AI Data Foundation Audit

AI Data Foundation Audit

Use When

• Use this skill when the requested outcome is specifically a evidence-backed audit report and the supplied brief falls within ai data foundation audit.

Do Not Use When

• Use ai-readiness-diagnostic when its narrower output is the real deliverable; do not use this skill as a generic substitute.
• Do not use it to publish, send, spend, alter a live account, or make unsupported legal, platform, performance, or certification claims.

Required Inputs

Capability and Permission Boundaries

Default to read-only: inspect supplied material and report findings. Editing, publishing, contacting people, spending, or changing live systems requires separate explicit authority. Minimum capabilities are read access to supplied files and search across the authorised evidence set. Use only the files, tools, accounts and evidence made available for the engagement, expose every unassessed check, and obtain explicit authority before any mutation.

Degraded Mode

Fallback: if files, network access, platform data, language review or production tools are unavailable, return the narrowest useful qualified evidence-backed audit report; mark unavailable checks not assessed and never convert them into a pass.

Decision Rules

Workflow

1. Confirm the exact evidence-backed audit report, consumer, market, channel and approval boundary; route to ai-readiness-diagnostic if it is the closer match.
2. Inventory supplied facts, source provenance, constraints and missing inputs; stop if the objective, audience or authority is unknowable.
3. Select the domain method and record the material decision behind it before drafting.
4. Produce the smallest complete evidence-backed audit report; keep facts traceable and placeholders visibly unresolved.
5. Test the result against the decision table, domain quality criteria and anti-slop gate; recover by narrowing or qualifying unsupported portions.
6. Deliver the artefact with evidence, assumptions, unassessed checks and the next approval or verification step.

Outputs

Evidence Produced

Quality Standards

• Preserve the domain guidance and East African market context below; replace it only when the requester names another market.
• Use British English unless the target language or market requires otherwise, and verify names, figures, quotations and platform rules before use.
• Make the key choice visible, cover failure and edge cases, and keep the result ready for its named consumer.
• Run the repository's anti-ai-slop ship gate; a blocking factual, cultural, safety or permission defect stops release.

Anti-Patterns

• Writing before the objective and audience are known. Fix: stop and obtain the missing brief fields.
• Reusing a neighbouring skill's template because the headings look similar. Fix: route by the requested evidence-backed audit report, not vocabulary overlap.
• Adding a price, result, quotation, platform limit or cultural claim without a traceable source. Fix: verify it or qualify/remove it.
• Treating missing access, evidence or native-language review as approval. Fix: mark the check not assessed and narrow the result.
• Publishing, sending, spending or changing a live account from drafting authority alone. Fix: obtain explicit action-specific authority and retain the approval record.

References

• ai-readiness-diagnostic is the nearest routing comparison for this skill.
• Repository agent guide defines the engine-wide market, safety and anti-slop gates.

Required Input

Ask for the following before generating any deliverable:

1. Client business name — trading name as it appears to customers
2. Industry — e.g. retail, hospitality, healthcare, professional services
3. Country/city — default: Uganda / East Africa if not specified
4. All customer data sources currently in use — tick all that apply: WhatsApp chat history | Facebook Page DMs and comments | Email (Gmail / Outlook) | CRM software (name it) | Excel / Google Sheets | Accounting software (e.g. QuickBooks, Wave) | Other (describe)
5. Primary AI use case the data will support — select one: Content personalisation | Chatbot knowledge base | Predictive analytics | Audience segmentation
6. Approximate number of customer records — total across all systems (even a rough estimate is useful)

Why Data Quality is Step 1

Three independent books converge on the same conclusion: data quality is the single most important prerequisite for any AI marketing investment (Venkatesan and Lecinski, 2026; Lamplugh, 2024; Ltifi, 2025). AI tools amplify what they receive — clean data produces better outputs; dirty data produces worse outputs at speed.

For East African clients specifically, the data challenge is structural. Customer data is fragmented across WhatsApp chat histories, Facebook Page DMs, manual Excel spreadsheets, phone contacts, and verbal records. Before any AI tool can add value, this fragmentation must be addressed.

The investment in data hygiene is not a technical exercise — it is the commercial prerequisite for every AI use case. Skipping this step does not save time; it wastes the cost of the AI tool.

The EA Data Reality

Map the typical Ugandan SME data environment honestly before proceeding:

• WhatsApp — customer enquiries, orders, and complaints are stored in chat history; rarely exported or structured; impossible to query or analyse at scale
• Facebook Page — post comments, DM threads, and page insights are exportable via Meta Business Suite but rarely exported; valuable for sentiment and engagement data
• Email — enquiries and newsletter replies sit in Gmail or Outlook, rarely in a CRM; no linkage to purchase data
• Excel spreadsheets — customer names, phone numbers, and purchase history recorded manually; often incomplete, inconsistent formatting, and not regularly updated
• Verbal / in-person — walk-in customers and cash transactions frequently go unrecorded entirely
• Result — the same customer may appear in three or four separate systems with different name spellings, no unique identifier, and no purchase history linkage. AI tools cannot perform matching, personalisation, or segmentation on this data without prior consolidation.

Data Hygiene Checklist (20 Items)

Complete this checklist for the client. Score each item Yes / No / Partial. Flag any No or Partial item as a remediation priority.

Completeness

• Are all customer records complete — name, contact detail, and transaction history at minimum?
• Are there records with missing phone numbers or email addresses?
• Are product and service names consistently recorded — not "cake" in some rows and "birthday cake" in others?
• Are dates recorded in a consistent format (DD/MM/YYYY throughout)?

Accuracy

• Are phone numbers in the correct format for Uganda (+256 or 0)?
• Are prices recorded in UGX consistently — not a mix of UGX and USD?
• Are customer names spelled consistently across all systems?
• Are inactive or churned customers flagged and separated from active customer records?

Consistency

• Is each customer recorded once across all systems — not duplicated across WhatsApp, Excel, and email?
• Are product categories consistent — not "electronics" in one place and "gadgets" in another?
• Are campaign names consistent across Meta, email, and CRM records?
• Are staff names or salesperson records consistent for attribution purposes?

Accessibility

• Is data stored in a format the AI tool can read — CSV, Excel, or plain text — not only in printed records or chat messages?
• Can the data be exported from its current system? Some WhatsApp and Facebook data requires manual export steps.
• Is the data accessible to the team members who will operate the AI tools?
• Are data access credentials documented and stored securely?

Governance

• Is there a named data owner — one person responsible for data quality?
• Is there a documented process for adding new customer records?
• Is customer data held in compliance with Uganda's Data Protection and Privacy Act (2019)? Confirm that consent has been obtained and data is not shared without permission.
• Is there a data retention policy — documenting how long customer records are kept and when they are deleted?

Data Mapping Framework

Follow this step-by-step process to consolidate fragmented data into a single master customer list:

1. List all data sources — WhatsApp, Facebook, email, Excel, accounting software, and any others identified in the Required Input.
2. For each source, record: what data it contains; its format; estimated record count; date last updated.
3. Identify the primary identifier — the field that will link records across systems. Phone number is the most reliable identifier in Uganda, because it is more consistent than email across both urban and peri-urban customers.
4. Export all sources to Excel or CSV — request exports from WhatsApp (chat export), Meta Business Suite (contacts and insights), Gmail (contacts), and any CRM or accounting software in use.
5. Standardise the phone number field across all exports — remove spaces, remove hyphens, ensure all numbers use the +256 international format.
6. Use VLOOKUP or a de-duplication tool to identify records appearing in multiple sources. Google Sheets has a built-in Remove Duplicates function; Excel has a Remove Duplicates tool under the Data tab.
7. Build a master customer list — one row per customer, linking all data points: name, phone number, email (if available), source systems, last transaction date, and any segment tags.

30-Day Remediation Plan Template

Apply this plan after the hygiene checklist is complete and the top three issues are identified:

Week 1 — Map and assess Complete the data mapping framework. Run the hygiene checklist for all 20 items. Identify the top three issues (typically: missing contact details, duplicate records, and inconsistent product naming). Assign a named owner for each issue.

Week 2 — Fix completeness Fill missing contact details by following up with customers via WhatsApp or in-person. Add missing transaction dates from accounting records or receipts. Ensure all product and service names follow a single agreed naming convention.

Week 3 — Fix accuracy Standardise all phone numbers to +256 format. Correct misspellings in customer names. Merge duplicate records into the master list. Flag and separate inactive customers.

Week 4 — Implement governance Assign the named data owner formally. Document the new record process: how new customers are added, which fields are required, and who is responsible. Confirm Uganda Data Protection and Privacy Act (2019) compliance — consent records in place, retention policy documented. Export the final master customer list and confirm it is ready for AI tool import.

Output: a clean master customer list ready for import into the AI tool of choice, plus a documented data governance process that prevents regression.

Tool Options

For most Ugandan SMEs, Google Sheets is the recommended starting point: it is free, collaborative, accessible on mobile, and sufficient for up to several thousand customer records.

Handoff — Connecting Clean Data to AI Tools

Once the audit and remediation are complete, connect the master customer list to the relevant AI tools:

• RAG knowledge base (see ai-rag-brand-knowledge-base) — upload the product catalogue and FAQs derived from the clean data
• Chatbot knowledge base (see ai-whatsapp-chatbot-design) — upload approved response templates and policies built from the structured data
• Segmentation tool — import the master customer list with segment tags for audience targeting
• Predictive analytics tool (see ai-predictive-analytics-social) — import transaction history and engagement history for churn and upsell modelling

Do not connect any AI tool until the Week 4 governance step is complete. A tool connected to unclean data will produce unreliable outputs and undermine client confidence in AI marketing.

Quality Criteria

Good output from this skill meets all of the following standards:

• All customer data sources mapped — no source overlooked, including informal sources such as verbal records and WhatsApp chat histories
• Data hygiene checklist completed in full — all 20 items scored Yes, No, or Partial with priority flags applied to every No and Partial item
• Data mapping framework completed — master customer list built using phone number as the primary identifier, with all sources consolidated
• Top three hygiene issues identified with a specific, actionable remediation step for each — not generic advice, but concrete next actions for this client
• 30-day remediation plan documented with named owner and weekly milestones — not a suggestion but a committed plan
• Data governance framework in place — named data owner, new record process documented, and responsibility assigned
• Uganda Data Protection and Privacy Act (2019) compliance confirmed — consent documented for all records, data sharing restrictions understood, retention policy set
• Clean data handed off to at least one AI tool with a verification step confirming that outputs have improved compared to pre-audit baseline

References

• Venkatesan, R. and Lecinski, J. (2026) The AI Marketing Canvas, 2nd edn. Stanford University Press.
• Lamplugh, M. (2024) The AI Marketing Playbook, 2nd edn. Mercury Learning.
• Ltifi, M. (ed.) (2025) Advances in Digital Marketing in the Era of Artificial Intelligence. CRC Press.