Almost every owner I talk to has a marketing dashboard somewhere. Someone built it, there was a walkthrough call, it looked good. Then it got opened twice in the first month and never again.
When I ask why, the answer is always some version of the same thing. I looked at it, the numbers went up and down, and I did not know what I was supposed to do about any of it.
That is a design problem rather than a discipline problem. The dashboard put every number on one screen at one frequency, so a number that should change your Tuesday sits next to a number that only means something after six months. Both blink. Neither asks for anything.
The dashboard nobody opens
Here is the usual contents of an abandoned dashboard at a 10 to 50 person B2B company. Sessions. Users. Bounce rate. Pages per session. Average session duration. Traffic by channel. Top landing pages. Keyword positions. Social followers. Email open rate. Form submissions, buried somewhere near the bottom.
Of those, exactly one describes something that happened to the business. The rest describe website behaviour, and website behaviour at a company doing a few thousand visits a month is noisy enough that week-to-week movement carries almost no information.
An owner who opens that screen is being asked to do the analyst's job: work out which of the twenty numbers matters this week, decide whether the movement is real or random, and then translate it into a decision. Nobody has time for that at 7am, so the tab closes.
The fix is to stop building one report. Build three, on three schedules, with different jobs.
Sort every number by how fast you could act on it
Before you decide how often to look at something, ask one question about it. If this number moved this week, what would I do differently on Monday?
Enquiries dropped to zero for a week. You would ring the office, check the contact form still works, check the phone forwards correctly, and check whether an ad account ran out of money. All of those are same-day actions. So enquiries is a weekly number.
Website traffic dropped 8 percent this week. What would you do? Nothing responsible. Traffic at small-company volumes swings on a statutory holiday, one competitor's newsletter, or a single page getting reshuffled in search results. You cannot separate a real decline from ordinary noise inside a single week, which means a weekly traffic number invites you to act on nothing. So traffic is a monthly number.
Run that test on every number you currently track. A surprising amount of what sits on the dashboard fails it at every frequency, which is the honest signal to remove it.
The three-number weekly check
Weekly is for the things that break. It should take an owner under five minutes, and it should arrive in an email or a message rather than behind a login.
The first number is new enquiries this week, counted from every route. Form submissions, phone calls, direct emails, referrals that came in through a person. One total, with last week beside it and the average of the last eight weeks beside that. Eight weeks is enough to tell a quiet week from a broken week.
The second number is enquiries still waiting for a first reply, and how long the oldest one has been sitting. This is the number that catches the most money at small companies. An enquiry that nobody answered for four days is usually gone, and nobody in the business will volunteer that it happened. It shows up in no traffic report and no revenue report until the quarter closes badly.
The third number is deals that moved this week. Proposals sent, proposals accepted, jobs that went quiet after a promising call. This does not need a system. A five-line list in an email from whoever runs sales is enough, and the list is more useful to an owner than a pipeline chart, because the names are what prompt the useful question.
Three numbers, one screen of text, no charts. If your weekly check needs charts, it is doing a monthly job.
The monthly review: four numbers and one conversation
Monthly is where the slower measures become readable, because a full month is enough volume to separate a trend from a wobble.
Look at total enquiries for the month against the same month last year, rather than against last month. Seasonality at a B2B company is real and strong. Construction-adjacent businesses in Metro Vancouver see it hard in December and again in July. Comparing December to November tells you about the calendar. Comparing December to last December tells you about the business.
Look at where enquiries came from, grouped into five or six buckets at most: search, referral, repeat customer, paid, events or outbound, and unknown. Expect unknown to be large. A large unknown bucket is honest, and it is more useful than a tidy chart built on guesses about which touch mattered most. If you want to understand why that bucket refuses to shrink, the thinking behind attribution at a small company is worth reading once and then not worrying about again.
Look at the conversion step that hurts most: enquiries that became a real conversation, and conversations that became a proposal. Two percentages. If enquiries are up and conversations are flat, you have a follow-up problem or a lead-quality problem, and those two have completely different fixes.
Look at total website traffic and the pages that brought people in. Monthly is the right frequency for this. It answers whether the work of the last few months is reaching anyone, and it moves slowly enough that month-over-month movement means something.
Then have the conversation, which is the actual point of the monthly review. One hour, whoever runs marketing and whoever runs sales in the same room. What are we changing based on these four numbers, and who is doing it by when. A review that ends without one written decision was a reading exercise.
The quarterly question
Quarterly is for the two things that are too slow to see any other way.
First, what did we spend and what came back. Add up every marketing cost for the quarter: agency fees, ad spend, software, the share of a salary that goes to this work, the contractor invoices. Put it against closed revenue where you can trace the origin, and be honest about the traceable portion rather than inflating it. A number like this only has to be directionally right to be useful, and it is the only number that answers whether the whole programme deserves next quarter's budget.
Second, who are we winning and who are we losing. Pull the last quarter's won and lost deals and read them as a list. Size, industry, how they found you, why the lost ones went elsewhere. This is where positioning decisions get made, and it is the input that improves the next three months of marketing more than any dashboard. A repeatable version of this sits in a quarterly review of who your best customers actually are.
Neither of these is automatable. Both require someone to sit and think for two hours. That is why they are quarterly.
What to ignore
Bounce rate. At a B2B site, a visitor who lands on your service page, reads it, gets your phone number and calls you is recorded as a bounce. The metric measures something, and it is not what you want to know.
Average position in search results. It mixes hundreds of different searches into one average, so it moves for reasons that have nothing to do with the searches that bring you work. Track positions for the five or six phrases a real buyer would type, and ignore the average.
Social followers, impressions, and reach, unless you sell through social. At most Metro Vancouver B2B companies these numbers exist because they are easy to export.
Email open rate has become unreliable since mail providers started loading images automatically on behalf of the recipient, which registers as an open the recipient never performed. Replies and clicks still mean something.
And any number where you cannot say, out loud, what you would do if it moved 20 percent in either direction. If you cannot answer that, it belongs in a file somewhere rather than in front of you.
Who prepares it and how long it should take
The weekly check should take one person 10 minutes and it should not require an owner to log in anywhere. Whoever handles enquiries pulls the three numbers and sends them. If that takes longer than 10 minutes, the enquiry records are scattered and that is the problem to fix first.
The monthly report should take under two hours to prepare, and it should be the same shape every month so the comparison works. If it takes two days, someone is rebuilding it from scratch each time and the effort is going into assembly rather than thinking. That is the point where a standing report built once, using something like a connected dashboard that refreshes itself, pays for the setup.
One person owns the file. Not a committee, not an agency that also grades its own work. If an outside provider prepares your numbers, they should prepare the inputs and someone inside your business should write the interpretation. What belongs on the page is covered in how to measure marketing.
The takeaway
Three numbers weekly, four monthly, two questions quarterly. Weekly catches what is broken. Monthly shows what is working. Quarterly decides what to fund.
If your current reporting does not sort itself into those three boxes, it is a single screen doing three jobs badly, and that is why it stopped getting opened. Cut it apart, send the weekly one as plain text, and put a written decision at the end of every monthly meeting.
The test of good reporting is whether anything changed because of it. If nothing changed in six months, the report is a habit rather than a tool.