Customer success story · Education
CheckMates — Ireland
How CheckMates Got AI to Tell Its Story Accurately
How an Irish school chess provider used Kojable to clarify its positioning, organise educational content, and improve how AI describes its services.
- Potential time savings
- 72 hours*
- Potential labour savings
- $4,320*
- Content drafts generated
- 28
- To client-reported improvement
- 6 weeks
*Estimated from documented production activity, reported review time, and manual-work benchmarks. Central labour scenario in USD. See savings methodology.
What Kojable implemented
From educational questions to reviewed content
A connected workflow to explain the school offering, keep people in control and check what AI says next.
- 01
Topic discovery
Identify educational questions and subjects relevant to CheckMates, parents and schools.
- 02
Topic clustering
Group related questions into organised content clusters, with main questions and supporting subjects.
- 03
Content generation
Generate drafts explaining the educational approach, school services and areas of expertise.
- 04
Human review
Check facts, correct unsuitable descriptions and reject content that does not meet editorial standards.
- 05
Website integration
Pass accepted content toward the website through the documented webhook integration.
- 06
Follow-up assessment
Review subsequent AI descriptions against the company's positioning. The current audit came later.
Programme records: 9 June–26 July 2026, covering 48 calendar days. The customer feedback refers to six weeks of use.
The customer's perspective
“After just six weeks with Kojable, the way AI describes us has improved dramatically.”
The challenge
A school chess programme needs more than a generic chess description.
CheckMates provides instructor-led after-school chess clubs for primary-school children in Ireland. Its story centres on a structured educational approach: a 50/50 balance between instruction and gameplay, educational and developmental benefits, delivery in schools, and licensing options for schools that want to run their own clubs.
Parents and schools needed that context when researching the business through AI. Kojable's initial assessment reported that some descriptions overlooked important parts of the school offering and did not explain how the programmes worked.
An incomplete answer can leave a parent unsure whether a club suits their child, or a school unaware of its delivery options. The task was to make the actual service easier to understand and support it with relevant, reviewed content.
Recorded production activity
An organised content library, with a reviewer at the centre.
Topic clustering connected main questions with supporting educational subjects. The workspace covered school club formats, teaching methods, the coach's role and wider educational questions, rather than treating every draft as an isolated article.
| Metric | Recorded value |
|---|---|
| Keywords | 35 |
| Topic clusters | 27 |
| Keyword-to-cluster links | 62 |
| Content drafts generated | 28 |
| Clusters marked approved at export | 21 |
Human editorial review
Generating content was only one part of the programme. A reviewer checked whether each piece reflected the customer's actual services, corrected unsuitable descriptions and rejected off-topic material.
For example, saved positioning focused on chess learners of all levels and checkmate patterns. The website-supported service was after-school chess clubs for primary-school children. That distinction shows why human judgement mattered; the exact article edit is unavailable.
The cluster records included exploratory topics. Cluster approval does not establish that AI systems recognised topical authority, and generated drafts should not be read as a publication total.
| Count | Disposition | Evidence boundary |
|---|---|---|
| 17 | Records retained approved flags | After identified exclusions; final acceptance is unverified. |
| 9 | Identified rejections | Five in the export plus four additional rejections confirmed by Kojable. |
| 2 | Records remained in draft status | No final decision recorded. |
| Unknown | Corrected articles | The available history has no separate correction count or before-and-after revisions. |
The original export showed 21 approved and five rejected records. Kojable clarified that four of those approved records were also rejected. Corrections and acceptance can overlap; the accepted-without-edit versus corrected-and-accepted split cannot be recovered.
Current AI perception audit · 3 November 2026
What AI now says about the school offering.
The later domain audit sampled Google AI responses about CheckMates' identity and services. Its public interface did not accept custom buyer prompts or expose an exact model version. These are review criteria applied to the output, not a controlled before-and-after study.
Where AI descriptions align with the website
- School programmes
- AI describes instructor-led after-school chess clubs for primary-school children.
- Educational approach
- The 50/50 balance between instruction and gameplay matches the company website's lesson format.
- School licensing
- AI identifies an option for schools to run their own chess clubs using the CheckMates programme.
The snapshot corroborates specific points, not the whole answer. The current audit capability was not available during the original programme, according to Kojable. Accurate representation needs ongoing verification as services and products change.
Commercial value and efficiency
The potential value of a more efficient content workflow
For a business explaining a specialist service, organising questions, preparing drafts and reviewing them in one workflow can reduce production effort. The documented activity lets us model that potential value, while keeping the assumptions in view.
Documented and reported inputs: 28 content drafts, a reported 15-minute daily review routine and a 48-day recorded programme. Planning assumptions: three manual production hours per draft and assumed hourly labour rates.
- Assumed manual production28 × 3 hours = 84 hours
- Estimated review time48 × 15 minutes = 12 hours
- Potential time savings84 − 12 = 72 hours*
| Scenario | Hourly rate | Potential labour savings |
|---|---|---|
| Low | $45 | $3,240 |
| Central scenario | $60 | $4,320* |
| High | $75 | $5,400 |
*Potential labour value, not measured time saved, verified cash savings or net ROI. These USD scenarios are not CheckMates' actual costs or the price charged by Kojable. The rate range varies rates only. Read the assumptions and limitations.
Methodology FAQ
How to read the estimates
How were the time and cost savings estimated?
The programme generated 28 drafts. An assumed manual production benchmark of three hours per draft gives 84 hours. Applying the reported 15-minute daily review routine to the 48-day observation period gives 12 estimated review hours. The calculated difference is 72 hours of potential time savings.
The three-hour benchmark comprises 0.75 hours of SEO analysis and planning, 1.5 hours of writing and factual checks, and 0.75 hours of CMS preparation and publishing. These are planning assumptions. Actual manual production hours were not measured.
Valuing 72 hours at assumed rates of $45, $60 and $75 per hour gives low, central and high scenarios of $3,240, $4,320 and $5,400. The central scenario values manual work at $5,040 and review at $720. The client is based in Ireland; these are US-equivalent labour scenarios in USD.
The manual allowance includes publishing work for every draft, although rejected drafts were not published and the available records do not establish the final publication count. This may overstate publishing effort avoided. Review time is reported, not an independently timed study, and the dollar range varies rates rather than the manual-hours assumption.
Kojable fees and setup costs are excluded. The figures represent potential labour value, not verified cash savings or net ROI.
What the evidence supports
The supplied case study documents activity and reconciled review records; the quotation records the customer's experience. The 3 November audit supports limited observations about current descriptions. There is no controlled baseline, independently measured visibility uplift or verified causal effect connecting the programme to those later answers.
Source basis: the supplied CheckMates customer case study and saved AI perception audit. Original programme: 9 June–26 July 2026. Later audit: 3 November 2026. The customer feedback's six weeks and the model's 48 days describe different periods.
What this customer story demonstrates
Monitor → Diagnose → Improve → Verify
CheckMates' story brings the workflow into focus: understand how AI describes the business, identify missing or inaccurate information, organise relevant content and evidence, prepare improvements with human oversight, then review subsequent descriptions.
The commercial opportunity is a clearer account of what the business actually does, supported by a more organised production process. Customer feedback, recorded activity and later audit observations each contribute evidence. They do not establish an independently verified causal uplift.
Keep checking as company information changes. Explore Answer Intelligence to understand the monitoring and diagnosis workflow.