{"site":{"name":"Koji","description":"AI-native customer research platform that helps teams conduct, analyze, and synthesize customer interviews at scale.","url":"https://www.koji.so","contentTypes":["blog","documentation"],"lastUpdated":"2026-11-17T07:51:34.534Z"},"content":[{"type":"blog","id":"ffef407a-d531-4834-85e6-3553ba3bfd50","slug":"ai-readiness-assessment-interviews-consultancy-playbook","title":"AI Readiness Assessment Interviews: A Consultancy Playbook for Hearing the Whole Workforce in Week One","url":"https://www.koji.so/blog/ai-readiness-assessment-interviews-consultancy-playbook","summary":"A playbook for consultancies running AI readiness assessments. It recommends a three-layer read: 5 to 8 live sponsor interviews, 8 to 12 AI-moderated manager interviews per function and 10 to 15 per frontline role family, fielded in days 3 to 10 of a four-week engagement. It gives a nine-question guide, cites MCA 2026, Gartner 2025 and 2026, BCG 2026 and McKinsey 2025 data on AI use and change adoption, and shows the Koji setup: Change Impact Assessment template, context files, structured questions, voice or text, Remove personal details and one study per layer.","content":"AI readiness assessment interviews are the part of an AI readiness engagement where you find out how a client's people actually work with AI today: who uses it, for what, what stops the rest, and what would have to change for the roadmap to land. Most consultancies run them as a dozen executive calls plus a survey. Running them as AI-moderated interviews lets you hear 60 to 150 people across every layer in the first two weeks, with the same questions and probes for each, so the readout rests on the workforce and not on the leadership team's guess about it. By the end of this playbook you'll have a week-by-week plan, a sample design, an interview guide to copy, and the exact setup in Koji.\n\n## Why AI readiness interviews matter now\n\nClients are buying this work. The Management Consultancies Association's 2026 Client Survey, run by Savanta with more than 350 UK leaders, found that two in five organisations are looking to consulting firms for help with digital and technological transformation. Asked what holds AI back, 40% named data security and privacy, and 34% a lack of skilled internal teams ([consultancy.uk, 2026](https://www.consultancy.uk/news/44009/uk-consulting-clients-turning-to-sector-for-ai-transformation-support)).\n\nThe work is about people more than tools. In a December 2025 Gartner survey of 110 CHROs, 78% agreed that workflows and roles will need to change for AI investments to pay off, and just over half of organisations had already redesigned roles because of AI ([Gartner, March 2026](https://www.gartner.com/en/newsroom/press-releases/2026-3-16-gartner-identifies-top-change-management-trends-for-chros-in-age-of-ai)). BCG's fourth AI at Work survey of nearly 12,000 people found regular AI use among frontline employees jumped to 74%, up 23 points in a year, yet 66% of those regular users get limited or no guidance on what to do with the time it saves ([BCG, June 2026](https://www.bcg.com/publications/2026/ai-at-work-why-strategy-matters-more-than-tools)).\n\nLeadership's picture of all this is usually wrong. McKinsey asked US employees and executives how much of their daily work employees do with generative AI. 4% of C-suite respondents thought employees used it for more than 30% of their daily tasks; 13% of employees said they already did. Looking a year ahead, the figures were 16% and 34% ([McKinsey, Superagency in the Workplace, January 2025](https://www.mckinsey.com/capabilities/tech-and-ai/our-insights/superagency-in-the-workplace-empowering-people-to-unlock-ais-full-potential-at-work)).\n\n![Column chart: C-suite estimates vs employee reports of using gen AI for over 30% of daily tasks, 4% vs 13% today and 16% vs 34% within a year.](/visuals/leaders-underestimate-employee-ai-use-mckinsey-2025/leaders-underestimate-employee-ai-use-mckinsey-2025.png \"If you only interview the executives, you inherit their estimate. · Source: McKinsey, Superagency in the Workplace, January 2025\")\n\nAnd the change itself rarely lands well. Only 32% of mid-to-senior leaders told Gartner the last change they led achieved healthy adoption by employees ([Gartner, July 2025](https://www.gartner.com/en/newsroom/press-releases/2025-07-08-gartner-hr-research-finds-just-32-percent-of-business-leaders-report-achieving-healthy-change-adoption-by-employees)), and an April 2025 Gartner survey of 2,850 employees found 79% have low trust in organisational change. Prosci's long-running research puts the stakes plainly: 88% of projects with excellent change management met or exceeded their objectives, against 13% of those with poor change management ([Prosci, Best Practices in Change Management, 12th edition, 2023](https://www.prosci.com/blog/the-correlation-between-change-management-and-project-success)).\n\nFor a consultancy, that adds up to one risk. If your readiness score comes from ten executives and a 1-to-5 survey, the roadmap you sell next is built on the group that misjudges usage most. The follow-on work stalls in month three, and the client remembers whose diagnosis it was.\n\n## What good looks like\n\nWe run AI readiness interviews in three layers, and we call the method the **three-layer read**: sponsors tell you the ambition, managers tell you the constraints, and the frontline tells you what is really happening. A readout is credible when every recommendation can show quotes from at least two layers.\n\n![Consultants around a table with laptops and notes, planning which client teams to interview in an AI readiness assessment](https://images.unsplash.com/photo-1499750310107-5fef28a66643?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHw3fHxjb25zdWx0YW50cyUyMHdvcmtzaG9wJTIwdGFibGUlMjBsYXB0b3BzJTIwbm90ZXN8ZW58MXwwfHx8MTc5NDkwMTgxMHww&ixlib=rb-4.1.0&q=80&w=1080 \"Plan the layers before you write a single question.\")\n\n_Photo by [Andrew Neel](https://unsplash.com/@andrewtneel?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\n**Who to interview, and how many.** This is our rule of thumb from the method, not a statistical sample:\n\n- **Sponsors and the executive team (5 to 8):** live, by a partner or principal. These are relationship conversations, and they set the hypotheses.\n- **Function and line managers (8 to 12 per function):** AI-moderated, voice or text. Managers decide whether a tool gets used on Tuesday morning.\n- **Frontline role families (10 to 15 per role family):** AI-moderated. A role family is a group doing the same work, such as claims handlers or account managers.\n- **IT, data, risk and legal owners (3 to 5):** live or AI-moderated. They know what is allowed, and the frontline often doesn't.\n\nStop adding interviews in a layer when the last five add no new theme. For a 1,200-person client with four functions in scope, that lands at roughly 80 to 120 interviews.\n\n**Timing.** Field in week one and two, not week three. A typical four-week assessment goes: days 1 to 3, sponsor interviews and brief sign-off; days 3 to 10, AI-moderated interviews in the field; days 10 to 12, synthesis; week 3, the roadmap draft; week 4, the steering committee.\n\n**Question design.** Anchor on the last real week, not on attitudes. \"Walk me through the last time you used an AI tool for work\" gets a story; \"How do you feel about AI?\" gets a press release. Ask about what people do when the official tool falls short, because that is where unapproved tools and the real use cases show up. Keep it to 15 minutes.\n\n**Confidentiality.** Promise it, and make it true. Tell people that answers are reported by group, not by name, that nobody at their employer will see a transcript with their name on it, and say it in the invitation from the client sponsor, not only on the welcome screen.\n\n**Common mistakes:**\n\n1. Interviewing only the people the sponsor nominates. You get the enthusiasts. Ask HR for a list by role family and invite all of it.\n2. Treating a survey score as readiness. A 3.2 out of 5 tells the steering committee nothing it can act on.\n3. Asking \"Do you use AI?\" People who paste text into a chatbot every day often answer no, because it isn't \"the AI tool\".\n4. Mixing layers in one analysis. A theme that 40% of managers raise and 5% of the frontline raise is two findings, not one.\n5. Leaving the shadow-AI question out because it feels awkward. It's usually the most useful thing you hear.\n6. Reporting themes with no evidence trail. Clients now ask where AI-assisted claims come from, and \"the interviews said so\" needs quotes behind it.\n\n## Copy this interview guide\n\n**One-line research brief** (paste it into \"What do you want to research?\"):\n\n> Find out how employees in [function] at [client] use AI in their work today, what stops them using it more, and what would need to change in their tools, rules, training and workflows for the AI roadmap to succeed.\n\n**Opening line:** \"Thanks for taking 15 minutes. This is part of an independent review of how AI fits into your work. Your answers are reported by group, never by name, and there are no right answers.\"\n\n1. **Open.** \"Talk me through a normal working day. Which parts take the most time?\" Probe: \"Which of those would you happily hand to someone else?\"\n2. **Single choice.** \"How often did you use any AI tool for work in the last two weeks?\" (Not at all, Once or twice, Weekly, Most days, Several times a day.) Probe: \"What was the last thing you used it for?\"\n3. **Open.** \"Walk me through the last time you used an AI tool for work, from start to finish.\" Probe: \"What did you do with the result?\"\n4. **Open.** \"When the approved tools don't help, what do you do instead?\" Probe: \"Are there tools people use that aren't official?\"\n5. **Multiple choice, shuffled.** \"What gets in the way of using AI more in your role?\" (I don't know what's allowed, The tools don't fit my work, I don't trust the output, No time to learn, My manager doesn't encourage it, Data I need isn't available, None of these.) Probe: \"Tell me about a time that happened.\"\n6. **Scale, 1 to 10.** \"How confident are you that you could use AI well in your role in six months?\" Probe: \"What would move that up by two points?\"\n7. **Open.** \"If AI took over part of your job tomorrow, what would you want to spend that time on?\" Probe: \"What worries you about it?\"\n8. **Open.** \"What would your manager need to do differently for this to work?\" Probe: \"Has anyone done this well around you?\"\n9. **Closing.** \"If you could tell the leadership team one thing about AI here, what would it be?\"\n\n![Question bank for AI readiness assessment interviews in three sections: how work gets done today, what gets in the way, and what would need to change.](/visuals/ai-readiness-assessment-interview-questions/ai-readiness-assessment-interview-questions.png \"Nine questions that take about 15 minutes, by voice or text.\")\n\n## How to set it up in Koji, step by step\n\nSay you've just signed a four-week AI readiness assessment for an insurer with 1,200 people across claims, underwriting, operations and customer service. Here's the build.\n\n1. **Start the study.** Open the template gallery and start from **Change Impact Assessment** under Business Analysts, or paste the one-line brief above into \"What do you want to research?\" on the home screen. The research assistant drafts the goal, questions and follow-up approach. Ask it for changes in plain words (\"make question 5 a multiple choice question with these options\").\n2. **Give it the engagement context.** Upload the SOW summary, the client's list of approved AI tools and its AI policy as context files (PDF, TXT, JSON or Markdown, up to 25 MB each). The assistant reads them when it drafts and changes the brief, so follow-ups use the client's own tool names. Add guardrails for what the interviewer must not do, such as naming colleagues or speculating about job cuts.\n3. **Add the structured questions.** Make question 2 a single choice, question 5 a multiple choice with option shuffling on and \"None of these\" kept at the end, and question 6 a 1 to 10 scale. Each gets its own chart in the report. The [structured questions guide](/docs/structured-questions-guide) covers each type.\n4. **Choose voice or text, and the language.** Turn on both and let people pick. Voice suits frontline staff on the move; text suits open-plan offices, and people can dictate answers into the text box. If the client works in several languages, turn them on: Koji interviews in more than 30, and each person's interview opens in the language they most likely read.\n5. **Turn on Remove personal details.** Names, email addresses, phone numbers, postal addresses and employers are taken out of each transcript when the interview ends and replaced with labels such as [NAME]. Reports and exports only see the cleaned text.\n6. **Make one copy per layer.** Finish the frontline version, preview it yourself (previews are free and never counted), then duplicate it for managers and for IT and risk owners and adjust the wording. One study per layer gives you one report per layer, which is how you keep the three-layer read clean.\n7. **Invite people.** For full confidentiality, send one shared link per layer through the client sponsor's email, so you never upload a list of names. When you need reminders and response tracking, import the HR list as a CSV and each person gets a private link, or email an audience from Participants.\n\n![Seven-step process for AI readiness interviews in Koji, from the template and context files to removing personal details, one study per layer and invites.](/visuals/ai-readiness-interviews-setup-steps/ai-readiness-interviews-setup-steps.png \"Build the frontline study first. The others are copies of it.\")\n\n### Inviting a client's staff from an HR list\n\n![Koji Import Respondents dialog with a CSV upload and a personal interview link for each person](/email/shots/shot-csv-import.jpg \"Each person on the client's list gets a private link, so you can send reminders to those who haven't answered.\")\n\n**Which plan you need.** Voice and text interviews, multilingual studies, structured questions, context files, guardrails, Remove personal details, duplication and report downloads are on every plan, including Free, which comes with 10 credits at signup: enough to pilot the frontline study. A text interview costs 1 credit and a voice interview 3. Recruitment email needs a paid plan (Insights, Interviews or Enterprise), private links from a CSV need Interviews or Enterprise, and your consultancy's own branding on the interview needs Enterprise. See [pricing](/pricing) for current plans.\n\n## What you get\n\nA report per layer. Each finding shows the share of participants behind it and opens the interview quotes that support it, so every line in your readout has an evidence trail. Each structured question has its own chart: usage frequency, the barriers ranked, the confidence score. The personas tab groups people by how they work with AI. Only interviews scoring 3 or more on Koji's 1-to-5 quality score enter the report.\n\n![Koji report summary with a highlighted claim backed by 11 participants (26%) and a side panel of supporting interview quotes](/email/shots/shot-evidence-panel.jpg \"Click any claim to see the quotes behind it before it goes in the steering committee deck.\")\n\nHow consultants use it the next day:\n\n- **In the steering committee deck,** download each report as an editable PowerPoint deck and build a readiness heat map: function by barrier, with the share of each layer raising it and one quote per cell.\n- **For the roadmap,** the answers to \"when the approved tools don't help\" become your use-case backlog, ranked by how many people already do it unofficially.\n- **For the change plan,** the manager layer's answers tell you which enablement to fund first: policy clarity, training or workflow redesign.\n- **For the client sponsor,** invite them as a workspace viewer. Viewers can read every report and download exports without taking a paid seat.\n\n## The same playbook elsewhere\n\nThe three-layer read works wherever a change depends on people further down than the sponsor. HR teams use a version of it to [hear the people who stayed after a layoff](/blog/layoff-survivor-interviews-hr-playbook). Business analysts use it to [decide which work to hand to an AI agent](/blog/ai-agent-workflow-triage-business-analyst-playbook). Product managers use the frontline layer to [find out why users don't trust an AI feature](/blog/ai-feature-adoption-research-product-managers). And agencies use the same week-one rhythm for [new-client stakeholder discovery](/blog/new-client-stakeholder-discovery-agency-playbook). Every Koji playbook for consultancies is collected at [/use-cases/consultancies](/use-cases/consultancies).\n\n## Why Koji for AI readiness interviews\n\nHere is how the realistic ways of hearing the workforce compare on what matters in a fixed-fee assessment.\n\n![Table comparing consultant-led calls, readiness surveys, workshops and Koji on reach, depth, consistency, candour, languages and analysis effort.](/visuals/ai-readiness-assessment-methods-compared/ai-readiness-assessment-methods-compared.png \"Use each method for what it does best. Checked November 2026.\")\n\n|                           | Consultant-led calls         | Readiness survey        | Workshops                       | Koji (us)                              |\n| ------------------------- | ---------------------------- | ----------------------- | ------------------------------- | -------------------------------------- |\n| People heard in two weeks | 15 to 25, limited by diaries | Hundreds                | 30 to 60, in groups             | As many as you invite, in parallel     |\n| Depth                     | High                         | Low: scores, no reasons | Medium, loudest voices dominate | Follow-up probes on every answer       |\n| Consistency               | Varies by consultant         | High                    | Low                             | Same guide and probes every time       |\n| Candour on shadow AI      | Low with a stranger on video | Medium                  | Low in front of peers           | Private, with personal details removed |\n| Languages                 | Your team's languages        | Translated form         | Your team's languages           | More than 30                           |\n| Analysis effort           | Days of note coding          | Charts, no reasons      | Sticky notes to type up         | Themes with quotes and charts          |\n\nOther AI-moderated tools court consultancies too. Usercall publishes monthly plans and runs voice interviews with change and AI-readiness consultancies in mind; Outset quotes per contract and offers managed services. Both are credible options, and you should trial more than one.\n\nWhy we think Koji fits this job:\n\n1. **Evidence you can defend.** Every finding links to the quotes behind it, which answers the client who asks where a claim came from.\n2. **Confidentiality built into the study.** Remove personal details strips names, contact details and employers from transcripts before analysis.\n3. **One design, many layers.** Duplicate a study per layer or per country and keep the guide identical, in more than 30 languages.\n4. **Client-ready output.** Reports download as editable PowerPoint decks or PDFs, and client sponsors can read them as free viewers.\n5. **You can start today.** A free account with 10 credits is enough to pilot the frontline guide on the client's own team before the steering committee.\n\nThe [/for/agencies](/for/agencies) page shows how consultancies and agencies run client work in Koji.\n\n## Questions you might have about AI interviews\n\n**Will client employees speak candidly to an AI?** Candour depends on what people believe happens to their answers, so make the promise specific. Say who will see what, in the sponsor's invitation. Keep groups large enough that nobody can be singled out: never report a group smaller than five. Each person answers alone, at a time that suits them, with no colleague or consultant on the call, and personal details are removed from the transcript when the interview ends.\n\n**Will the client accept AI-moderated evidence?** Present it as what it is: structured interviews, with the method, the numbers per layer and the quotes. Every claim traces back to what people said, and low-quality interviews don't enter the report.\n\n**How is client data handled?** Koji offers a pre-signed GDPR Article 28 data processing agreement with EU Standard Contractual Clauses, at [/compliance/dpa](/compliance/dpa). Upload only the context files the interviewer needs.\n\n## When Koji may not be the right fit\n\nIf the assessment is mainly technical (data architecture, model governance, security controls), you need document review and specialist interviews, not 100 conversations. If the client has under 40 people, talk to them all yourself. And if a works council or union has to be consulted on the AI rollout, AI interviews can inform that consultation but do not replace the formal process.\n\n## Metrics to track\n\n- **Response rate per layer,** against a target you agree with the sponsor before fielding. A low rate in one function is a finding in itself.\n- **Median interview length** against the 15 minutes the guide is built for. Much shorter means people are skimming.\n- **Share of interviews in the report** after quality scoring.\n- **New themes per five interviews,** to know when a layer is done.\n- **Gap between layers** on usage and confidence: the headline of most readouts.\n- **Re-run in six months:** the same guide shows whether the roadmap moved anything.\n\n## FAQ\n\n**What is an AI readiness assessment interview?**\nA structured interview with an employee about how they use AI in their work, what blocks them and what would need to change. A set of them across every layer of the organisation forms the evidence for an AI readiness assessment.\n\n**How many interviews does an AI readiness assessment need?**\nOur rule of thumb is 5 to 8 sponsors, 8 to 12 managers per function and 10 to 15 per frontline role family, stopping when five interviews in a row add no new theme. For a 1,200-person client with four functions in scope, that is roughly 80 to 120.\n\n**Should we still run a readiness survey?**\nYes, if you need a number for every team. Use the survey for breadth and the interviews for the reasons behind the scores. The interview guide here includes three structured questions, so you get a number next to the reasons.\n\n**How long does it take?**\nPlan two weeks from brief sign-off to synthesis inside a four-week engagement: three days for design and sponsor interviews, a week in the field, and two days to read and synthesise.\n\n**Can we brand the interviews as our consultancy?**\nBrand themes with your logo and colours are on Enterprise plans. On other plans, the client sponsor's invitation carries the context and your name.\n\n## Start your first AI readiness study\n\nStart a free study from the **Change Impact Assessment** template, or paste the one-line brief from this playbook into \"What do you want to research?\", and pilot the frontline guide before your next steering committee. To see how consultancies and agencies run client research in Koji, visit [/for/agencies](/for/agencies). For the in-house version of this research, read [Employee AI adoption research](/docs/employee-ai-adoption-research), and for pairing it with a survey, see [measuring change readiness with employee surveys](/docs/change-management-survey-guide).\n","category":"Use Cases","lastModified":"2026-11-17T07:50:10.846301+00:00","metaTitle":"AI Readiness Assessment Interviews: Consultancy Playbook","metaDescription":"Run AI readiness assessment interviews across every layer of a client in two weeks: sample design, a question guide to copy and the setup in Koji.","keywords":["ai readiness assessment interviews","ai readiness assessment","ai readiness stakeholder interviews","ai readiness audit","consulting stakeholder interviews","employee ai adoption interviews","change readiness interviews"],"aiSummary":"A playbook for consultancies running AI readiness assessments. It recommends a three-layer read: 5 to 8 live sponsor interviews, 8 to 12 AI-moderated manager interviews per function and 10 to 15 per frontline role family, fielded in days 3 to 10 of a four-week engagement. It gives a nine-question guide, cites MCA 2026, Gartner 2025 and 2026, BCG 2026 and McKinsey 2025 data on AI use and change adoption, and shows the Koji setup: Change Impact Assessment template, context files, structured questions, voice or text, Remove personal details and one study per layer.","aiKeywords":["ai readiness assessment interviews","ai readiness assessment questions","how many interviews ai readiness assessment","consultancy ai readiness","three-layer read","shadow ai interviews","ai moderated interviews consulting"],"aiContentType":"guide","faqItems":[{"answer":"A structured interview with an employee about how they use AI in their work, what blocks them and what would need to change. A set of them across every layer of the organisation forms the evidence for an AI readiness assessment.","question":"What is an AI readiness assessment interview?"},{"answer":"Our rule of thumb is 5 to 8 sponsors, 8 to 12 managers per function and 10 to 15 per frontline role family, stopping when five interviews in a row add no new theme. For a 1,200-person client with four functions in scope, that is roughly 80 to 120.","question":"How many interviews does an AI readiness assessment need?"},{"answer":"Yes, if you need a number for every team. Use the survey for breadth and the interviews for the reasons behind the scores. The interview guide here includes three structured questions, so you get a number next to the reasons.","question":"Should we still run a readiness survey?"},{"answer":"Plan two weeks from brief sign-off to synthesis inside a four-week engagement: three days for design and sponsor interviews, a week in the field, and two days to read and synthesise.","question":"How long does it take?"},{"answer":"Brand themes with your logo and colours are on Enterprise plans. On other plans, the client sponsor's invitation carries the context and your name.","question":"Can we brand the interviews as our consultancy?"}],"relatedTopics":["ai readiness assessment","change readiness","stakeholder interviews","employee ai adoption","consulting diagnostics"]}],"pagination":{"total":1,"returned":1,"offset":0}}