Automation and AI can only work reliably when the underlying information is accurate, consistent and appropriately governed. A data-health review helps a charity understand what it can trust and what needs attention first.
Charities are increasingly exploring how Salesforce automation, analytics and Agentforce could reduce administration and improve the way teams support donors, volunteers and beneficiaries.
The opportunity is real, but the quality of the result depends heavily on the quality of the information underneath it. An automated process can move incorrect data more quickly. An AI agent can retrieve an out-of-date answer with complete confidence. A dashboard can present a precise figure that is still wrong.
Before expanding automation or AI, a charity needs an evidence-based view of its data: what is reliable, where the risks sit and which improvements will create the strongest foundation.
Why charity data becomes difficult to manage
Relevant information often arrives from many different places. Donations may come through digital campaigns, telephone fundraising, door-to-door activity, events, lotteries and external agencies. Volunteer, grant, service and membership information may live in separate systems or spreadsheets.
Each source can use different formats, identifiers and rules. Over time, that creates duplicate supporters, inconsistent addresses, missing consent information, unreliable income coding and manual reconciliation.
The problem is not simply untidy data. It affects how the organisation treats people and makes decisions.
- A supporter may be thanked twice or contacted against their preferences.
- A donation may not be matched to the correct person or campaign.
- Gift Aid opportunities may be missed or claims may need additional checking.
- Fundraisers may spend time reconciling files rather than developing relationships.
- Reports may give leaders different answers depending on who produced them.
- Automation may trigger the wrong action because an important field is incomplete.
What should a charity assess
A useful data-health check should examine the information in the context of the charity’s work. Sweet Potato Tec’s approach covers 13 areas across four connected groups.
Supporters and income
- Contacts and households, including duplicates, uncontactable people and deceased records.
- Donations and Gift Aid, including claim validity, unclaimed income and coding.
- Recurring giving, including lapsed payments, mandates and processor synchronisation.
Compliance and operations
- Consent and GDPR, including opt-ins, suppressions and retention rules.
- Cases, including ownership, ageing and appropriate access to sensitive information.
- Volunteers, including training, DBS information and relevant consent.
- Grants, including restricted income, milestones and due diligence.
Data foundations
- Standardisation of names, addresses, picklists and placeholder values.
- Completeness of critical fields used for processes and reporting.
- Duplicate controls, validation rules, rollups and audit history.
Connections and control
- The freshness and reconciliation of integrations and incoming feeds.
- Governance, including owners, standards, access and the system of record.
- Readiness for Nonprofit Cloud, Data 360, automation and AI agents.
A review should start with evidence the charity can share
Data assessment does not have to begin with unrestricted access to personal records. It can start with a questionnaire, client-run checks, metadata or a guided screen-share. More detailed analysis can then use anonymised or record-level information where it is appropriate and properly governed.
The important point is to label the strength of the evidence. A self-reported answer is useful, but it should not be presented as though it has been verified directly in the data.
The resulting assessment should provide more than a list of problems. It should explain what each issue means, the risk or cost it creates and the practical order in which it should be addressed.
From fragmented files to one reliable supporter view
A UK animal-welfare charity provides a useful example of the scale of the challenge. More than 50 agencies and fundraising channels were supplying information, with over 90% of those feeds handled manually. Teams were spending more than 200 hours each week managing fundraising files, while duplicate records and data errors disrupted campaigns.
Sweet Potato Tec developed an automated supporter-data engine that collects incoming files, validates them, matches and de-duplicates supporter records, standardises the information and brings it into Salesforce and CRM Analytics. Donations are matched and reconciled through FinDock.
The published outcomes include:
- Manual data effort reduced by approximately 80%, from more than 200 hours a week to fewer than 40.
- Campaign-disrupting errors reduced from more than 15 a month to zero.
- Duplicate and error records reduced by approximately 75%.
- New agencies and channels brought live in under one day rather than weeks.
- 15.73 million supporter transactions handled automatically each year.
- More than 170 fragile data links replaced by one reusable engine.
The value is not only cleaner information. Fundraisers receive useful data sooner, supporters are less likely to receive inappropriate communications and the charity has a stronger platform for analytics, automation and future AI use.
Why AI readiness is a governance question
Reliable data is necessary for AI, but it is not the only requirement. The charity must also decide which information an agent can access, which actions it may perform and when a person must review or take over.
Sensitive beneficiary, safeguarding and supporter information should not become available to an agent simply because it exists in Salesforce. Permissions, purpose, escalation and monitoring all need to be defined.
A data-health review therefore needs to assess governance as well as cleanliness. The organisation should know who owns important data, which standards apply, how issues are corrected and how automated decisions will be monitored.
A practical route from data health to innovation
- Assess the information and the strength of the available evidence.
- Fix non-negotiable risks involving consent, security, income or critical processes.
- Agree common definitions, owners and standards.
- Improve the most valuable integrations and recurring data feeds.
- Establish reliable reporting and monitoring.
- Introduce a focused automation or AI use case with clear human oversight.
- Measure the result and expand only when the foundation is working.
This approach avoids treating data preparation as an endless exercise. The objective is to make the information reliable enough for the next useful outcome, while creating controls that prevent the same problems returning.
Know what you can trust before you automate
Automation and AI can create valuable capacity for charities, but they amplify the information and rules they are given. The safest starting point is therefore not a product demonstration. It is understanding the quality, ownership and readiness of the data the organisation already holds.
Sweet Potato Tec helps charities assess data health, prioritise improvements and build connected data foundations across Salesforce, fundraising platforms, finance systems and other operational tools. That creates a more reliable base for reporting today and responsible automation tomorrow.