Measure, iterate, and know when to quit or double down

Most sites die of not looking

Not from bad content, not from competition. From the founder never opening the numbers, or opening all of them and seeing nothing. A site that is measured badly and a site that is not measured at all end the same way: with a quiet decision to stop, made months after the data would have said either "stop" or "this part is working, do more of it".

This step sets up the smallest measurement that works, defines the three numbers that matter, and gives you a fixed calendar for looking at them and a rule for deciding.

The tools, and which to pick

Google Search Console is not optional. It is free, it shows what Google does with your pages, and nothing else can. If you set up one tool, it is this one.

For visitor analytics, pick one of two:

Google Analytics 4 is free, deep, and more than a solo site needs. It answers almost any question if you learn its reporting model, which takes a weekend. It requires a cookie consent banner in the EU, and its default data retention is short (two months for event data unless you change it to fourteen). Use it if you will run ads, sell products with a checkout funnel, or want to see exactly which article led to which sign-up.

Plausible is a one-script, cookie-free analytics tool that shows visitors, sources, pages and goals on a single screen. It costs $9 a month for up to 10,000 monthly pageviews, scaling with traffic, after a 30-day trial; check the current page. It needs no consent banner in most readings of EU law, it loads in a millisecond, and you will actually look at it because it takes ten seconds to read. We use it for exactly that reason.

Decision: Plausible if you want to look at your numbers weekly and get on with the work; GA4 if you need funnels and are willing to learn it. Either one plus Search Console. Never both analytics tools; you will trust neither.

Whichever you choose, set up three goals on day one: newsletter sign-up, outbound click on an affiliate or product link, and contact form or purchase. A goal turns a visit into a result.

The three numbers that matter

Traffic is a vanity number. It goes up when something goes viral and down when Google sneezes, and neither tells you whether the site is working. These three do:

One: subscribers gained per week

Not total subscribers; the weekly change. It combines traffic and trust into one number: people came, and enough of them believed you to hand over an email. If this number is flat while traffic grows, the promise on the form is wrong, or the traffic is the wrong people. If it grows while traffic is flat, you are getting better and the rest will follow.

Two: money per thousand visitors

Take everything the site earned this month (affiliate, ads, sponsorships, products, services attributed to the site) and divide by monthly visitors in thousands. This one number tells you whether growth is worth chasing. A site earning $2 per thousand visitors needs a million visitors to matter. A site earning $80 per thousand can be small and still work. It also shows, instantly, whether a change you made (adding ads, removing ads, a new product) made the site better or just busier.

Three: return rate

The share of visitors who have been before, from the "returning visitors" report in your analytics tool or, better, the number of people who typed your address or came from the newsletter. Search sends strangers; a site with no return visitors is an appendage of a search engine and will fall when the algorithm shifts. Return rate is the only number that measures whether you are building something people want to come back to, which is the only thing that lasts.

Write these three on a card. Check them on the same day each week. Ignore everything else unless one of them moves.

The 30/60/90 review

A fixed calendar stops the two failure modes: never reviewing, and panicking over a bad Tuesday.

At 30 days: everything is small and that is fine. The only question is whether the machine works. Is each page indexed? Do goals fire? Did the newsletter send on its day? Fix mechanics; do not judge results.

At 60 days: look for the first signal. Which two or three pages get most of the search impressions in Search Console? Which one article produced the most sign-ups? That is your reader telling you what to write next. Write three more pieces around the winner before you write anything else. Most sites spread themselves across ten topics at this point; the ones that grow narrow to two.

At 90 days: the first honest verdict. Compare the three numbers to day 30. All three up: double down, which means the same things, more often, and cutting whatever did not move a number. Two up: keep going and fix the third. One or none up after three months of real weekly work: the problem is the idea or the audience, not the effort, and the next section applies.

Then repeat every 90 days. A year is four verdicts.

The twice test

Here is the question we ask of every project, and it is harder than any dashboard: would someone pay for this twice?

Not once. Anyone can be curious once. A first purchase, a first subscription, a first click is proof of interest. A second one, a renewal, a repeat order, a reader who stays on the paid tier after month one, is proof of value. It is the difference between selling a thing and having a business.

Apply it to the site as a whole and to each thing it sells:

• A newsletter passes when readers stay subscribed past the third issue and reply.
• A product passes when buyers of the first thing buy the second.
• A service passes when a client comes back or sends a friend.
• The site passes when the return rate rises without you doing anything new.

If nothing you make passes the twice test after two 90-day cycles, quitting is the correct decision, not a failure. You will have learned what your readers want from the pages that did work, and the next site starts with that knowledge and a list. Many successful creators are on their second or third project. The first one taught them what to measure.

If one thing passes, that is the business. Double down means cutting everything that does not feed it, even things you like. A site that does one thing people pay for twice beats a site that does ten things they enjoy once.

Checklist

• Search Console plus one analytics tool live, with three goals tracking
• The three numbers written down and a weekly review day chosen
• 30-day mechanics check, 60-day winner identified, 90-day verdict scheduled
• Content narrowed toward the pages that produced sign-ups
• Every product, service and the site itself scored against the twice test