Last week’s best number probably felt good and changed nothing. A post collected a few thousand impressions, someone screenshotted it for the team chat, and on Monday the plan stayed exactly as it was. That number was real. It just wasn’t connected to any decision.
Most analytics tools show you dozens of metrics, and a small brand has the time and attention to act on very few of them. This article builds a short scorecard around the questions a small team actually faces, and it is upfront about how much a small audience can and can’t tell you.
The metric that made you feel good last week and changed nothing
Vanity metrics get criticised for being shallow, but shallow isn’t the real problem. Follower counts and impressions do carry information. The problem is that nothing you will do this month depends on them. A number you can’t act on is trivia, however accurate it is.
The usual advice is to ignore likes and watch conversions instead. That swaps one list for another and still leaves you with a dashboard full of numbers you check out of habit. A better filter is to ask what a metric is for. A small brand makes a limited number of decisions in a month: what to publish next, which channel deserves another hour, which page to fix, whether a paid tool or a collaboration is worth keeping. A metric belongs on your scorecard only if it informs one of those.
This is also why the search for marketing metrics that matter for small business can’t end in a universal list. The right numbers depend on the decisions in front of you. A newsletter-led brand and a service business with a contact form will legitimately track different things.
The one-line test: what will you do differently if this number moves?
Take any metric and ask: if this rose by a fifth next month, what would I change? Then ask the same about a fall of a fifth. If both answers are “nothing”, remove it from the scorecard. You can still look at it occasionally, but it shouldn’t take up a monthly review.
Here is how that plays out with a few common numbers:
- Total pageviews. If they climb, what changes? Usually nothing, because pageviews don’t say who visited or whether it helped. Fails the test.
- Share of visits from people in your target audience. If it drops, you would look at which channel or topic is pulling in the wrong crowd, and adjust. Passes.
- Returning visitors. If they rise after you launch a series, you make more of that series. If they stall, you question whether the content earns a second visit. Passes.
Decide the response before you see the result. “If enquiries fall below what I’d need to cover this month’s costs, I’ll rewrite the contact page before publishing anything new” is a rule. Deciding afterwards invites you to explain away whatever the number says.
Are the right people finding us?
Reach matters only when it reaches people who could become your audience or customers. Ten thousand views from people who will never buy, subscribe or return tell you less than two hundred from people who match the profile you’re building for.
Three measures are usually available without extra tools:
- Where visitors come from, by channel. If your best customers say they found you through search but most of your traffic comes from a social post that went wide, the two groups deserve separate attention.
- The queries you appear for. Google Search Console lists the searches that showed your pages. Read them as language. Do they sound like the questions your audience asks, or like something adjacent that happened to match your wording?
- Location and device, where they matter. A local service that gets most of its visits from other countries has a targeting problem, whatever the total says.
The decision this question informs is where to put your next hour: more of the content that attracts the right people, or a different angle for the topics that attract the wrong ones. The distinction between what people type and what they mean is covered in more depth in search intent vs keywords.
Do they trust us enough to stay?
Pageviews count arrivals. Trust shows up in what people do after they arrive: whether they read, whether they come back, whether they go further into the site.
Two signals work better than raw traffic. The first is the share of visitors who return, which most analytics platforms report as returning users or a similar label. Someone coming back unprompted has judged you worth a second visit, which is a stronger signal than any single page’s traffic. The second is engagement on the pages that matter: do people scroll and spend time on the article, or leave within seconds? Use whichever engagement measure your tool offers, but compare it only against your own pages. Definitions differ between tools, and benchmarks from other sites rarely transfer.
The decision here is about format and depth. If one series brings people back and another doesn’t, you have a reason to publish more of the first. If strong rankings come with weak engagement, the page may be answering the query but failing to earn belief, a problem explored in why a site can rank but not convert.
One caution: a high bounce or short visit isn’t always a failure. A page that answers a question in one screen has done its job. Judge engagement against what the page is meant to do.
Are they taking the next step?
Traffic is something visitors give you passively. A subscription, an enquiry or a click on a recommendation is something they chose to do, which makes actions the most decision-ready numbers on the scorecard.
Define each action explicitly and count it in your analytics as an event:
- Newsletter sign-ups
- Contact or booking form submissions
- Clicks on a specific link, such as a recommended tool or a product page
Then look at the rate: actions divided by the visitors who could have taken them. Compare it across pages, because that tells you where to work. A page with plenty of visitors and almost no actions is a better candidate for a rewrite than a page nobody sees. If you suspect a page is losing people but can’t see where, how to find conversion friction on your website walks through the places to check.
If you use affiliate links, count clicks as a measure of interest, and treat commission as a separate matter. A click shows that the recommendation was relevant to someone. It doesn’t show that the tool suited them, which is one reason to read what readers write back as well as what the dashboard says.
Is any of this paying for itself?
This is the question small brands most want answered and the one the data answers least cleanly. Connecting content to revenue is difficult even for large companies, and a small brand has fewer tools and fewer data points. Pretending otherwise leads to confident wrong conclusions.
Start with the cost side, which you control. Add up what content costs you each month: subscriptions, tools, any freelancers, and your own hours at a rate you assign yourself. People often skip the hours, which makes everything look cheaper than it is.
On the revenue side, use imperfect but honest signals:
- Ask new customers or enquirers how they found you, with a single optional field on the form. Answers are fuzzy, but they surface paths analytics can’t see, such as someone who read three articles and then searched your name.
- Note which pages a visitor viewed before converting, where your tool allows it.
- Track revenue from leads that came through content, even if you can’t prove content caused the sale.
None of this is clean attribution. It gives you a rough, defensible picture that is good enough for the actual decision: keep investing at this level, scale back, or change the approach. A worked example of framing content spend against outcomes is in the AI content ROI case study, though this scorecard applies to any kind of content.
What small numbers can’t tell you: sample size, noise and how to stay honest with your own data
Small audiences produce noisy numbers, and noise looks a lot like insight. A few honest limits are worth keeping in mind.
Rates built on small counts are wide guesses. Suppose 3 of 100 visitors to a page send an enquiry. That’s 3%, but a standard 95% confidence interval for that result runs from roughly 1% to 8%. If next month the same page converts at 5%, you haven’t necessarily improved anything. The result sits comfortably inside the range of what chance alone could produce.
Counts bounce around too. When you’re counting something that happens about ten times a month, a swing of three either way is ordinary random variation. A fall from ten to seven is not evidence that anything broke.
Short windows mislead. One strong week usually reflects a single post or a single share. Look at three months before calling something a trend, and look at the direction rather than any single point.
Don’t run A/B tests on a few hundred visitors. You won’t gather enough data to separate a real difference from chance. Make one change at a time, note the date, and watch the trend over a longer period. If you change the headline, the layout and the offer in the same week, you won’t know which mattered.
Write down the count behind every percentage. “4% conversion” means one thing from 25 visitors and another from 2,500. Keeping the raw number next to the rate stops you from over-reading it.
Small numbers still count. Qualitative evidence, like replies to your emails, questions in your inbox, and comments that quote your own words back to you, can outweigh a month of statistics at this scale. Use the data to narrow your guesses, not to claim proof. This scorecard shows correlation and direction. It won’t tell you what caused what.
Your one-page scorecard: five questions, one metric each, reviewed monthly
The fifth question is the one that keeps the other four honest: can we believe what this month’s numbers say? Its metric is simply the raw count behind each figure, kept beside it.
Question One metric Decision it informs Are the right people finding us? Share of visits from your target audience, by channel or location Which channel or topic gets your next hour Do they trust us enough to stay? Returning visitor share Which content series to extend or retire Are they taking the next step? Rate of one defined action (sign-up, enquiry or click) on your key pages Which page to rewrite first Is any of this paying for itself? Revenue from content-sourced leads against monthly content cost Whether to invest more, hold steady or scale back Can we believe this month’s numbers? The raw count behind each rate above How much weight to give any change
Review it once a month, at a fixed time, for about thirty minutes. For each row, write the number, the previous month’s number, and one sentence on what you will do about it. If you can’t write that sentence, the metric hasn’t earned its place yet, so replace it or move on.
Resist the urge to add rows. Each new metric splits your attention, and the review turns back into a dashboard tour. When a decision comes up that none of the five rows covers, add one metric for it and say what decision it serves.
Pick one question from the scorecard and track its metric for 30 days before adding anything else. If you’d like a second pair of eyes on what your numbers are really saying, get in touch about a conversion review.