# AI Perception Maturity Assessment URL: https://www.onsomble.ai/tools/ai-perception-assessment Updated: 2026-09-14 A free self-assessment workbook for reviewing how AI represents your business, your information foundations and your improvement plan. Progress is stored in your browser. This document contains the blank workbook's questions and guidance; use the interactive page to save entries and download your own results. ## Using the workbook Free assessment tool Work through the sections at your own pace. Entries are saved on this browser and are not submitted to Onsomble. Download a backup to keep a copy or continue on another device. Your workbook You will finish with a picture of how AI represents you, the information and working practices that need attention, and up to three actions to take next. Have your website, an AI tool and any existing business information handy. Some checks need help from your website provider. This is a full assessment across 13 components, so you can complete it over several sittings. Start with the three fields marked “Needed for the AI prompts”. The other setup details can be added as you work. Write what is true about your business in your own words. Later, compare AI’s answers with this profile to spot missing or incorrect information. You do not need to paste this profile into the test conversations. Work through these two tests in order. Keep the results from each test, then review and score the combined observations in step 3. You can edit the details here without returning to Scope. Your previous entries are kept here unchanged. New results go under the relevant test above. Step 3 · Review your observations Use both sets of answers below. The first checks whether you appear without being named; the second also shows whether AI recognises your name and explains you accurately. Repeat the tests for the important customer situations and AI tools in your scope before judging a consistent pattern. For anything not tested yet, record “Not checked” in the finding and use U for an outcome you cannot score. Evidence strength is Low for a small sample, Medium for repeated checks of most priority situations, and High for repeated coverage across your chosen scope. ## Scoring | Score | Level | | --- | --- | | 1 | Unmanaged | | 2 | Reactive | | 3 | Established | | 4 | Managed | | 5 | Adaptive | U means unknown; N/A requires a reason and is available only where applicable. Record evidence for every score. The management headline uses the lowest management capability score, rather than averaging away a weakness. ## Does your brand appear without being named? Describe a customer looking for a provider. Keep your business name and website out of these details so you can see whether AI suggests you on its own. Copy this into a new ChatGPT, Claude, Gemini or other AI conversation. Run it, then paste the full answer and its source links into the results below. Record the AI tool and date. Copy the prompt, open your AI of choice (such as ChatGPT, Claude or Gemini), and paste it into a new conversation. Send it and wait for the full answer. Then copy the answer into the results box in step B below, including any source links. Record which AI you used and the date. Use a separate new conversation for each test. Keep each AI conversation in its own record. Include the prompt, full answer, AI tool, date and source links. Paste one AI conversation here, including the question and full answer. Add the AI tool, date, source links and whether web search was used. Long answers scroll inside this box. Add another observation for your next conversation. ```text Help me find suitable providers for the customer brief below. Customer situation: [customer situation] Requirements and priorities: [customer need] Recommend a shortlist of providers. For each, explain how it addresses this situation and meets the requirements. Compare the main differences to help me choose, and include the sources you use. ``` Did your business appear? Which other providers appeared and why? Use the questions below to record your findings. ## How does AI describe your brand? Start a separate, fresh conversation. This time, ask about your business by name and compare the answer with what your business actually offers. Copy this into a new ChatGPT, Claude, Gemini or other AI conversation. Run it, then paste the full answer and its source links into the results below. Record the AI tool and date. Copy the prompt, open your AI of choice (such as ChatGPT, Claude or Gemini), and paste it into a new conversation. Send it and wait for the full answer. Then copy the answer into the results box in step B below, including any source links. Record which AI you used and the date. Use a separate new conversation for each test. Keep each AI conversation in its own record. Include the prompt, full answer, AI tool, date and source links. Paste one AI conversation here, including the question and full answer. Add the AI tool, date, source links and whether web search was used. Long answers scroll inside this box. Add another observation for your next conversation. ```text Describe the business below to a potential customer. Business name: [business name] Organise your answer under these headings: 1. What it offers — its main products, services or experiences. 2. Who it serves — the customers and needs it is best suited to. 3. Where it operates — its locations and service areas. 4. Practical details — relevant pricing, availability, duration, suitability and restrictions. 5. Reasons to choose it — what distinguishes it and why a customer would consider it. Use specific details and include the sources supporting your description. ``` What did AI get right, leave out or misunderstand? Use the questions below to record your findings. ## Scope ### Business or brand name Use the name customers know you by. This name is also used in your branded AI prompt. ### Website address Paste the main address of your business website, such as https://yourbusiness.co.nz. ### Assessor and assessment date Write your name and today’s date so your team knows who completed this assessment and when. ### Offerings, audiences and locations in scope Choose what you are assessing this time: which products or services, which customers and which places. You do not have to cover the whole business at once. ### Customer situation to test Describe a typical customer and their situation. For example: a parent planning a rainy-day activity with two children. ### Customer need and important requirements What does that customer need? Include details that would affect their choice, such as budget, accessibility or available time. ### Location for the customer question Enter the place where the customer needs the service, such as Rotorua or New Zealand’s North Island. ### AI systems, observation dates and review period List the AI tools you will test, when you tested them and the period you want to compare. For example: ChatGPT and Claude, tested this week and again next month. ### Pages and external profiles checked; coverage gaps List the website pages and public listings you looked at. Note anything important you have not checked yet. ## Business profile ### What you offer and who it serves Describe your actual products or services and the people they are designed for. Use the words you would use with a new customer. ### Where you operate List where customers can visit you or use your services. Include your base and the areas you serve. ### Prices, duration, hours, suitability and restrictions Write down the practical facts customers need to decide: cost, opening times, duration, who can take part and any limits. ### Reasons to choose you and supporting evidence Why would a customer choose you? Add something that supports each reason, such as a qualification, review, award or specific feature. ### Customer decisions and alternatives What is the customer choosing between? Include other providers or other ways they could meet the same need. ### Intended perception supported by business evidence Describe how you want AI to explain your business. Make sure this matches what you can actually deliver. ## AI answers ### Observation record: prompts, full answers, dates, systems, sources and context Paste the question you asked and the full AI answer. Add the tool’s name, date, source links and whether it used web search or a previous conversation. ### Intended versus observed perception and customer significance Compare the AI answer with your business profile. What was missing or wrong, and how might it affect a customer’s decision? ## Plan and review ### Diagnosis: finding, customer situation, business evidence, public information, working explanation, unknowns and significance Choose a finding and explain what may be causing it. Point to the website text or other evidence you checked, and note what you still need to find out. ### Priority 1: finding, action, owner, delivery date, intended change and follow-up observation Choose your most important improvement. State what will change, who will do it, the deadline and how you will check whether it helped. ### Priority 2: finding, action, owner, delivery date, intended change and follow-up observation Choose your second improvement. Give it an owner, a deadline and a clear way to check the result. ### Priority 3: finding, action, owner, delivery date, intended change and follow-up observation Choose your third improvement. Keep it specific and achievable, with an owner, deadline and follow-up check. ### Outstanding checks: evidence needed, owner and due date List anything you still need to verify. Say who will check it and by when. An unanswered question is a task to investigate, not automatically a problem to fix. ### Review: what changed, what you learned, next action and next review date After making changes, repeat the relevant checks. Record what improved, what did not, what you will do next and your next review date. ## Contextualise Assessment area: management - C1 — Has the organisation documented what the business is and the category or categories in which it competes? - C2 — Are the products, services or experiences in scope clearly defined? - C3 — Are the priority customer groups and their relevant needs documented? - C4 — Are the markets, locations and service areas in scope clearly defined? - C5 — Has the organisation connected each priority offering to the customer needs it genuinely serves? - C6 — Are the capabilities, strengths and limitations supporting the intended position substantiated by current evidence? - C7 — Are the customer decisions and journey stages in which the brand should be considered clearly identified? - C8 — Has the organisation selected the AI systems and competitive alternatives relevant to those situations? - C9 — Is the business profile reviewed when an offering, audience, market or position changes? ## Scoring anchors - 1: The assessment has no agreed business profile or customer scope. - 2: People have an informal view, but audiences, offerings, limitations or relevant decisions are incompletely recorded. - 3: A documented, substantiated business profile defines the scope and is used to shape the assessment. - 4: An owner reviews it regularly and when the business changes, updating the assessment scope accordingly. - 5: Learning from assessments and customer decisions improves the profile and the situations examined in later cycles. ## Understand your business and customers Use the business profile you recorded earlier. Check whether the team agrees on the offerings, customers and decisions you want AI to understand. [Read the guide](https://www.onsomble.ai/guide/applied/define-your-business-profile) ## Observe Assessment area: management - O1 — Are observation questions derived from the defined customer groups, needs, offerings and markets? - O2 — Does the question set include branded prompts that test recognition and interpretation? - O3 — Does it include non-branded prompts that test whether AI introduces the brand independently? - O4 — Does it cover the customer journey stages where AI may shape awareness, comparison, selection or action? - O5 — Are observations collected across the AI systems that matter to the target audience? - O6 — Are prompts, responses, dates, systems and relevant context retained? - O7 — Are citations, retrieved sources and evidence captured with the responses? - O8 — Are competitors and alternatives recorded alongside the brand? - O9 — Are questions repeated enough to distinguish a recurring pattern from an isolated response? - O10 — Are observation conditions sufficiently consistent to support later comparison? ## Scoring anchors - 1: No usable observation practice exists. - 2: People ask occasional questions and save selected answers, with inconsistent scope or records. - 3: A defined question set covers priority situations and systems. Full observations and relevant context are retained and repeated comparably. - 4: An owner reviews coverage and consistency, runs the agreed schedule and follows up gaps in the evidence. - 5: Learning from previous rounds improves the question set and observation method while retaining a comparable core for tracking change. ## Check how AI represents you Look at your observation records. Check whether your team repeats the tests, records the full answers and compares results across relevant customer situations. [Read the guide](https://www.onsomble.ai/guide/applied/assess-your-ai-perception) ## Presence Assessment area: outcome - P1 — Is the brand correctly recognised when customers name it? - P2 — Does it appear when customers describe a relevant need without naming the brand? - P3 — Does it appear for the priority audiences, offerings, markets and locations? - P4 — At which customer journey stages does it appear or disappear? - P5 — Is it merely mentioned, or is it presented as an option to consider? - P6 — Is it shortlisted or recommended in situations where the evidence supports its relevance? - P7 — Which competitors or alternatives appear more prominently, and in which situations? - P8 — Which presence patterns recur across systems and repeated observations? ## Scoring anchors - 1: The brand is generally absent or not reliably recognised in the situations assessed. - 2: The brand is recognised when named but rarely enters relevant non-branded decisions, or appears only incidentally. - 3: The brand appears in some relevant decisions and is sometimes presented as an option, with noticeable gaps across situations or systems. - 4: The brand is consistently considered in the priority situations where the evidence supports its relevance. - 5: The brand is consistently and prominently considered or recommended across the priority scope when it genuinely fits, with the pattern maintained through ongoing observation. ## Interpretation Assessment area: outcome - I1 — How does AI categorise the business? - I2 — What products, services or experiences does it say the business provides? - I3 — Which customers, needs and use cases does it associate with the brand? - I4 — Which markets, locations and service areas does it associate with the brand? - I5 — Which capabilities, restrictions and limitations does it present? - I6 — Which attributes, strengths and weaknesses does it emphasise? - I7 — Which important facts or capabilities does it leave out? - I8 — How does it compare the brand with relevant alternatives? - I9 — What reasons does it give for recommending, qualifying or excluding the brand? - I10 — Which claims are supported by suitable evidence, and which are unsupported, outdated or contradicted? - I11 — How does the representation change across customer groups, markets, systems or repeated observations? ## Scoring anchors - 1: The representation is materially incorrect, misleading or commercially harmful. - 2: AI recognises parts of the business but presents an incomplete, outdated or inconsistent picture. - 3: The representation is broadly accurate but lacks important detail, differentiation or contextual fit. - 4: The brand is accurately and distinctly represented across the priority scope, with claims generally supported by suitable evidence. - 5: The representation remains accurate, differentiated and contextually appropriate across audiences, needs, systems and time, with material changes detected through ongoing observation. ## Brand clarity Assessment area: readiness - BC1 — Does the website state clearly what the business offers? - BC2 — Does it identify who each important offering is for? - BC3 — Does it clearly state where the business operates or provides the offering? - BC4 — Are prices or useful price ranges provided where customers need them? - BC5 — Are hours, seasons, duration, availability and other practical details clear? - BC6 — Are restrictions, exclusions and suitability requirements stated directly? - BC7 — Does the website explain what makes the business meaningfully different? - BC8 — Does it give customers evidence-based reasons to choose the business? - BC9 — Is the next action to book, buy or enquire clear? ## Scoring anchors - 1: A visitor cannot establish the core offer, intended customer or operating area from the published information. - 2: Some pages explain the offer, but important facts such as suitability, location or price are missing or vague. - 3: Important offerings clearly state the relevant facts, supported reasons to choose them and the next action. A repeatable content checklist is used. - 4: A named owner reviews clarity across the important pages and checks it against the current business profile and customer questions. Gaps are followed up. - 5: Repeated customer feedback and AI assessments inform improvements to the messaging, and the team checks whether understanding improves. ## Brand clarity Pick your homepage and one important product or service page. Check whether someone new can tell what you offer, who it is for and where it is available. Use the brand-clarity exercise to compare AI’s reading with yours. [Try the exercise](https://www.onsomble.ai/guide/ai-discoverability-foundations/can-ai-understand-your-brand#try-it-yourself) [Read the guide](https://www.onsomble.ai/guide/applied/readiness-brand-clarity) ## Content completeness and structure Assessment area: readiness - CS1 — Do page titles and descriptions identify the subject, offering and location where relevant? - CS2 — Do headings organise pages around recognisable customer questions and topics? - CS3 — Are important facts expressed in clear paragraphs, lists or tables? - CS4 — Do links describe the information or destination they lead to? - CS5 — Do meaningful images have useful text descriptions? - CS6 — Are essential facts available as text rather than only inside images, video or downloadable files? - CS7 — Are organisation, location, offering and practical-information pages connected logically? - CS8 — Can customers and machines reach important information through ordinary navigation and links? - CS9 — Is essential information available when the page initially loads? - CS10 — Is essential information available without first opening a pop-up, tab, accordion or booking widget? ## Scoring anchors - 1: Important facts are missing from readable page content or the structure prevents people and machines from finding them. - 2: Some pages use useful titles, headings and text, but important page types or facts are inconsistent or difficult to extract. - 3: Important pages use a consistent structure with descriptive headings, useful links and accessible text. New pages follow the same approach. - 4: An owner checks new and changed pages, including their readable content, and tracks structure or extraction problems to resolution. - 5: Evidence from repeated use and extraction checks improves the templates and publishing process; the improvements are checked again. ## Content completeness and structure Check the same pages for useful titles, headings and practical details. Write down any facts that are missing or only visible after an interaction. [Try the exercise](https://www.onsomble.ai/guide/ai-discoverability-foundations/can-ai-understand-your-brand#try-it-yourself) [Read the guide](https://www.onsomble.ai/guide/applied/readiness-content-structure) ## Crawler access Assessment area: readiness - CA1 — Has the current `robots.txt` file been reviewed? - CA2 — Are rules applying to all crawlers understood? - CA3 — Have rules for the relevant search and AI crawler identities been checked? - CA4 — Are the homepage and primary navigation pages permitted? - CA5 — Are important offering, location, pricing and visitor-information pages permitted? - CA6 — Are booking, purchasing and enquiry paths permitted where they contain useful public information? - CA7 — Have path-level restrictions been checked for unintended effects? - CA8 — Have hosting, content-delivery and security controls been checked for unintended automated-access blocks? ## Scoring anchors - 1: Required public pages are confirmed blocked for relevant discovery crawlers, or the requested review activity does not happen. - 2: Only some rules or important paths have been checked, or permissions are inconsistent and fixes occur after problems appear. - 3: Checks cover relevant crawler identities, important paths and applicable security controls. Required public access works and the checks are repeatable. - 4: An owner reviews access after relevant site or hosting changes and at an agreed interval, following up unintended blocks. - 5: Repeated findings improve the access checks and release process, including how the team responds to changes in relevant crawler requirements. ## Crawler access Run the crawler-permissions exercise on your website. Keep its findings, then ask your website provider to check any hosting or security restrictions it cannot establish. Please review our robots.txt rules and hosting/security settings for unintended AI crawler blocks on our important public pages. Tell us what you checked, what is blocked and what needs changing. [Try the exercise](https://www.onsomble.ai/guide/ai-discoverability-foundations/ai-crawler-access#try-it-yourself) [Read the guide](https://www.onsomble.ai/guide/applied/readiness-crawler-access) ## Indexability, discoverability and extractability Assessment area: readiness - ID1 — Do important public pages return successful responses reliably? - ID2 — Do page-level indexing instructions permit the intended pages to be included? - ID3 — Do canonical instructions identify the correct version of each important page? - ID4 — Do internal links make important pages discoverable? - ID5 — Are sitemaps and other discovery mechanisms accurate where they are used? - ID6 — Are duplicate, conflicting and obsolete pages controlled? - ID7 — Can the main page content be extracted as meaningful text? - ID8 — Does JavaScript rendering preserve access to essential information? - ID9 — Can important information be reached without authentication or a private session? - ID10 — Do embedded booking or commerce systems expose enough public information to support discovery and understanding? ## Scoring anchors - 1: Important public information cannot be reached, is excluded from indexing unintentionally or cannot be read meaningfully. - 2: The homepage works, but some important pages have incorrect instructions, weak links or missing readable content. - 3: Important page types have working responses, suitable indexing instructions, useful links and readable content. Publishing includes repeatable checks. - 4: An owner reviews new pages and site changes, investigates failures and records their resolution. - 5: Recurring findings are used to improve templates, navigation or rendering, with follow-up checks showing the improvements work. ## Can AI find and read your pages? Choose important pages and ask your website provider to check whether search systems can find them and extract their main information. Use U for anything you have not checked yet. Please check our important pages for indexing instructions, canonical URLs, internal links and readable main content. Identify information that only appears after JavaScript runs or a visitor interacts. [Read the guide](https://www.onsomble.ai/guide/applied/readiness-indexability) ## Machine-readable meaning Assessment area: readiness - MM1 — Does the site use appropriate structured data for the organisation or business? - MM2 — Are locations and service areas represented accurately? - MM3 — Are important products, services, experiences or events represented where appropriate? - MM4 — Are prices, offers, availability and opening hours represented where appropriate? - MM5 — Are relationships between the organisation, its locations and its offerings explicit? - MM6 — Are stable identifiers and official profiles connected where appropriate? - MM7 — Does the structured data agree with the visible content? - MM8 — Is the implementation technically valid? - MM9 — Is the structured information maintained when the business changes? - MM10 — If an `llms.txt` file is used, does it direct systems to useful, current public content without replacing the site's core information structure? ## Scoring anchors - 1: Required structured information is absent or materially incorrect. - 2: Basic information exists, but important entities, facts or relationships are incomplete or inconsistent. - 3: Appropriate structured data describes the important facts and relationships, agrees with the page and passes relevant validation checks. Updates are repeatable. - 4: An owner checks coverage and accuracy when pages or offerings change and resolves validation or consistency issues. - 5: Repeated findings improve how structured information is produced and maintained, with follow-up validation across affected pages. ## Clear labels for machines Ask your website provider whether structured data on your pages describes the business and offerings correctly and agrees with the visible content. Please review the structured data on our important pages. Does it accurately identify our business and offerings, match the visible information and pass the relevant validation checks? [Read the guide](https://www.onsomble.ai/guide/applied/readiness-structured-data) ## Freshness and consistency Assessment area: readiness - FC1 — Are current products, services and experiences accurately represented? - FC2 — Are expired offerings removed, redirected or clearly archived? - FC3 — Are prices, hours, availability, locations and contact details current? - FC4 — Do important facts agree across related website pages? - FC5 — Does structured data agree with visible information? - FC6 — Do important external platforms receive material updates? - FC7 — Is there a named owner for each important information source? - FC8 — Is there a defined review or expiry process for time-sensitive information? ## Scoring anchors - 1: Important published facts are outdated or contradictory and updates are unmanaged. - 2: Corrections occur when someone spots a problem, leaving gaps between pages or platforms. - 3: Important facts are current and consistent, supported by a repeatable update routine. - 4: Each important source has an owner; expiry dates or review triggers prompt checks and corrections. - 5: The team uses recurring discrepancies to improve how updates reach all relevant sources and checks whether those improvements reduce gaps. ## Fresh and consistent information Compare prices, hours, locations and availability across your website and important listings. Check who updates them when something changes. [Read the guide](https://www.onsomble.ai/guide/applied/readiness-freshness) ## Wider public account and authority Assessment area: readiness - WA1 — Is the business accurately represented on important destination, tourism or industry websites? - WA2 — Are important business directories and official profiles accurate and consistent? - WA3 — Do partner, distributor and booking-platform profiles describe the current offering correctly? - WA4 — Is there credible customer evidence supporting important experience and performance claims? - WA5 — Do reviews and public discussions provide useful evidence about the customer experience? - WA6 — Does relevant media, editorial or comparison coverage reflect the current business? - WA7 — Are the business identity, locations, offerings and major claims consistent across important external sources? - WA8 — Are outdated or incorrect third-party descriptions identified and addressed with suitable evidence? - WA9 — Do the sources cited in AI answers provide relevant support for the claims made? ## Scoring anchors - 1: Required external profiles are absent or materially wrong, or important claims lack supporting evidence. - 2: Some useful listings and customer evidence exist, but coverage, accuracy or follow-up is inconsistent. - 3: Relevant external sources accurately describe the business, important claims have credible support and a repeatable correction process exists. - 4: An owner reviews priority sources, investigates unsupported descriptions and tracks correction requests and outcomes. - 5: Recurring customer and AI findings guide improvements to external information and supporting evidence, with subsequent review of what changed. ## What other sites say about you Review important directories, review sites, partner pages and other public references. Compare their descriptions with your current business profile. [Read the guide](https://www.onsomble.ai/guide/applied/readiness-public-account) ## Connected and actionable information Assessment area: readiness - AC1 — Is current availability accessible through a reliable structured source where required? - AC2 — Is live or frequently changing pricing available where required? - AC3 — Can systems access relevant inventory, capacity or location-specific information? - AC4 — Are booking, purchasing or enquiry capabilities structured for reliable interaction? - AC5 — Are suitable APIs or structured feeds available and maintained? - AC6 — Where MCP or another tool interface is used, is access controlled and aligned with a defined customer need? - AC7 — Where UCP or another transaction mechanism is used, are the supported actions and limits clear? - AC8 — Are authentication, permissions, monitoring, failure handling and operational ownership defined? ## Scoring anchors - 1: A required information feed or customer action is unavailable or does not work. - 2: A limited implementation works for some offerings or situations, with manual updates or unreliable coverage. - 3: The required information or action works across the agreed scope, with a repeatable operating process and defined access and failure handling. - 4: A named owner reviews reliability, freshness and failed interactions and follows up problems. - 5: Evidence from repeated interactions improves the information or action and its operating process, with the changes checked afterwards. ## Connected information and next actions Check whether customers can move from information to an enquiry, booking or purchase. Review feeds or integrations only where they are relevant to your business; otherwise explain N/A. Which feeds or integrations publish our business information, availability or booking options? Please check their accuracy, ownership and maintenance. Note which capabilities are not relevant to us. [Read the guide](https://www.onsomble.ai/guide/applied/readiness-connected-information) ## Diagnose Assessment area: management - D1 — Does diagnosis begin with recurring observed findings rather than assumptions about what AI systems may be doing? - D2 — Are observed answers compared with the substantiated business profile and intended position? - D3 — Can the organisation distinguish presence, understanding, association, differentiation, trust, sentiment and recommendation problems? - D4 — Are factual errors distinguished from gaps in emphasis, audience association or contextual fit? - D5 — Is the customer or commercial significance of each finding assessed? - D6 — Are citations, retrieved sources and the surrounding evidence inspected? - D7 — Are the relevant Information readiness components examined? - D8 — Does the investigation consider the business itself, its own explanation and the wider public account? - D9 — Are established facts, working explanations and unresolved uncertainty recorded separately? - D10 — Does each prioritised finding lead to a proportionate next action to assess? ## Scoring anchors - 1: Actions are chosen without investigating observed problems. - 2: Some problems are investigated, but conclusions are informal or rest on assumptions. - 3: A repeatable diagnosis compares observed answers with business and public evidence, records explanations and prioritises the customer impact. - 4: An owner reviews diagnosis quality and priorities, and important unresolved questions have follow-up actions. - 5: Later observations and intervention results improve the explanations and investigation method used in subsequent cycles. ## Understand what is causing gaps Choose an inaccurate or missing AI answer. Trace it back to public information and record what you know, what you suspect and what still needs checking. [Read the guide](https://www.onsomble.ai/guide/applied/diagnose-perception-gaps) ## Enhance Assessment area: management - E1 — Is every intervention connected to a prioritised finding and supported diagnosis? - E2 — Does the organisation address genuine product or service gaps rather than merely strengthening the wording of unsupported claims? - E3 — Can it improve its own explanation through clearer content, stronger evidence and accurate limitations? - E4 — Can it address technical readiness issues such as crawler access, indexability, rendering or structured data? - E5 — Can it coordinate corrections and improvements across important external sources? - E6 — Is the intended change in Presence or Interpretation defined before work begins? - E7 — Does every intervention have an accountable owner and delivery date? - E8 — Is the observation that will assess the result defined in advance? - E9 — Are interventions prioritised according to customer and commercial importance? - E10 — Can teams deliver improvements across brand, content, search, product, technology and reputation where required? ## Scoring anchors - 1: Relevant findings do not lead to action. - 2: Occasional fixes happen without consistent priorities, ownership or intended results. - 3: Prioritised findings lead to delivered actions with owners, intended perception changes and planned follow-up observations. - 4: Delivery and follow-up are reviewed, and actions across the business, website and external sources are coordinated as needed. - 5: Results from completed interventions improve how subsequent actions are chosen and delivered. ## Make and follow through on improvements Review recent changes made in response to findings. Check whether each had an owner, a clear intended result and a follow-up check. [Read the guide](https://www.onsomble.ai/guide/applied/plan-perception-improvements) ## Continuous learning loop Assessment area: management - L1 — Is there a recurring observation and review cadence? - L2 — Are post-intervention observations comparable with the original observation set? - L3 — Does the organisation check whether updated information is available in the relevant sources? - L4 — Are changes in AI systems, source material and customer context recorded when results are compared? - L5 — Does the organisation distinguish an observed change from proof that a particular intervention caused it? - L6 — Is the intervention register maintained with outcomes and learning? - L7 — Are unsuccessful interventions used to refine the diagnosis and choose the next action? - L8 — Do brand, content, search, product, technology and reputation owners review the evidence together where relevant? - L9 — Are priorities updated according to customer and commercial significance? - L10 — Does the organisation return to Contextualise when its offering, audience, market or positioning changes? ## Scoring anchors - 1: The assessment ends with the initial report or changes. - 2: Follow-up happens occasionally, without consistent comparisons or retained learning. - 3: The team returns to comparable observations, records results and uses them to choose the next action. - 4: An owner maintains the review cycle, evidence and action register, with relevant people reviewing the results. - 5: Multiple completed cycles show how learning changed priorities, diagnosis or the business profile and informed subsequent work. ## Learn and repeat Look for evidence that earlier findings changed how your team works. Check whether you revisit results and agree the next action and review date. [Read the guide](https://www.onsomble.ai/guide/applied/review-ai-perception-progress) ## Improvement plan ### Finding to address Describe the problem and why it matters to a customer. You can bring a low-scoring finding into this field using the picker above. ### What will you change? Write one specific change someone can deliver, such as publishing accurate opening hours. ### Who will do it? Name the person or team responsible for making this change. ### Target date Choose the date you want the change completed. ### What should improve? Describe the difference you want to see, such as AI giving customers the correct opening times. ### How and when will you check? Say how you will check the result and when. For example, repeat the same AI questions two weeks after the update. [Open the interactive workbook](https://www.onsomble.ai/tools/ai-perception-assessment)