"We need more traffic" is a wish, not a task. It becomes useful once you take it apart. For visitors from search, the arithmetic looks like this:
Clicks equals search demand times visibility times click-through rate.
Three factors, each costing a very different amount to move. Most sites start with the expensive one.
Lever 1: positions four through fifteen
By a wide margin the cheapest lever, and the one used least.
Click-through rate does not decline smoothly down the results page, it falls off a cliff. Roughly, and depending on query type: position one takes around 27 percent of clicks, position three about 11 percent, position eight roughly 3 percent, position fifteen about 1 percent. The exact figures move around. The shape of the curve does not.
What follows from that is unusually concrete. A page sitting at position eight for its main query with a thousand impressions a month collects around thirty clicks. The same page at position four collects around a hundred. Four positions triples the visitors without a single new page being written.
How to find those pages: open Search Console, set the range to three months, filter average position between 4 and 15, sort by impressions descending. The first five rows are your quarter.
These pages have already proven that Google considers them relevant. They no longer need the benefit of the doubt. Usually they just need a better answer to the question actually being asked. How to find out which question actually lands on a page is covered in analyzing user behavior.
What you actually do to a page like that, in this order:
- Click the page in Search Console and switch to the queries tab. It lists every search term this one page appears for. The five with the most impressions are the yardstick for the text, rather than the keyword you had in mind while writing it.
- Check whether the strongest query appears in the title and the main heading, in the searcher's own words. Someone searching "make website faster" does not recognise themselves in "performance optimisation".
- Pull the answer to the top. The first paragraph under the heading answers the question directly; the reasoning comes after.
- For every query on the list the text does not answer, add a section of its own with a subheading. That is usually the biggest gain, because Google is already looking for an answer on your site there.
- Link to the reworked page from two or three of your strongest pages, with anchor text that names the query. "Click here" throws the signal away.
- Check whether a second page shows up for the same query. If it does, the two compete, and usually neither wins. Make one the main page; the other points to it or gets merged into it.
The full routine with thresholds and examples is in striking distance keywords. If the page has plenty of impressions but strikingly few clicks, the problem often sits one step earlier; impressions but no clicks untangles that. And if the page is simply old, follow updating existing content: refresh numbers, examples and screenshots instead of writing a new article on the same topic.
Lever 2: the click-through rate of the result itself
Same position, different title, more clicks. Sounds like fiddling at the edges. It is not, arithmetically.
A page with 20,000 impressions a month at a 2.4 percent click-through rate brings 480 visitors. Lift that to 3.8 percent and it brings 760. That is 280 extra visitors a month with the ranking completely unchanged.
What holds up in testing: a number in the title, a year on anything that could look outdated, and a promise concrete enough to sound checkable. What does not hold up: superlatives and exclamation marks.
One limit people forget. On pages already sitting at position one or two with a healthy click-through rate, leave the title alone. There is more to lose there than to gain.
How to go about it: sort pages by impressions in Search Console and hold each row's click-through rate against the curve above. A page at position three with a 4 percent click-through rate collects far less than its spot allows. For exactly those pages, write a new title, note the date, and compare the click-through rate after four weeks at a similar position. Change one thing per page at a time, otherwise you will not know afterwards what worked. How to build a title that gets clicked, and when Google rewrites it anyway, is in title and meta description.
Lever 3: covering new demand
The most expensive lever, which is why it sits third rather than first.
A new page starts with no history. Google typically needs three to six months to form a view of whether it is any good, and the outcome is not guaranteed. Despite that, it is the first reflex almost everywhere: another article, another landing page.
The right moment for new pages is when the existing ones have run out of room. And the good starting point is not a keyword list, it is the questions your current pages raise without answering. Those sit in Search Console among the queries with plenty of impressions and a dismal position: Google is already trying to show you for something you have no proper page for.
Concretely: filter Search Console to position greater than 20 and at least 50 impressions over three months. For each query, check which of your pages shows up. If it roughly fits, that is a job for lever 1 and a new section. If it does not fit at all, you have found a topic for a new page, with the advantage that demand is already proven. If you sell locally, the cheapest new demand often hides in queries containing place names; what counts there is in getting found locally.
Lever 4: being visible in AI answers
A growing share of searches ends in an answer rather than a list of links. ChatGPT, Perplexity and Google's AI overviews cite sources, and being cited brings visitors.
After watching this for a while, what these systems favor comes down to three things: pages that answer a question inside one self-contained paragraph, a clear structure, and being mentioned elsewhere on the web. A paragraph that makes sense without the rest of the page around it gets quoted far more often than the same content spread across four sections.
Volume per citation is small. Quality is unusually high, because these people have half-decided before they arrive.
The test costs two minutes. Ask one of these systems the question you want to rank for and see who gets named. If the same three competitors show up every time, you know which pages you are being measured against. What to change about your writing so that you are one of the named sources is in getting found in ChatGPT. The cheapest technical addition alongside it is an llms.txt: half an hour of work that shows these systems which of your pages hold the answers.
How Google assembles its own AI answers, and why different pages get named there than in the results underneath, is explained in Google AI overviews. Search Console tells you little about whether the work brings visitors: it does not separate clicks from Google's AI overviews, and it knows nothing about visitors from ChatGPT or Perplexity. In GA4 you can count them with a dedicated channel group; the walkthrough is in measuring AI traffic in GA4.
The fifth lever hardly anyone thinks of: how the result looks
Between visibility and a click sits more than the title; it is the whole appearance of your result in the listing. Two things you can influence and most people leave alone.
The first is the meta description, the grey text under the title. Google does not always use it, but when it does, it is your free advert. A description that picks up the question and promises a concrete benefit pulls measurably more clicks at the same position than a truncated sentence lifted from the first paragraph. The second is structured data: review stars, prices, availability that show up as an extra line in the result. They do not change your position, but they make your result bigger and more noticeable than the competition sitting in the same spot. Both are fine detail, but they work on every page that already ranks, without you gaining a single position.
The mistake that wastes the most time
Traffic without matching intent. A post that catches on brings twelve thousand visitors in a day, zero enquiries, and a beautiful curve in the report. Then it goes quiet again.
Before starting a page, answer one question: would the person searching for exactly this ever buy from you? If not, the page is not a mistake, but it does not belong on the list while more urgent things are open. How intent shapes what happens once somebody lands is covered in understanding bounce rate.
Where to start
Search Console, last three months, position between 4 and 15, sorted by impressions, write down the first five rows. Open the page behind each row and check whether it really answers the query. What is usually missing is a concrete number, a worked example, or a list somebody can skim.
It is unglamorous and it works faster than any new article. If you discover along the way that one of those pages is not indexed cleanly in the first place, the work starts one level earlier, at finding SEO problems. If an important page does not appear in Search Console at all, see page not indexed. And if you do not have Search Console yet: setting it up takes ten minutes and is covered in setting up Search Console. Without it, this whole article is guesswork.
One month, week by week
To turn the levers into a plan, here is a split that fits alongside day-to-day work:
- Week one: pick the five pages from the 4 to 15 position band and note the query list for each. No changes yet, just stocktaking.
- Week two: rework the two pages with the most impressions using the six steps from lever 1.
- Week three: give the three pages with the biggest gap between position and click-through rate new titles and descriptions.
- Week four: rework the remaining three pages from week one and add internal links from your strongest pages.
Every change gets a date, in a spreadsheet or wherever you will find it again. Rankings take two to six weeks to move, and without a date you cannot tie a movement to a change later. If a page drops sharply after a change, the diagnosis in lost rankings helps.
How Traffalyzer helps
Traffalyzer takes over the part that sounds like spreadsheet work. The free scan below checks your site's technical foundation, meaning everything that holds a lever back before it can work: indexability, titles and descriptions, load time, structured data.
With Search Console connected, Traffalyzer collects the pages from the 4 to 15 band as striking distance keywords, shows pages whose click-through rate sits below what their position would suggest, and flags when two of your pages compete for the same query. You log your changes as measures in the history, and Traffalyzer compares the four weeks before with the four weeks after. Which page needs the better answer, and what that answer looks like, is still your call.
Frequently asked questions
How long does it take for traffic to increase?
For pages that already rank, first movement after two to six weeks is typical, depending on how often Google recrawls the page. New pages usually need three to six months before their position settles. That is why starting with existing pages pays off.
How do I increase website traffic for free?
The two cheapest levers cost only time: rework pages in positions 4 to 15 and improve the titles of pages whose click-through rate lags their position. Both work with the free Google Search Console. Ads bring visitors immediately, but they also stop the moment the budget does.
Do more blog posts bring more traffic?
Only if they cover demand you have no fitting page for yet. Another article on a topic an existing page already half covers often competes with that page, and both get stuck. Exhaust the existing pages first, then write new ones.