How To Use LavaReach AI Research in Email Prompts
Table Of Contents
- LavaReach AI Research
- Using Researched Data in Outreach Sequence
- Monitoring Outputs in Active Campaigns and Edge Cases
If you're using LavaReach's AI Research Enrichment, you are probably saving yourself hours of manual research. To take advantage of the research output even more, you can include the findings in your LinkedIn and Email sequences to personalize your outreach at scale.
LavaReach AI Research
If you haven't tried using AI to research prospects, give it a shot! For example here we are using the LavaReach AI to research if the given companies have facilities in Mexico. You can type the question in natural language and give it details about how the research should be outputted.
Here's the output in the list view:
To use this output in your outreach, you can use AI once again in the email templating step to format the research in a way that the prospect on the other side can understand.
The key is to always ensure your prospects feel that you've done the research yourself, and are genuine when reaching out to them.
Using Researched Data in Outreach Sequence
The first thing to know is that once you use AI Enrichment to create a new column, the column header becomes a dynamic variable that you can pull into the sequence editor.
From our previous example, the column header created is “Facilities in Mexico?”. This now becomes the variable {{Facilities in Mexico?}} on the sequence templating side for both LinkedIn and Email sequences. (Since AI Research is enriched at a list level, make sure to select the list which you've enriched the research and select contacts to preview outputs).
The variable {{Facilities in Mexico?}} outputs exactly what you would see in the enriched column in the list view. Obviously, we wouldn't send this to a prospect in its current form – that would be a pretty strange message to receive.
We want to use AI Prompting in the template builder to format the research into information that the prospect can respond to. Let's try the following prompt:
Left side: anything in purple is prompt, and you can wrap variables within prompts
Right side: output preview for John Witkowski from the Medical Devices list
Hmmm, the preview is showing quotations, and it sounds too robotic and disingenuous.
Let's try again:
This time, I've added 2 instructions in the prompt and now it sounds less robotic because we are not greeting the prospect. We've also solved the quotation problem.
However, if the prospect received this message they might find it too abrupt and out of the blue. Let's tell the prompt to mention that we've done some research and that's why we are reaching out.
Let's try again:
I like this one, but here's what's still missing:
- No friendly greeting
- We haven't introduced what we do
- We are missing a CTA!
We don't need AI to do this for us; we can just use some static templates and variables.
Ok, here we go. There's not much to test here because the output will just perfectly match with the variables. Since this is a LinkedIn message, let's just fix the formatting by collapsing the spaces:
Ok, I'm finally happy with this. We have a greeting, a personalized hook with researched information, a quick intro on our value prop, and a CTA.
Make sure to click "Save Changes" at the top right corner. Finally, add a few follow ups in the sequence, and you're done!
Overall this would have taken me 5-10 mins, and it might take you a bit longer if you're unfamiliar with this process. Nevertheless, keep in mind you only have to do this once, and all your outreach could be hyper-personalized like this one.
Monitoring Outputs in Active Campaigns and Edge Cases
As you monitor your campaigns, you may notice that some messages don't sound as good as this one, and you can always come back here to tweak the instructions and edit. You can also select more contacts from the list to preview outputs until you feel like you've covered a good amount of variability.
Yuri - Company Name Too Long
For example, if I test with Yuri, the company name here is very long, and it's a tell to the prospect on the other side a robot wrote this.
Within our prompt libraries we have some presets, and one of them is “Company Name Abbreviation”
Prompt library will drop down a list of prompts commonly used that you can borrow!
Circled in orange is a prompt preset that abbreviates any long company names. Note I had to add a 3rd guideline in the previous prompt where I use research as well.
Nathan - AI Citing Sources
For Nathan, the AI is saying where we are getting the research from. Sometimes this could be great, but for this campaign I don't want to mention it so I'll provide a 4th guideline in the prompt to ensure this doesn't happen.
Here's the output. Now this edge case has been covered.
Tait, Michael, Clint - Looks All Good!
I tested with 3 more contacts, and nothing looks weird. I think we've covered a good amount of edge cases for this campaign and I feel comfortable running it now. Keeping in mind that edge cases may still occur, we need to actively monitor our campaigns and LinkedIn outbox to reiterate on the template if need be!
We understand that this could be a steep learning curve. Reach out to us founders: Yiming (CTO) or Daniel (CEO) at LavaReach any time if you're stuck and we'll help you out!
About Daniel Zhao
Daniel Zhao is a multiple time founder with years of first-hand experience in B2B sales and revenue leadership. He has a consistent track record of helping companies experiment and implement outbound in SaaS and other industries. Throughout his career, Daniel has set up numerous outbound motions for the first time for companies that previously had not found success with sales led customer acquisition.