled on a decision or a piece of content from the client, and a report that names it politely every month is more effective than three emails.

The maintenance section, and how to make it honest

If the retainer includes looking after the site, the report has to show it, and this is where inflation is most tempting because the underlying data is easy to generate and impressive to look at.

Three lines are enough, and each one should be a fact you can produce evidence for.

Availability, stated with its measurement conditions rather than as a bare percentage. A figure without the interval and the definition behind it is not a measurement, and a client who later compares it to their host's dashboard will find a different number: what a percentage does and does not commit you to.

Incidents, with detection time and resolution time, including the months where the answer is none. The detection figure is the one that describes your service rather than the host's, and it is the number worth putting in front of a client who is wondering what the monitoring line is for.

Updates applied, and specifically the ones that were held back and why. "Eleven updates applied, one deferred because it changes the checkout template and we want to test it against your payment provider first" is a sentence that demonstrates judgement. A count on its own demonstrates a cron job.

Automate the gathering, never the sending

The economics of reporting on thirty clients push everything towards a scheduled export. That is the right instinct for the data and the wrong one for the document.

A report that arrives automatically, with no human sentence in it, teaches the recipient that nothing in your email requires attention. Within four months it is a filter rule. Then, on the month when something genuinely needs a decision, the message goes to the same folder.

The compromise that works is to generate everything and write two paragraphs. Ten minutes per client. Those two paragraphs are the entire product from the client's point of view, because they are the only part that proves a person looked.

One test for whether a report is worth sending: could it have been produced without anyone reading the site's data? If the answer is yes, you have sent a receipt, and a receipt for a retainer invites a conversation about the retainer.

What to leave out

Anything the client cannot act on and cannot verify. Keyword position tables running to three pages, when four positions actually matter. Bounce rate without a question attached to it. Domain authority scores from a vendor, presented as though they were a property of the site rather than an estimate by a third party.

And any number you cannot reproduce on request. The awkward moment is not being wrong, it is being asked where a figure came from and discovering that it came from a panel whose methodology you have never read. Everything in a report should be traceable to something you could show them, which is a stricter rule than it sounds and eliminates about half of what most tools export.

The reason all of this repays the effort is that the report is the only regular contact most clients have with the work. Between one report and the next, the site works, nothing arrives, and the retainer looks like a subscription to nothing: what a client believes they are buying, and what they are actually paying for.

What one number looks like when it is the only one

Our monthly report carries seven values and no more: total checks, uptime percentage, seconds down, average response time, the per monitor breakdown, the incident list, and the day by day series. You can count them yourself in src/Report.php. It goes out on the first of the month by default, which is the one editorial decision in the file.

Seven is not a magic number, and four would not have been either. What matters is that a payload you can list in one sentence is a payload a client can argue with, and a client who can argue with your report is a client who read it. Twelve charts produce agreement without reading, which is the same thing as no reporting at all.