SEO strategy consulting that measures all three visibility channels, then hands you a prioritised 90-day plan your team can actually run.
Every logo is a business we can name and a relationship we can describe. Search happens to be one thread of it.
A strategy engagement produces the diagnosis, the map and the specification. Writing the articles that fill it is content marketing work, bought separately.
What you get
Search visibility split into three channels and most programmes still measure one. A buyer can reach you through a classic Google result, through an AI assistant that cites you without sending a click, or through whichever language version matches the one they searched in.
Still the largest source of qualified visits for most businesses, and still governed by the same fundamentals: crawlable, fast, relevant, linked to. What changed is that a good position no longer guarantees the click.
Assistants answer directly and cite a handful of sources. Being cited is not the same as ranking. It depends on whether the crawler can reach the page, whether the answer survives being lifted out of context, and whether the claim carries a source.
Access is where most sites fail without knowing. Each engine uses a named crawler controlled through robots.txt and, separately, by whatever sits at your CDN or firewall. A site can allow every bot in the file and still be blocked at the edge.
Where an audience searches in more than one language, the versions almost never perform alike, and rarely in the direction anyone expects. Two mechanical faults cause most of the damage: hreflang that contradicts itself, and language versions chasing the same queries as each other.
Prima publishes in English, Arabic and German. Our own three versions do not perform alike either, which is how we came to measure this.
| Engine | Crawler | What blocking it costs you |
|---|---|---|
| ChatGPT | GPTBot, OAI-SearchBot | No citation in ChatGPT answers or its search results |
| Perplexity | PerplexityBot | Absent from a service used heavily for supplier research |
| Claude | ClaudeBot, Claude-SearchBot | No citation in Claude answers |
| Google AI Overviews | Google-Extended | Excluded from AI summaries while still ranking normally |
| Crawler names per each provider's published documentation. Check yourdomain.com/robots.txt, then test again from outside your network, because edge rules never appear in that file. | ||
What teams need: target terms selected on how close the searcher is to buying, not on impression counts.
How Prima solves it: the keyword map scores commercial intent separately from traffic and drops the rest.
What teams need: a ruling on which page keeps which query.
How we solve it: every overlap is registered, then resolved as consolidate, redirect or differentiate.
What teams need: to know whether the problem is access, structure or sourcing.
How we solve it: crawler access is tested from outside your network, not just read from the file.
What teams need: each version measured on its own before budget is committed.
How we solve it: a like-for-like comparison of every language version, plus an hreflang integrity check.
What teams need: a recommendation, not a chart pack.
How we solve it: every review ends by naming one thing to start and one to stop.
What teams need: technical debt quantified and ordered by effect.
How we solve it: the audit measures against Google's published Core Web Vitals thresholds and writes developer-ready specifications.
Most engagements begin with the first track. The other two are available when the diagnosis warrants them, and neither is a condition of the first.
| Feature | What It Means for Your Operations |
|---|---|
| Technical audit against published thresholds | You get a measured list of faults with owners attached, rather than a score you cannot act on. |
| Keyword and category mapping | One primary term per URL, so your pages stop bidding against each other for the same result. |
| Cannibalization register with rulings | Each conflict comes with a decision, which is the part that takes judgement and the part most audits omit. |
| AI crawler access testing | You learn whether assistants can physically reach your pages, tested at the edge rather than assumed. |
| Language-version comparison | Translation budget goes to the version that moves, which is frequently not the one being funded. |
| Sequenced 90-day roadmap | Work is ordered by expected effect, so the first month produces something visible. |
| Feature | What It Means for Your Operations |
|---|---|
| Developer-ready technical specifications | Your engineers, or our website development team, implement from the document instead of booking a call to interpret it. |
| Schema specified in JSON-LD | Structured data ships correctly first time rather than being validated after release. |
| Migration oversight | A rebuild stops being the event that costs you two years of accumulated position. |
| Feature | What It Means for Your Operations |
|---|---|
| Position and citation tracking | You see movement across classic results and AI answers together, rather than only the half that is easy to measure. |
| Crawler re-testing after infrastructure change | A new firewall rule stops being able to remove you from AI answers silently. |
| Review that ends on a commitment | Each month closes with an agreed change, which is how the retainer survives an internal budget challenge. |
| Step 1: Crawl |
We take a full technical picture and test every AI crawler from outside your network. | OUTPUT: A fault register with severity and owner against each item. |
| >>> | ||
| Step 2: Map |
Every URL is matched to one primary term and one topic category. | OUTPUT: The keyword map and the cannibalization register. |
| >>> | ||
| Step 3: Prioritise |
Findings are ordered by expected effect against effort, with the reasoning shown. | OUTPUT: A sequenced 90-day roadmap. |
| >>> | ||
| Step 4: Ship |
Changes go out in order, each with a named owner and a written specification. | OUTPUT: Implemented fixes plus the specs behind them. |
| >>> | ||
| Step 5: Track |
Positions, AI citations and enquiries reviewed against what we predicted. | OUTPUT: A monthly report ending in a recommendation. |
Where a prediction turns out wrong, we say so in step five rather than quietly re-baselining. At each stage you know what has been completed and what comes next.
Karachi, Lahore and Islamabad, including firms ranking simultaneously at home and in export markets.
Dubai, Abu Dhabi and Sharjah. Paid competition is fierce here, which raises the value of organic and AI-answer visibility rather than lowering it.
Publish in a language not listed and the method is unchanged. Each version is assessed on its own performance.
Riyadh, Jeddah and the Eastern Province, where search runs across Arabic and English inside one buying journey and the Arabic side is usually the underworked one. Recommendations are shaped around PDPL expectations.
European entities whose German and English sites usually need separating properly.
Sourced claims and a named reviewer are required anyway. Answer engines reward the same discipline.
Intake calendars compress demand into short windows, and the roadmap works backwards from them.
Nothing goes live until compliance clears it, so the first version is already built to get through them.
Documentation frequently outranks the marketing site. Better planned deliberately than discovered by accident.
Category structure carries the ranking weight, and duplication is almost always the fault.
Small query volumes and long consideration. Ten right visits beat a thousand wrong ones, so intent scoring carries the work.
Buyers search by part number and specification, so category mapping does most of the lifting.
Highly local and seasonal. Location pages have to be genuinely distinct rather than templated.
Across a six-page audit of one Gulf professional-services group in September 2026, the Arabic version of every page outranked its English equivalent, the widest gap running to 24.7 positions on the same topic in the same month. The organisation had assumed English was the stronger side and had invested accordingly.
verified Confirmed against the same dataset. The Arabic IAS 36 page holds position 7.11 with 314 clicks, ahead of most English service pages on the domain.
In one sixteen-month dataset, twenty question-format queries ranked inside the top ten and produced zero clicks between them. Two of those sat at position 1.6 and position 2.0. The pages were winning the ranking and losing the click, which is a title and snippet problem rather than a ranking problem, and no monthly report had ever separated the two.
verified Questions and AEO export, sixteen months to 10 August 2026. 128 question-format queries, 11,011 impressions, 19 clicks.
Sixty-six queries carrying more than 1,500 impressions each produced 182,314 impressions and twenty clicks between them across sixteen months. Two of those queries ranked on page one. The organisation had been reporting impression growth as progress for a year.
verified Search Console export to 31 August 2026, queries with 1,500+ impressions and one click or fewer.
One professional-services group had its best-ranking hub page sitting at position 5.8 and converting 9,230 impressions into 38 clicks, a rate of 0.41%. Four levels below it, pages nobody had linked to were carrying 405,782 impressions. Authority and demand were sitting on different pages, and the internal linking had never been mapped.
verified Page-level export to 31 August 2026, cross-referenced against 110 service URLs in the live sitemap.
Search authority compounds both ways. A competitor publishing consistently for twelve months does not finish slightly ahead. They finish holding positions that now cost multiples to recover.
The AI channel compounds faster. Assistants learn which sources they cite for a topic and keep returning to them, and the buyer never sees the suppliers who were not mentioned.
The quietest cost is the diagnosis nobody has run. Cannibalization, a blocked crawler and an unmeasured language version are all invisible from the inside, and all three keep charging you until somebody looks.
SEO strategy is not an administrative cost. It is the infrastructure that determines whether buyers already searching for what you sell can find you at all.
Most diagnostics surface at least one thing nobody in the business knew: a crawler quietly blocked, two pages competing, a language version outperforming the funded one.
It is the research and planning work that decides which changes a website needs in order to appear for the searches its buyers run. It covers technical health, keyword and topic mapping, content priorities and measurement. It stops short of writing the content or building the site.
Ours contains six things: a technical review, a keyword and category map, a register of pages competing with each other, an AI crawler and citation assessment, a review of each language version where more than one exists, and a 90-day sequence. Each finding names its owner. An audit that lists problems without saying who fixes them tends to sit in a drawer.
Technical corrections can move things within weeks. Competitive commercial terms usually take between six and twelve months of consistent work. Anyone offering a guaranteed timeline on a competitive term is not being straight with you.
No. Most engagements start with the audit and roadmap, which is bounded and delivered once. Some organisations take the roadmap and run it internally, which is a legitimate outcome. Ongoing work is available where it helps, not as a condition.
Three conditions have to hold at once. Their crawlers must reach the page, which means checking robots.txt and the network edge separately. The answer must survive being lifted out of the page. And the claim needs a source attached. Most sites fail the first condition without knowing.
Yes, and the two are connected. Assistants draw heavily on pages that already rank, so classic performance feeds AI citation rather than competing with it. What changes is measurement, because a well-ranked page can be summarised without ever being clicked.
Each language version is measured on its own first, because the assumption about which is strongest is frequently wrong. Then we check hreflang integrity across the set and confirm the versions are not competing for the same queries. This holds whether you run two languages or five.
Usually three things nobody has run: a cannibalization register, an AI crawler test from outside the network, and a like-for-like comparison of the language versions. All three are diagnostic and none requires changing supplier. Several clients have taken the findings back to their existing agency.
It is when several pages target the same term, so the engine picks between them and authority splits rather than accumulates. We register every case, then rule on each: merge, redirect or rewrite around a genuinely different term. The ruling is the valuable part.
Read-only Search Console and analytics, the site URL, and a short conversation about which enquiries you want more of. Nothing needs preparing. If Search Console was never set up, that is itself a finding and we will help configure it.
Yes, and most of this work is not financial. We work across professional services, manufacturing, technology, healthcare, education, real estate and retail. Regulated-sector experience is useful because those environments demand accuracy, not because the method is sector-specific.
Ask each to name something specific about your site before you engage them, because a firm that has looked will mention a real fault. Ask what happens to the work if you stop. And ask how they measure visibility inside AI assistants, because many still cannot answer.
Yes. We draft to pass review the first time rather than treating review as a hurdle afterwards: claims sourced, no superlatives that cannot be evidenced, a documented trail behind each statement. That discipline also happens to be what answer engines reward.
Every month without a diagnosis is another month of a competitor compounding, a crawler possibly blocked, and a language version nobody has measured.
The Search Visibility Diagnostic covers all three channels: how you perform in classic results, whether assistants can reach and cite you, and how each language version performs against the others. Findings arrive in writing and are yours regardless.
Search is one part of the wider digital presence services. The diagnostic will say plainly if the constraint sits somewhere else in it.
Book Your Free Visibility Assessment
A 30 minute conversation, nothing owed after it, and a diagnostic written against your own domain.
Ready to elevate your business with Prima Consulting? Fill out the form below to discuss your needs with our experts and discover how we can help you achieve your business goals.
We Schedule a call at your convenience
We do a discovery and consulting meeting
We prepare a proposal
