Lesson 1.4.2
Write the emails off the fields you enriched
What this costs
The copy tools run per row. Draft against ten rows before you generate against four hundred.
Before you start
A list with at least one research column on it, something the table knows that a database would not.
This is the lesson that decides whether the enrichment was worth buying. If your email would read identically with the research columns deleted, you paid for research and then wrote a generic email anyway. Write off the columns, and the difference is visible in one side by side comparison.
The test that settles it
Write the same email twice. Once from the company name alone, once from the enriched columns. Put them side by side. If they read the same, the column you bought was not worth buying, and that is a sourcing problem rather than a copy problem.
Write 4 emails. Email 1 opens on @signal_summary in one sentence, no compliment, no flattery. Email 2 references the same signal from a different angle. Email 3 is two sentences and one question. Email 4 is a two line break up. Subject lines under 4 words, lower case, no punctuation. Never use the words solution, leverage, synergy or exciting.
What a column earns its cost looks like
Worked example · The same opener, generic against enriched
| Company | Generic opener | Written off the columns |
|---|---|---|
| Northwind Logistics | I came across Northwind and was impressed | You have three SDR roles open in Rotterdam and none of them are filled |
| Peak Freight | I came across Peak Freight and was impressed | Your careers page has been advertising the same two roles since June |
| Atlas Shipping Group | I came across Atlas and was impressed | You opened a Hamburg office in March and the team page still shows four people |
The right hand column is only possible because somebody paid for a hiring signal and a headcount check in course 2.1. The left hand column is what you send when you skip that, and it is the reason most cold email gets no reply.
Every email opens the same way
- What you see
- The generated copy varies in wording but not in substance
- The fix
- The prompt did not reference a column. Name the column with an @ reference so the model reads the row.
Some rows produce nonsense
- What you see
- Openers referencing signals that are not there
- The fix
- Those rows have an empty or Not Found value in the referenced column. Filter them out before generating, or the model invents something.
Check yourself
0 of 2 answered1.How do you tell whether an enrichment column was worth buying?
2.Some generated openers reference signals that are not there. Why?
Try it yourself
Generate the same opener twice on ten rows, once from the company name and once from an enriched column, and put them side by side.
- You have done it when
- You can point at a specific fact in every enriched opener that the generic one could not have known.
- Credit budget
- Ten rows, twice.