A teacher at an IELTS prep centre in Almaty drops a Search Console screenshot into the team chat: clicks from Google have tripled over the summer, from 310 in June to 920 in September. In the spring she wrote an article on irregular verbs for the website, and now it brings in more people than every other page put together. The owner opens the CRM and sees the same number of leads from the site as before. "So SEO doesn't work. Shall we stop writing articles and put the money back into ads?" The centre's numbers in this lesson are made up.
They are arguing because they are both looking at one total. It mixes people searching for IELTS courses with adults who need a verb table for work or travel. Until you separate them, you can't say SEO is growing or that it has failed. Below we cover how a search engine finds pages, which groups to split search traffic into, and where to start SEO on a site where nothing has been set up yet.
How a page gets into search
Google works in three stages. First a crawler finds the page and downloads it; that is crawling. Then the system processes the text and decides whether to keep it in its index. Only a page in the index can show up in results for someone's query. Google doesn't promise to crawl a page, index it or show it in results, even if the site meets all its requirements, and it doesn't accept money to crawl more often or rank higher (how Google Search works).
That draws the line around your job. You are responsible for the crawler being able to open the page, for nothing blocking it from the index, and for the page answering the query. The search engine decides the position. A contractor who guarantees the top spot is promising something they don't control, and in its guidance on choosing an SEO Google advises finding someone else in that case.
Kazakhstan has two search engines that matter. According to StatCounter for September 2026, Google accounts for 68% of search referrals and Yandex for about 30% (search engine shares in Kazakhstan). So you work with both: Google Search Console and Yandex Webmaster. Search Console shows which queries the site appeared for in Google, how many clicks it got and which pages didn't make it into the index. Webmaster has its own list of pages in search and pages excluded from it.
The quickest way to verify the site in Search Console is through a Google Analytics tag or Google Tag Manager container that is already installed, as long as you sign in with the same Google account. That verifies a URL-prefix property. A domain property, which covers http, https and every subdomain at once, can only be verified with a DNS record (verification methods).
Split the traffic by task
The centre's marketer opens the Performance report in Search Console, goes to the Pages tab and looks at which queries bring people to each page. Leads come from GA4: the Engagement > Landing page report, filtered to sessions from the Organic Search channel, with the Key events column, where the centre has a single event, generate_lead.
| Page | Queries that bring people in | Google clicks, June | Google clicks, September | Search leads, September |
|---|---|---|---|---|
| Home page | the centre's name | 130 | 160 | 5 |
| Article "Irregular verbs: table and examples" | general English grammar | 120 | 680 | 0 |
| Article "IELTS Academic or General Training" | questions about the exam | 60 | 80 | 2 |
| Course page | not a single impression | 0 | 0 | 0 |
| Total | 310 | 920 | 7 |
The clicks here are Google only, while GA4 counts leads across all organic search, Yandex included. That's good enough for the decision.
Nearly all the growth, 560 of the 610 extra clicks, came from the verb article. The queries show who reads it: "irregular verbs table", "present perfect rule". These are people who need English in general, not the exam. When the centre sorted 120 leads, the "need English, not IELTS" group produced 30 leads and not one payment. The article brought no leads at all in September, and judging by the queries, most of the new clicks came from people the centre has nothing to sell to.
Search leads were 6 in June and 7 in September. The difference is a single lead, and it tells you neither that search is growing nor that it has stopped working. The leads come from the home page, which people reach when they already know the centre's name, and from the article on choosing between Academic and General Training.
The course page row matters more than all the others. People search for "IELTS courses Almaty", the centre pays Google for ad clicks on that very query, and the course page doesn't appear in search at all. Until it gets into the index, no article will lead anyone to the place where they can sign up. What is blocking it comes out in a technical check of the site, and that is where to start.
So the owner and the teacher get this answer. There is nothing to celebrate in the total clicks: the growth came from a topic the centre earns nothing from. Giving up on search is premature as well, since the page that sells has never once taken part in it. There's no reason to delete the verb article, and no reason to write ten more like it.
The picture shows the path the centre wants from its exam articles: a person gets an answer to their question and follows a link to the page where they can sign up.
What counts as a result
An SEO result is pieced together from three sources, and each answers its own question:
| Level | Where to look | What it answers |
|---|---|---|
| Impressions and clicks | Search Console, Yandex Webmaster | Whether the page appears and for which queries |
| Sessions and leads | GA4, Landing page, Organic Search channel | Which landing page the people who submitted the form came in on |
| Payments | CRM | Which of those people became students |
Search Console clicks and GA4 sessions will never match exactly. Search Console counts clicks on Google results, while GA4 only sees visitors whose tag fired on the site. There's no point reconciling them line by line; it's enough that they move in the same direction.
Payments are harder. If the CRM only receives UTM tags with each lead, a lead from search lands in the "no tags" row together with direct visits, because a click from search carries no tags. To tell organic apart, the site passes the channel or the landing page to the CRM along with the lead. How you set that up depends on the form and the CRM, but linking a lead to its source is the same thing end-to-end analytics rests on.
How long to wait and what it costs
Google says some changes show up in search within a few hours and others take several months, and it advises waiting at least a few weeks before judging the result (Google's starter guide). Timelines like "results in six months" are practitioners' estimates, not a search engine rule, and you can't promise them to a client as fact.
SEO only looks free because nobody charges you for a click from search. The work costs money. The centre needs a developer to remove whatever is keeping the course page out of the index and to fix the old addresses. Going by the technical check, that's about four hours, or 20,000 ₸ at 5,000 ₸ an hour. The marketer needs a working day to rewrite the course page for the query. The centre keeps 48,000 ₸ per student over three months, after paying the teacher and materials, so the developer's work pays off if search brings one student a quarter.
Whether it will, nobody knows in advance. So the plan has a check date: four weeks after the course page gets into the index, the marketer looks at which queries it shows up for and in what position. Leads get added up over the quarter, because there are too few per month.
Where to start on a new project
The order for a site where nobody has done any SEO yet:
- Run the technical audit from its checklist: connecting Search Console and Webmaster, the pages that sell, and whatever keeps them out of the index. It comes first because a page outside the index won't get a single impression.
- Bring the selling pages up to standard: the title, description and text are built around the query people arrive with.
- Build a semantic core: which queries people search for in the niche and which page answers each one.
- Write articles on the questions customers ask before buying. Take them from the CRM and WhatsApp chats, the way the centre took "Academic or General Training".
- Once a month put together a table by page, like the one above, and show the client that rather than an overall click chart.
Exercise: search traffic by task
Take the website of your own project or a client's, one you have Search Console access to.
If most clicks land on one or two pages with an unrelated topic, the overall chart grows without helping sales. If clicks come mostly from the business's name, search works as a directory for people who already know the company, and other channels bring in the new people. With five to seven search leads a month, months can't be compared by leads, because the difference rests on one or two. Then the three sentences for the client cover impressions and clicks for the selling pages, leads over the quarter, and no total click count without a breakdown.
Video: SEO basics
An introductory overview for a first look at the topic.