By KYN AI Advisory Team — AI implementation specialists, Singapore
To build a target account list from a registry export, load the file, clean the names and statuses, filter to live entities that match your customer profile, enrich the survivors with website and hiring data, score them against written rules, and have a person review the top of the list. Registry data tells you who exists. Your rules and the enrichment decide who is worth contacting. Expect to discard most of the file at the filter step.
The steps below work for any country that provides a file. Singapore publishes open datasets on data.gov.sg. Other registries may require a paid extract or restrict bulk use. Read the terms first. See company data sources for Southeast Asia.
What does a registry export contain?
Usually a flat table: entity name, registration number, entity type, status, registration date, a primary activity code and a registered address. Some include former names or a postal code.
It does not contain headcount, revenue, a website, an email address or a phone number. The export is a skeleton. You add the rest.
Check the file before you start:
- Row count and the date the file was produced.
- Duplicate registration numbers.
- Blank or malformed status values.
- Encoding. Names with Chinese characters or accents often break in a spreadsheet.
Keep the raw file unchanged. Work on a copy so you can always trace a row to the source.
Step 1: How do you clean the file?
Cleaning removes noise that would cause bad matches later.
- Normalise names. Lowercase, strip punctuation, collapse spaces, remove legal suffixes into a separate field.
- Standardise status. Map the registry's labels to three: live, inactive, unknown.
- Parse dates. Convert the registration date to ISO format. Compute age in years.
- Split the address. Pull out the postal code and area. You may use it to filter by district.
- De-duplicate. One row per registration number.
Log how many rows each rule changed. If a rule touches more than a few percent, look at why.
Step 2: How do you filter to your customer profile?
Write the profile as rules before you look at the data. Otherwise the file will persuade you of a profile that fits it.
Examples of rules an SME seller can apply from registry fields alone:
- status is live;
- entity type is a private limited company, not a sole proprietorship, if you sell to companies;
- registered at least two years ago, if you need an established operation;
- primary activity code is in your list of industries;
- registered address is in your service area.
Each rule is a yes or no with a reason. Keep a count of how many rows each rule removed. A common result is that 90 to 99 in 100 rows fall out. That is the point of this step.
Step 3: How do you add a website and size?
The registry gives no website. You find one, then confirm it belongs to the entity.
- Search by name and country, then check the site footer or contact page for the registration number or the legal name. A match on the number is a confident match.
- If no number appears, require two other matches, such as address and a product line that fits the activity code.
- If there is no confident match, mark the row "website not found" and move on. Do not guess.
From the website, record: what the company does, the locations, whether it has a careers page, and the count of open roles. See how to research a lead with an agent for the record format. Run this for the filtered list only, since it costs time or money per row.
Step 4: How do you score the list?
Scoring ranks rows so a person looks at the closest-matching accounts first. Use points you can explain.
| Rule | Points | Source | |---|---|---| | Industry in core list | 3 | Activity code and website | | Open role matching your offer | 3 | Careers page | | New entity or address change in the last 12 months | 2 | Registry | | Operates in more than one country | 2 | Website | | Website found and matched by number | 1 | Matching step | | No website found | -2 | Matching step |
The weights are yours to set. Start simple, review the top 50, and adjust. The aim is a ranking a salesperson can read and challenge, not a hidden score. Each point has a reason you can show. See buying signals for SMEs for signals to add.
Step 5: What should a person review?
Review the top of the list before any message is drafted.
- Read the top 50 rows. For each, open the source link and confirm the entity and the claim.
- Mark each row keep, drop or unsure, with a reason.
- Use the drops to tighten a rule. If six dropped rows share a cause, write the cause as a filter.
- Only then move the kept rows to the outreach queue.
A review of 50 rows takes a person about an hour at first and less as the rules improve. The time spent here is cheaper than a campaign to the wrong list.
Step 6: What do you keep and what do you delete?
- Keep the registry fields, the source URL and the date for each row.
- Keep the score and the rule that produced each point.
- Delete personal data you do not need. Officer names are personal data. For outreach, a role is enough, and the company record can stand alone.
- Re-run the status check before each campaign. Entities close.
Singapore's PDPC PDPA overview sets the rules on personal data. Other markets have their own.
What are the common mistakes?
- Matching on name alone. It sends outreach to the wrong company.
- Treating the registry as a prospect list. Most entities are dormant, holding vehicles or one-person firms outside your target.
- Scoring without checking. A score the team cannot challenge will not be trusted.
- Skipping the freshness check. The file was a snapshot on its date.
- Automating access against the terms. Use the open dataset or the documented API.
Where KYN fits
KYN builds list-building and research agents that run in the client's own cloud account. The registry file is loaded into the client's database, the steps above run as logged jobs, and each score carries its reason. Agents use their own restricted accounts. Outbound messages wait for a named person.
The first 14 days are a review of your current export files and how your team ranks accounts today. A full build with the list pipeline ships in under 90 days. See sales systems and marketing systems.
Frequently asked questions
How many accounts should a target list have?
Enough for your team to work properly. For a small sales team that is often a few hundred accounts per quarter. A short, well-matched list beats a long one that nobody can review.
Can I use a registry export for cold email?
You can use it to identify companies. You still need a lawful, sourced business contact before you send, plus an unsubscribe route. See how to find business emails legally in Singapore.
How often should I rebuild the list?
Re-check status before every campaign and rebuild the list each quarter. New entities and closures change the file continually.
Checked 30 October 2026.
Curious whether Marketing agents fits your business? Talk to KYN on WhatsApp — no forms, just a conversation.
