How do I map old and new product positioning?
The first step after a product positioning change is to write a side-by-side map of what the company used to mean and what it means now. Record the old category, new category, audience, problem, alternatives, proof points, product names and claims that must no longer lead. Do not start by rewriting every page, because broad edits can remove evidence that still helps buyers understand the product.
Mark each statement as retained, revised, retired or newly introduced. Include language customers use, not only language chosen by the brand team. A repositioning often changes the category label while leaving product documentation, comparison pages and support content attached to the old one. That creates a split signal: an assistant may recognize the company but describe it using the previous position.
Check whether the new position is specific enough to distinguish the product from adjacent categories. If the new statement could describe several types of software, later visibility fixes will have no stable target. Save this map as the reference for every prompt, page edit and measurement decision that follows.
Rebuild the buyer-language prompt set
The second step is to replace prompts built around the former position with questions real buyers would ask under the new one. Group prompts by category discovery, problem diagnosis, shortlist comparison, implementation, proof and alternatives. Keep a small sample of old prompts so the team can see whether the former position is fading, persisting or still useful.
Check every prompt for intent and language. A prompt that merely asks whether the brand exists measures recognition, while a prompt asking which option suits a stated situation measures fit. Include brand-led, category-led and competitor-led wording, along with questions that use customer vocabulary rather than the new positioning statement. Avoid treating one answer from one engine as a market verdict.
Cituna asks ChatGPT, Perplexity, Gemini, Claude, Grok, Google AI Overviews and Google AI Mode the questions a brand's buyers ask every day. Its records show who each answer names and cites, at what position, plus the competitors and pages that appear instead. That makes prompt design the measurement foundation, not a final reporting task.
Measure recognition separately from recommendation
The third step is to separate whether an engine understands the new position from whether it recommends the company for it. Recognition means the answer describes the company or product accurately. Recommendation means the company appears when a buyer asks for a suitable option. Citation adds another test: whether the answer points to a page that supports the claim.
Check these signals independently across the seven engines. A company can be named but described with its old category, cited but omitted from a shortlist, or recommended without a source that supports the new promise. Each outcome calls for a different response. Incorrect description points to entity and page language. Missing recommendation points to relevance, comparison and proof. Missing citation points to accessible evidence.
Record the answer wording, named position, cited URL, competing names and the reason the response appears wrong. Do not compress these into one visibility score before reviewing the underlying answers. The most useful early result is often a mismatch between recognition and recommendation, because it shows whether the problem is understanding the repositioning or trusting it.
Audit evidence for consistency across the buyer journey
The fourth step is to check whether the new position is supported consistently from the homepage through comparison, product, documentation and proof pages. An assistant needs more than a repeated slogan. It needs clear descriptions, specific use cases, product relationships, limitations and evidence that can be connected to the claim.
Check page titles, headings, structured data, internal links, visible copy and update dates for contradictions. Look for old category terms that remain prominent in navigation or metadata, and for new claims that appear only on a campaign page. Review whether product names, company names and capabilities are written consistently enough to connect the pages.
The check also applies beyond the main site. Partner pages, review pages, directories and press coverage may continue to describe the former position. Do not assume every third-party statement must be changed immediately. Prioritize sources that engines cite in the answers you are investigating, then distinguish a harmful contradiction from a merely older description.
For a repositioning that changes product relationships, use the same method as an audit of AI visibility across product lines, because engines may merge or separate offerings differently after the change.
Choose the smallest fix that resolves the observed gap
The fifth step is to match each observed gap to one corrective action instead of rewriting the site by instinct. Use a page change when the right evidence exists but is unclear. Use schema or FAQ markup when important relationships are difficult for crawlers to interpret. Use a new explanatory page when the new category or use case has no credible source. Use a correction outside the site when a cited third-party page is materially wrong.
Check that the proposed fix addresses the answer's failure mode. Adding the new positioning statement to every page will not solve a missing comparison, an unsupported outcome or a competitor's stronger evidence. Conversely, publishing a new article will not help if the product page still leads with the retired category.
Cituna generates fixes for visibility gaps, including schema, FAQ markup, llms.txt and page changes. Its AutoSEO can write articles from those gaps and from Search Console demand, then publish them to WordPress, Shopify, a GitHub repository or another CMS by webhook. Treat those outputs as proposed responses to measured gaps, not substitutes for deciding what the new position must prove.
Apply changes in a controlled order
The sixth step is to change the source of truth before adding supporting content. Start with the company and product descriptions that define the new position, then update the pages that explain use cases, comparisons, proof and implementation. Add supporting markup only after the visible wording is accurate. This order reduces the chance that structured data reinforces a claim the page itself does not clearly make.
Check one positioning pathway at a time where possible. A useful pathway might run from a category page to a product page, then to a proof page and a comparison page. Confirm that each link answers the next buyer question and that no step sends the reader back to retired language. Keep a dated record of changed URLs, claims, markup and publication status.
Do not interpret a short-term drop in old-query visibility as automatic failure. Some loss is expected when obsolete demand no longer matches the product. The decision rule is whether qualified new-position prompts improve without creating inaccurate descriptions or unsupported recommendations. If a website migration is part of the repositioning, treat AI visibility after a website migration as a separate control because redirects and discoverability can obscure the positioning result.
Re-scan the same prompts and connect answers to outcomes
The seventh step is to rerun the same prompt set across ChatGPT, Perplexity, Gemini, Claude, Grok, Google AI Overviews and Google AI Mode after the changes have been available long enough to be encountered. Compare answer wording, names, citations, positions, competing pages and the specific gap each change was meant to address.
Check both movement and quality. A better position is not proven by appearing more often if the answer still uses the retired category or cites the wrong page. Review whether new citations land on pages that support the claim, whether competitors remain the preferred alternatives and whether answers differ sharply between engines.
Connect visibility observations to first-party demand. Cituna includes Google Search Console, so teams can compare changes with clicks and search queries rather than treating assistant answers as the only outcome. Keep those measures separate: an engine can improve its description before search traffic changes, and clicks can move for reasons unrelated to the repositioning. Record the timing, affected prompts and page changes so the next decision has a traceable basis.
Set a transition decision and rollback rule
The eighth step is to decide when the new position is established enough to retire the old one and what evidence would justify a rollback. Set the decision around answer quality, not a single visibility movement. The new position should be described accurately, associated with the intended buyer problem, supported by appropriate citations and free from major contradictions across the pages engines use.
Check unresolved prompts by failure type. If recognition is accurate but recommendations remain weak, improve comparison and proof. If citations point to old pages, repair redirects, links and source hierarchy. If one engine changes while others do not, verify whether the issue is retrieval, wording or page access before changing the core message.
Keep the old positioning map, prompt set, answer captures and page-change record. A rollback should restore only the last known accurate source, not automatically restore every retired claim. Re-scan after material edits and after major product, pricing or acquisition changes. A hosted MCP server can connect Claude or another AI agent to the same Cituna data, with read tools on every plan and additional scan and workflow actions on Pro.
Related reading
Sources consulted
- Google Search Central (developers.google.com)
- Google Search Console Help (support.google.com)
- OpenAI Platform (platform.openai.com)
- Anthropic (anthropic.com)
Drafted with AI assistance from our own research and Search Console data, and reviewed by Rahul A before publishing. Rules and prices change; check the linked official source before you act.