How We Drove $1M+ in Pipeline with Clay, Strategic Gifting & Direct Mail

Imagine this scenario: You’ve taken full responsibility for *gulp* 100% of qualified pipeline generation for the business, and your outbound motion has been underperforming. You have a small but mighty Marketing team with aggressive goals, and the clock is ticking. You need to turn things around yesterday. 

For me, this wasn’t imaginary — it was very, very real. More and more marketing leaders are finding themselves in this situation, taking full ownership of qualified pipeline generation with daunting goals and inheriting SDR teams + the question: “Does outbound even work?”. Our six-person team had to get creative and find ways to scale high-quality pipeline generation without additional resources or budget.

I recently spoke at Pavilion’s GTM 2024 conference about how we optimized our outbound strategy, which is led by Katie Penner and her team of 3 BDRs, in order to generate over $1M of sales-qualified pipeline a quarter. This outbound approach just helped us exceed our Q3 pipeline and revenue goals for a record-breaking quarter.

This isn’t a fluffy thought leadership post. Read on for a tactical breakdown of exactly what the team does and how we do it. We hope it helps you hit (and blow past!) your goals.

Automation with Personalization

We hadn’t seen success with our previous outbound motion — a manual, costly process.

Earlier this year, I wrote about my top go-to-market bets, one of which was programmatic, AI-enhanced outbound. 

We believe that personalized, thoughtful human connection doesn’t have to be fully manual or 1:1 to be done well. By adding Clay and UserGems into our outbound toolset, we were able to forge connections through personalized gifting in a way that was genuine—not transactional—at scale.

The ROI of Gifting and Direct Mail

Before I dive into our playbook, I know some folks reading this may be skeptical about the impact of gifting and direct mail. HockeyStack’s report on the true ROI of gifting found that when used strategically, gifting:

• Boosts meeting rates by 3x
• Boosts second-call rates by 6x
• Increases win rates by 2x
• Reduces MQL costs by 43%
• Helps deals close 30% faster

We’ve seen these results firsthand while drinking our own champagne, and they’re validated by our own customer data. I can confidently say that gifting is a force multiplier for outbound (and inbound) qualified pipeline generation and a powerful way to accelerate deal cycles.

What is Clay?

Let’s get into the nitty-gritty of what Clay is, how it works, and one of our most successful plays, which you can duplicate right now.

If you aren’t familiar with Clay, here’s a quick rundown: it’s a data manipulation tool that allows you to combine data enrichment, workflow automation, and personalized, creative outreach into one seamless process.

Using a table structure, Clay takes the inputs you provide and helps sculpt the process you use to support your campaigns. Some of our data comes from various signals, intent score surges, and 6sense score changes.

You can plug in the data sources and enrichment tools you already have in your tech stack. We use SFDC and Marketo as sources, as well as Clearbit, Nimbler, ContactOut, and People Data Labs for enrichment and email validation. Clay uses a waterfall approach, where it will continue checking data providers until it finds the information it’s looking for — saving you valuable time.

Integrating Clay with SmartSend

At the beginning of the year, we launched our SmartSend feature, which allows you to search for a personalized gift either using the recipient's email address or by providing a short description of their interests. 

By integrating Clay with our SmartSend API, the team automated the entire process of sending personalized gifts to the highest impact prospects, so each contact in a list will automatically receive a message and gift or selection of gifts specifically tailored to them (plus the ability to exchange it if they find something they like better).

Here’s how we did it: 

1. Import list of contacts: First, we imported a target list of contacts from our CRM into Clay (if you don’t already have one, Clay can create a list for you).
   
2. Enrich with personalized details: Clay then enriched contacts with contact information, LinkedIn profile URLs, latest posts, recent job changes or promotions, and other relevant signals. 
3. Design personalized direct mail campaigns: Using the enriched data and ChatGPT integration, Clay crafted personalized direct mail campaigns with tailored messaging. For example, offering gifts based on specific milestones or job changes.‍
4. Send direct mail through Sendoso integration: Using the direct integration with Clay, we automatically triggered personalized gift sends, such as local restaurant or spa gift cards, directly to the prospects’ addresses.
   

Outreach Play with UserGems

There are an overwhelming number of tools you can use to capture buying signals, and they all have their advantages. The signals we’ve found that actually convert fall into these categories:

• Buyer job changes
• Buying group changes
• New hires and promotions
• Past champions

This part didn't get updated to reflect my draft: We began using UserGems to help us track and act on these signals. Our outbound success depends on trusting our data and having a steady stream of high-intent ICP contacts to engage at just the right time. These signals consistently result in our highest converting outreach.

Along with UserGems, we use several other tools, all connected to Clay, to run this play efficiently:

• Sendoso for gifting
• ChatGPT/OpenAI to craft email copy
• Orum for parallel dialing prospects that either didn’t respond to the email or opened it at least twice without a response, pulling the phone numbers from Apollo
• Outreach for building the sequences
• Instantly/Smartlead for sending sequences
• SFDC to track meetings booked, no-shows, and closed/won deals

Playbook: New Job Congratulations

Using a tool like UserGems, we can monitor our users, customers, active contract signers, and other audiences to see when they move or change jobs. The play kicks off when we get alerted to one of those signals. Here’s the step-by-step process of how we use our Clay table to run it:

1. Qualify their new company and role as within our ICP
   
2. Develop personalization with enriched historical data, such as the name of their CSM from their former company
   
3. Using this information, ChatGPT writes a human-sounding personalized message
   
4. Our SmartSend API finds a gift that matches their interests and inserts a gift link into their email
   
5. The sequence is pushed to Smartlead to send out emails with personalization variables, as well as trigger calls and LinkedIn touches as necessary

If you need help building your table, Clay has plenty of templates you can use for building blocks.

Recipient Feedback

Not only has this play been a highly successful pipeline generator, but we’re actually getting love notes back from people receiving the messages — which, as a marketer, was unexpected but much appreciated. The responses from other marketers have ranged from amazement to delight, with many asking how we’re doing this, which is pretty cool.

"Would love to learn more about your tactics here. This is type of stuff I need to get my team doing."

"Thanks, I love this. Let me set up a call with my biz dev team and sales leadership, I think they'd all love this."

"Nicely done. I'm blown away by the technique you used! WOW!"

"Meeting booked! I look forward to speaking with you!"

Metrics matter, but getting feedback like this reinforces that we’re doing this in the right way and it’s resonating with our audience.

Using AI Responsibly

For every successful use of AI in marketing, there are at least ten examples of failure. Think of AI as an intern — it’s eager to learn but relies on you to give it the information it needs to succeed. You need to provide details, write samples, and think through all the things that could go wrong in order to create strong prompts and get the right output.

Through a lot of trial and error, we’ve figured out how to build prompts that work best for us. Yours may differ in the information you’re providing or expect to get back, but the foundational principles remain the same.

1. Create custom, detailed contact tables

Clay tables allow us to piece together information from lots of different sources to produce the result we’re looking for. Making these contact tables as detailed as possible will allow you to have enough information to pull from so your outreach doesn’t sound generic, robotic, or forced.

Here are some examples of the detail fields that you may not think of:

• Is their company B2B or B2C
• Normalize their job title, ex: “a growth marketing director” instead of “Director, Growth Marketing”
• Determine their specific department, ex: “Marketing - Demand Gen”, instead of just “Marketing”
• Years and months of tenure
• Department initiatives
• Interests as determined by SmartSend

This isn’t an exhaustive list, but it illustrates just how much information you can - and should - be giving your AI tools. Anything that provides context, normalizes language, or acts as a data point is fair game to be included.

2. Tell AI what NOT to do

People take a massive piece of the process for granted: direction on what to avoid is the most important part of your prompts. In most AI fails, the prompts weren’t thought through with a “What could go wrong here?” approach. I would love to say we haven’t made mistakes, but those mistakes have helped us learn a lot and refine our prompts to make them more successful and get more responses.

These are some of the things we come across that can derail your outreach and potentially damage your reputation:

• Getting too personal. Interests inside and outside of work, pets, local (to them) attractions, and experiences - these are all fair game. You want to avoid talking about subjects like people’s kids, family members, or life events perceived as negative or sad. These can come off as intrusive and inappropriate.
  
  
• Using specific information incorrectly. People often mention languages they speak in their profiles. While it’s a fun fact, it’s not their entire personality, and it’s likely irrelevant — but AI-written outreach can fixate on specific details like this, and an outbound email that references everything about being French because someone has a bullet point about speaking the language is a huge miss. Direct your AI to avoid using specific, small details to influence the entire message.
  ‍‍
• Scraping social posts for “relevant” signals. I saw this play out in real time recently. A member of my team lost her mother over the summer and wrote a thoughtful post aimed at helping other people in the same situation. Shortly after, she received an email with the subject line “Sendoso’s content strategy while you grieve” and a line crowbarred in the body referencing her post. The email's sender admitted she was using an automated sequence that she wasn’t checking regularly, which turned out to be a huge mistake.

3. Train your model to sound like a human team member

This outreach is being sent on behalf of a real person on your team, so the copy should match their voice and writing style. You can achieve this by training your model on writing samples both from your brand and the person sending the email. These can include:

• Emails from team members
• LinkedIn posts
• Brand guidelines

In addition to writing samples, you should have explicit instruction on the output you want from your model and the persona it’s writing to. For example, phrases like “You’re emailing a friend” and “Formulate a super casual, human-sounding email opening” are hugely effective in nailing automated personalization.

4. Continuously test your prompts and pay attention to what’s being sent

This bears repeating: Automation is not “set it and forget it” from the start. AI learns and improves over time, but only with feedback.

When we first started this campaign, we kept a close watch on the emails being sent to ensure nothing inappropriate or wonky was happening. This also allowed us to quickly make changes on prompts that weren’t working and test different types of variables to see how the output was affected. 

Now, we’re confident enough to allow the program to run, and we only need to spot-check emails regularly and keep an eye out for any responses that indicate there’s a problem.

The Results

When we started this program, we hoped to see incremental improvements, not necessarily runaway success.

Impressively, we’ve achieved the latter.

Here’s a breakdown of the results we’ve seen:

• 11-18% response rate — a 20% increase by leveraging enriched data and personalized direct mail campaigns.
• Over $1M in pipeline generated a quarter through a combination of personalized outreach and strategic gifting.
• Automated workflows allowed the team to handle 3x the number of prospects without increasing headcount.

These numbers are something to be proud of on their own, but when you combine that with how much we’ve spent, it looks even better:

• Clay subscription + pay-as-you-go credits = around $1k/month
• Email tool (Instantly/Smartlead) = $100/month
• Domains = $7-$14/month per domain

After seeing the results from such a low investment, we're doubling down and increasing our budget - while still saving hours of work, exceeding our pipeline goals, and continuing to prove that you can automate and scale personalized outreach.

If you're looking for an even deeper dive into our strategy, join me and Bruno Estrella, the Head of Growth at Clay, on December 4th (10am PST/1pm EST) for a live webinar where I'll talk through these playbooks from start to finish. Register now to save your spot!