(Part Two) How to Automate the “Best Sales Prospecting Email Ever”

Oct 02, 2017

Personalized Copy at Scale

(This is Part Two of a series on automating outbound. Click here for Part One)

Brian Anderson had written what was dubbed the “Best Sales Prospecting Email Ever”. In Part One, we explored automating the custom image he’d created, and detailed the technology required.

But Brian was still spending five to ten minutes researching and writing each email to go with the now automated image. Our goal was to find a way to scale Brian’s level of personalization to reach up to 200 prospects, without spending 18 hours to write them one by one.

Deconstructing the email

We took the same approach as the custom image, deconstructing Brian’s email for points of personalization we could automate. But there was a problem: it was too personalized. Using Brian’s copy directly, we’d end up culling every prospect on our list who hadn’t blogged about the low ROI of conference leads (spoiler: none had).

But, what if we broke down each point of personalization and looked for data to replicate the intent vs. simply the words? Then we might be able to automate the process.

When advising clients, we use several questions to convert one-off personalizations to scaleable merge tags:

  1. Replace: Can we find this data from existing data sources?
  2. Proxy: If no, what company / people data can we use to inform personalization?
  3. Insight: If no, can we combine multiple data points to create a more impactful insight?
  4. Revise: Edit based on learnings from 1–3?

REPLACE: Can we find this data easily at scale?

The first datapoint referenced a prospect’s upcoming conference. We found out the folks at DataFox had an upcoming conferences API that would return upcoming and past conferences a company had attended, including metadata on the conference. With some quick coding, we were able to look at all the conferences a company was attending in the near future, and only return the one closest to today.

Next conference identification

By using the next conference name and date, we could replace Brian’s personalization with a merge tag:

PROXY: Can company/people proxy data be used to inform messaging?

We knew from using enrichment data for custom images, we’d have the prospects role and seniority. But there wasn’t an obvious way to use that for the personalization. However, looking at the closing line and “your team”, we could change the language based on the prospect’s seniority.

For example, if the prospect was a manager instead of an executive, we’d say “prioritizing your team’s time” vs “prioritizing your time”.

More importantly, since Brian’s software is targeted toward marketers who directly manage conference handoff, we could ask executives for referrals, while asking directors and managers for a meeting.

This was a good start on using company/person data, but we suspected we could find more opportunities to pivot data as we optimized for scale.

INSIGHT: Can multiple data points combined make a more impactful insight?

Brian had thrown us for a loop on this one: since none of our sample prospects had blogged about conference follow-up ROI, we’d have to convert Brian’s insight into data we could find and then convert back into an insight.

Backing into data needed for outbound

What Brian was actually saying: “Based on something I know about your business, you’re unlikely to get the most out of your conference investment” — that was the real Insight. The Analysis was based on a deep read of Charlie’s blog post (Raw Data).

So, what data could we analyze to reasonably say, “You’re unlikely to get the most out of your conference investment”? We went back to the DataFox Conference Intelligence data. By comparing Y/Y conference sponsorship, we could know whether a prospect was increasing or decreasing its conference investment. If increasing, we could assume they’d invest in follow-up. But if they were decreasing conference investment, we knew they would need to do “more with less” and ensure a white glove experience for each lead.

Determining conference investment velocity by historical trend

While we could now use this for the merge tag, it felt like something was missing. So, we analyzed the type of conference a company was investing in. Then we could reference the type of prospect they were aiming for, similar to Brian’s original reference.

REVISE: Edit based on learnings

With just a company’s domain and a prospect’s email address, we could find nearly a dozen data points to personalize the message: next conference, conference trend and types, seniority and title, attendee type and more. Incorporating this new insight, we edited the template to add more than fifteen merge tags.

While the new template retained the core of Brian’s original template, more than 70% of the final text could be automated without human involvement.

Value

We ran the math again, Brian could now generate 200 messages in less than two hours. Said another way, that’s 16.5x faster than the original message and a custom image explaining the value of Brian’s product.

A step above the typical, “Hi {{first_name}}, we want to work with {{company_name}}!” that floods prospect’s inboxes.

What’s Next

Pushing the limits of Brian’s email has been really fun. We challenged ourselves with this example because it wasn’t immediately obvious how to automate the content. Part One even inspired a new Visual Prospecting service where we’re helping clients identify and automate custom visuals for scale.

Stay tuned for our post in the coming week exploring the possibilities for personalized voicemail at scale.

Dogpatch Advisors is a research and advisory firm in San Francisco. Follow us on Medium or email us: hello@dogpatchadvisors.com | dogpatchadvisors.com