From an audit to the fight for positions in AI results

ICProject is exactly the type of project in which SEO makes the most sense: a mature Polish product with a strong brand foundation and real, yet still untapped, potential in search. The client’s team knows its product inside out and speaks the same language as we do — so decisions are made quickly and rest on data, not on guesswork.

We do not promise miracles or “rankings in 30 days”. We start with an honest, hard diagnosis — we show what already works well, where the biggest challenges are and where the fastest lever for growth lies. This document is a record of that start: strengths, market context, real needs and the action plan agreed together with the ICProject team.

Let us start with what matters most: ICProject is a truly remarkable product — a mature, polished project management system that competes with the world’s biggest platforms without any complexes. What is more, it is Polish — and that is not a curiosity but a real asset. It understands the language, the realities and the way Polish teams work better than global tools do, and in SEO and in building customer trust that local relevance is an advantage no budget can buy. It is a great foundation to build on.

At the same time ICProject operates in one of the most crowded markets on the web. On the other side stand brands such as Asana, Atlassian (Jira), ClickUp and Monday — global companies with budgets for content and links that a Polish producer will not beat “on money”. And here comes the first challenge: this is not a race of “who is bigger”. If we took on the giants with their own weapons, we would lose at the start.

That is why we play for something narrower and far more valuable: specific, buying keywords on the Polish market, where a local, Polish-language and well-optimised site can beat a global colossus — because it understands the Polish context, language and user intent better.

Three areas with the greatest lever for growth

The good news is that these are not “problems that cannot be moved” but clearly countable needs — and each of them is a ready opportunity for growth:

• Quick technical improvements. 23 service pages will gain from simply adding H1 headings, and 32 existing pages have dormant potential ready to be recovered. It is the cheapest, fastest visibility gain in the whole plan.
• Order in the site structure. In places several subpages target the same keyword — putting this in order (one keyword = one page) means the site’s strength stops scattering and starts accumulating.
• Presence in generative search. Some users today get an answer straight from AI — and here ICProject already has a strong starting position that we want to expand (more on this in the AI Overview section).

Google does not “set” the results once and leave them alone. On top of the base ranking it applies a layer of reranking — functions that constantly rearrange the order based on how users actually behave. In testimony in the Google antitrust trial (DOJ, 2023), Google itself confirmed the existence of the Navboost system, which uses the history of clicks and their quality (so-called “good clicks”) to adjust results. In turn, the leak of the internal Content Warehouse documentation (2024) — widely analysed by the industry, though unofficial — describes dozens of signals and a layer of “twiddlers”, mechanisms that rerank the results just before they are shown. This direction is consistent with Google’s long line of patents on modifying ranking based on implicit user feedback. These are not our guesses — they are publicly documented facts on which we base the strategy.

The consequence is fundamental: it is not enough to get high once — you have to earn and maintain a positive trend. Reranking rewards sites whose signals (click-through rate, engagement, satisfaction with the result) are growing and penalises those that are weakening. That is why we look not at a single point in time but at the trend.

More and more often, when someone asks Google about “project management software”, they no longer see ten blue links at the top — just a ready answer generated by AI (the so-called AI Overview). Google stitches it together from fragments of a few trusted sources and cites them at the top. Assistants such as ChatGPT, Perplexity and Gemini do the same. The effect: being cited in an AI answer is becoming the new “position number 1”, and the user very often ends the search without a single click (so-called zero-click).

For a brand this means one simple thing: if you are not a source from which AI builds its answer, you practically disappear from this part of the market — even if you have decent positions in the classic results.

And here we have very good news: according to Senuto data, ICProject is already present in AI Overview for 586 keywords, with an average citation position of 5.7. This is not a hypothesis — the brand really does appear in AI answers to questions about project management. (Presence in AI Overview is variable by nature — it depends on the query, the location and the moment of measurement; the figures given are a snapshot from the tool, not a guaranteed value.) Our task is to widen this presence and move it up, to citations in the TOP 1–3, because that is where the user’s attention goes.

Where the greatest lever is

We broke the AI presence down by stage of the purchase decision (the funnel), using keyword intent classification. It is an approximation by nature, but the direction is unambiguous and strategic:

• Top of the funnel (educational content) — 87% of the presence. Here ICProject is strong: “what project management involves”, “project charter”, “resource allocation”. This builds recognition but rarely sells straight away.
• Bottom of the funnel (purchase keywords) — 0.5% of the presence. Where people actually choose a tool and compare options, the brand is practically absent. And this is our most important, most profitable opportunity.

The search engine stopped “counting words” long ago. Google (thanks to models such as BERT and MUM), and AI engines even more so, understand content through meaning, not through matching letters. They look at entities — concepts, brands, people, products — and at the relationships and context between them. To the algorithm, “project software”, “task management tool” and “work planning app” are largely the same semantic field, even though they share no word.

The conclusion is practical and has real consequences for content:

• Keyword stuffing no longer works — and can even do harm. What works is complete, clear topic coverage: an answer to every sensible question a user has around a given subject.
• Topical authority counts. A site that describes its area completely and coherently is a more credible source for Google and AI than one with a single “over-optimised” page.
• One concept = one place. If three subpages try to serve the same meaning, they dilute the signal and compete with one another.

The most common hidden brake on visibility is cannibalisation — several subpages compete with one another for the same keyword, so Google does not know which one to show and often shows none of them high up. The solution is not spectacular, but it is effective: we mapped the whole structure and assigned 1,625 meaningful keywords to 88 pages — each keyword to exactly one page. We based the assignment on the semantics described above (embeddings), not on intuition.

What exactly we are putting in order

• A 1:1 keyword map. Every existing page gets a clear goal and an assigned set of keywords — the end of internal competition.
• 32 “dormant” pages. They exist but rank for 0 keywords from their cluster. Once the content and linking are put in order, this is potential ready to be recovered — without building new addresses.
• 32 new subpages. Where the competition has content and we have gaps (271 content-gap keywords), we add the missing pages — with a ready brief, not “just in case”.
• Internal linking. We tie the clusters together so that authority flows to the pillar pages and the user (and the crawler) moves through the topic intuitively.

We ran ICProject through a checklist of 42 E-E-A-T points. And it is encouraging: the brand foundation is strong, and that is the hardest part to build from scratch. Some signals need refining and a few simply do not exist — although they are cheap to implement and highly scored at the same time. These are classic “quick wins”.

This is not cosmetics. The lack of an “About us” page or of a physical company address is marked in the quality rater guidelines as a signal of very high priority. Filling these gaps raises the credibility of the whole domain — and with it the chance that it is ICProject (and not a competitor) that gets cited in an AI answer.

For each of the 88 pages we prepared a specific proposal for the title / H1 / meta description and a list of primary and secondary keywords. Instead of generalities such as “write something about projects”, the copywriter gets a ready plan: what the page should be about, which questions it answers and how not to get in the way of other subpages. This shortens content production and eliminates randomness.

These are keywords with a clear, often purchase intent — and a good share of them also appear in AI Overview. This means every page won works for the classic results and for AI citations at the same time. One well-made text, two visibility fronts. Examples of priority keywords (Polish market):

• ms project alternatywa
• program do prowadzenia projektów
• aplikacja do planowania projektów
• karta projektu
• oś czasu projektu
• alokacja zasobów
• harmonogram realizacji
• narzędzie do zarządzania zadaniami