SDRs on Autopilot: How to run programmatic outbound campaigns (for next to nothing)

There is a huge amount of talk and hype about the future of outbound and the impact of AI on our outreach. As someone operating in the space, it certainly feels like there are some fundamental shifts in the way we’re going to GTM more broadly. Some are even going as far to question the different roles humans vs AI are going to play in go-to-market motions moving forward.

Over the past month working at the second company I co-founded, GoodFit, we began outbound outreach campaigns for the first time and got to tackle this new reality and its consequences head-on. 

Most interestingly, I got the chance to compare the experience, setup, infrastructure, and results to when I successfully ran and scaled these programs at Paddle ten years ago. 

Spoiler: We’re achieving some great results (quickly!) at GoodFit, however I was left feeling a little underwhelmed. Largely because in doing so it didn’t feel like we’re doing anything particularly fancy or groundbreaking. No generative AI was used, we made some minor use of GPTs and LLMs to speed-up some tasks but they were by no means required. 

The good news is - replicating our setup and results feels something easily replicable by many others rolling out these programs themselves. Whilst this world is changing fast, there are a clear set of steps to follow to drive highly repeatable results. 

The year is 2017 

‍

Some context on the world in which these programs were rolled out previously will be helpful to contrast what’s required in executing today. 

To do so we need to step back in time to the year 2017. Everything is good. Harry Styles releases his first solo album. Many of us didn’t know what a black swan event was…

Importantly I was running revenue at Paddle (with a lot less grey hair!). Although, nobody knew who or what Paddle was. Nobody knew what we did, why we did it, or how it differed to the competition. A set of circumstances not uncommon to other early stage businesses out there. 

As a result of our lack of brand or market awareness, we made the brave decision to go out ourselves and find customers that we felt would be a match for our product offering. We did so via cold outbound email.

We ran all of this outbound email manually through our sales CRM at the time, Close.io. Long before the lofty rise of SalesLoft & Outreach in creating and owning the sales engagement category. 

Candidly, things were going well. Paddle was named fastest growing software company in the UK, and named the 6th fastest growing company overall in the Deloitte Fast50 two years in a row. 

Behind our 3000% growth over a three year period was us running outbound email. Outbound, and email specifically, was our exclusive channel for the first 5 years of the business. 

Outbound email was embedded deeply into the culture of the organisation, there were a set of firmly held principles around how it should be run, and our growing sales organisation would be indoctrinated and trained on said principles as soon as they walked through the door. Many of these we’ve covered in previous webinars with Notion. 

Fast forward to 2024

If we fast forward to the state of outbound in 2024, the picture is wildly different. The good times in tech have ground to a halt, budgets have been cut, the era of growth at all costs is over, valuations have softened. 

During this time two perspective have come to the fore: 

‍

1. Outbound is dead - RIP 

How many LinkedIn articles have people read telling you outbound is dead in the past year?

In reality, there are more companies & competitors sending more bad email than ever before. From my perspective, outbound was always hard, but you can no longer make up for being bad with high volumes. Fortunately, outbound is very much alive if done well.

2. AI will change outbound forever 

‍

This perspective is a little trickier. AI is inevitably going to change the way we GTM. However, a more accurate statement is that how we go-to-market is evolving, has always evolved, and if you don’t also evolve you’re gonna get left behind.

‍

“How we go-to-market is evolving, has always evolved, and if you don’t also evolve, you’re gonna get left behind.”

I’d encourage folks not to get hooked-in or intimidated by the more extreme of the statements. I certainly did and was stunned into a stasis for a short while.

‍

Fortunately, the steps below will help you begin executing on low cost, effective outbound programs if rolling them out for the first time, or if optimising an existing motion. 

Why isn’t outbound working for folks?

Before we jump into what to do, let’s spend a little more time on why things aren’t working for many. 

Is it that outbound is dead?

Is it that AI has ruined it for everyone?

In reality, I think the writing has been on the wall for some time. There are a lot of tech businesses out there today - and the barrier to creating them is lower than ever. 

Meanwhile, a category of tools has emerged to make sending email to a lot of people really really easy. As such…. we’re receiving a lot of email (and this extends to other channels too…)

One of the things people tried to do to stand out is deliver highly custom, highly personalised email. Now, the introduction of AI has only increased our capacity to create “personalised” email and amp up the numbers of emails being sent & received. 

As mentioned, this is true for email but other channels too. With tools out there for AI driven ads,  automated LI touch-points, and more. 

‍

However, to say AI has us all doomed still is a little extreme for my tastes. 

I can’t tell the future….I don’t know what is going to be automated vs not in the future. But having tried and succeeded to set up zero-human touch outbound campaigns in the last month. I can share my perspective on what most folks can use AI for today and what they can’t for yet. Something that I’m sure will continue to evolve and surprise us. 

As of now, AI isn’t going to work out a repeatable / predictable / successful way to run outbound for you. When discussing launching outbound during our leadership meeting at GoodFit, bearing in mind we don’t have any SDRs/BDRs, a smart member of the team suggested, “Let’s go and use an agency, or AiDR tool and let them run outbound for us”.

This isn’t going to work. 

AI isn’t going to work out product-market fit for you…

• You need to understand the audience of customers who have a common pain.
• You need to have built a solution for them.
• You need to work out how to communicate the benefits of your solutions to those people.

Here AI can help. Drawing insight from customer conversations, but you still need to go and have those conversations. AI can help interpret the unstructured data - but still isn’t doing a better job  than the founder’s intuition having spoken to 25 companies just yet imo. 

‍

What is it going to do? 

AI is there to help scale something that is already working. The ability to automate well defined tasks / decisions / activities humans are able to do is certainly possible and something to take advantage of.

It’s this which allows us to run much more of our outbound or top-of-funnel (TOFU) activities, at higher volumes, lower costs, and with less human involvement than ever before.

So let’s explain how we did it. 

Case Study: GoodFit 

GoodFit provides data to sales & marketing teams. We’re in a highly competitive space, with both 100s of new entrants emerging each month, as well as large incumbents like ZoomInfo present.

Moreover, we’re in an industry where people have been oversold many times over. On the whole - folks aren’t super happy with the data they’re using to GTM. 

Everybody likes a challenge, right?

Despite those challenges, we’re growing very fast, but have not yet spent a $ on sales / marketing - aside from attending / hosting a handful of events. The sales team is myself and my VP of Revenue. 

For GoodFit, historically organic inbound has been the driver of demand. Think, referrals, previous buyers moving to new orgs, and intros from our network. 

This inbound deal flow has taken us well beyond $1m ARR —however we approach summertime 2024, we were ready to accelerate and made some big breakthroughs: specifically around outbound campaigns. 

Limitations 

In looking to set this up in 2024 vs 2017, I recognised I was working under some quite different limitations or constraints this time around. Specifically: 

‍

1. We had zero folks in sales / no SDRs.
2. We had zero budget.

‍

Some pretty extreme limitations, but they certainly sharpened our minds. For others, I think they’re helpful as they’re the same constraints all GTM leaders are facing to some degree. 

Nonetheless, making outbound work was a challenge I was up for.

Checklist: Running programmatic outbound 

In what felt pretty anti-climatic, we did everything that also made sense 8-10 years ago. The core concepts for outbound remain the same: 

• We mapped our market of companies & contacts we could sell to.
• Built the audiences of companies & contacts we wanted to reach.
• Configured our sequences / campaigns.
• Put the tooling to work (rather than our SDRs this time!). 
• Then iterated and scaled from there.

Everything has changed… but absolutely nothing has changed in that 8-10 year time horizon. 

‍

Market Mapping

The first step in the process is mapping your market of companies

One of your constraints to outbound (and revenue overall) is the size of your market, so you want to understand the size of your market early in your startup journey. 

In the case of running outbound, you need the market mapped to build the list of accounts you’re going to sell to. Having that list of high qualified accounts is critical - as your outreach is only as good as the people it gets sent it to. Something many have forgotten.

“Your outreach is only as good as the people it gets sent to. Something many have forgotten.”

For the purposes of outbound campaigns requiring no human intervention, we want to map the companies you want to sell to, as well as the contacts you want to reach in those businesses.

In the case of programmatic outreach, you can’t have a rep browsing LinkedIn Sales Navigator determining who they want to email - since we don’t want humans involved in this part of the process at all.

TL:DR - map your market of accounts & relevant contacts.

At GoodFit we used our own tools for this, however there are many others out there that folks will be familiar with. 

GoodFit Mapping

At Notion we’ve previously written about mapping your market here - specifically highlighting the importance of building insight into the set of companies you’re able to work, support and sell to today. 

These companies are your “serviceable” companies, and we want to map your market of serviceable companies, your “SAM”. 

This market mapping exercise will look different for every company given the unique profile of customer each busines is looking to serve. Some examples: 

… If you’re an expense management business: every company with 500+ employees, 3 in finance. 

….If you’re an employer of record: every company hiring for remote staff / outside office locations. 

It’s critical you get as tight as you possibly can here, to ensure you’re building a list or mapping the market of truly qualified companies. 

At GoodFit we had a deep understanding of our serviceable customer.

• They need to have an established sames team: +5 sales team members 
• We don’t want companies selling into niche audiences (i.e their own SAM must be 1k+ accounts). This excludes those selling heavy enterprise, only into local markets, etc. 

Rather than establishing a set of filters for this (as we would suggest for most people), we’ve trained a text-based ML model to qualify out companies selling into niches.

TL:DR - We model looks at the way a company describes itself (we scrape for the words on their website). We trained the model on 100s of qualified companies, and 100s of bad fit companies. The model identifies which companies look (or describe themselves) similarly to those who are qualified vs disqualified. The model works to 90% accuracy - helping us exclude the companies selling into niches which are a bad fit for us. 

Don’t let this confuse your own market mapping exercise - most companies can rely on pre-existing data and filters vs models. 

Most importantly for the purposes of programmatic outreach, our SAM (i.e +5 sales in sales not selling into niches) delivered me 3639 companies. Something I can flex up-down based on that minimum sales headcount. 

‍

Recycling & Funnel Assumptions 

Having identified the number of serviceable companies out there, you need to work through an important set of questions:

‍

How often per-year am I willing to recycle a company? 

‍

In this case I was happy with twice per-year.

‍

This means I can reach out to the total number of companies in my market multiplied by 2.

‍

3639 companies every 6 months, is 7278 per year.

‍

What is my disqualification rate on the mapped market? 

‍

You will always disqualify a % of companies mapped - this exercise is never perfect. In our case we’re disqualifying 10% of the mapped market (due to our model performance). This means we have 6550 accounts available to be worked per-year. 

‍

TL:DR: I can outreach to ~550 accounts p/m based on my market size & recycling cadence.

‍

If we apply some assumptions to the funnel using input figure, we can also predict how many replies / meetings we should expect off this volume of outreach. Also sense checking whether the time/investment is even worth it. 

With programmatic outbound, the number of accounts you can physically contact per-day / per-quarter is no longer the limitation. 

In 2017 when running this at Paddle, we’d constantly assess SDR/BDR activity levels & productivity. Asking ourselves, how many folks could they reach out to in any given quarter? 

In 2024 doing this without BDRs, your constraint is largely the size of your market. 

In order to get more juice out of outbound, you either need to improve funnel performance, or expand the size of the market you’re able to effectively service/target. 

Audience Building 

Having mapped your market of serviceable companies, we next need to determine the audience(s) we’ll be running our outbound campaigns at. 

There are lot of different types of campaigns you can run:

• Many folks start with competitive displacement campaignssome text
  • “You’re using competitor X - here’s where they’re weak - here’s what we do better…”
• Some people run lookalike campaigns..some text
  • “We have customer X in your industry, here’s what challenges they had and how we helped…”
• Some get as far as value-based campaignssome text
  • Targeting companies who look like they have a pain that you solve for. With messaging that suits.

To build out an example of a value-based campaign: Paddle offers payments infrastructure for software companies. Paddle’s strongest value proposition is helping clients sell their software globally. A value based-campaign / audience for these folks would be companies with global traffic, only supporting a single currency/payment method. 

Lookalike Audience Example: Cyber Security 

‍

At GoodFit we started with lookalike campaigns. One we’ve worked on is companies in Cyber Security. With some strong reference customers in the space, it seemed a sensible place to start. 

To identify the Cyber Security audience (a subset of your SAM) we used GoodFit itself, however you could use your CRM or whether your account data lives and is most accurate. 

For us, we maintained the filters we need to ensure a company is serviceable, and then used a longlist of cyber security related sub categories to capture the companies we wanted in our audience. 

This left me with 114 companies for our CyberSec lookalike audience. 

‍

‍

To expand on advice in building out these audiences:

‍

• Build the audience wherever it makes sense for you.

• You want to be able to tag the companies in your audience or put them in a labelled list - different tools do this in different ways albeit usually it's very simple.
• At GoodFit we map the market of every serviceable account in your market, as well as all the contacts. Many other tools don’t do this, and/or you’ll be charged per-contact enriched. 

• I’d suggest you only bulk enrich contacts against the accounts you want to run campaigns at (as opposed to your whole mapped market) in order to keep costs lower.

Audience Building: Key Learnings

I’d encourage you to experiment with a few different types of campaigns. I have a personal preference for value-based campaigns given they’re typically differentiated to competition and specific to what your product is uniquely positioned to do. 

Why did we try lookalikes first at GoodFit? 

Folks have been sold (via outbound email) data, lists of prospects and more a lot. However most folks are selling the same poor quality data. Whereas, GoodFit is immediately and visibly different to status quo based on data we supply. If I know what industry you’re selling to, I can give 3 example data points which are super useful, you’ve never had access to before. I can stick these three example data-points in every touchpoint I deliver and get some great results. 

“Your audience should be so tight, so specific… that your messaging writes itself”. 

What is absolutely critical is the tightness of your audience. Your audience should be so tight that the message writes itself.

 Eg. “You are in X industry, so we could offer Y data points that you don't have”, “You have international traffic, but aren’t localised - can I show you how localised checkout would look for X”, “You have a growing AR team - here’s how we help them / stop that team growing linearly”. 

You should know what to say to that potential customer just because they're in the audience. The other benefit is that it will mean later down the line we have less dynamic content in our sequence, makes things less risky.

As we highlighted earlier - folks got away with poorly defined, poorly targeted audiences historically and “solved” for this through mass volumes. This will no longer work. 

Sequences, Cadences & Messaging

As a recap, we’ve mapped our market: e ~3600 companies we can ping twice p/a. We’ve built tight lookalike audiences: those in ATS, CyberSecurity, DevTools, SalesTech…

In approaching the need to run outbound with no budget, no SDRs, I realised I needed to approach building the messaging / sequence in a very different way. 

Typically when trying to build our sequence or writing our messaging. We think about trying to write the very best sequence/campaign that we can - measured on the response rate / meeting booked / $ value of opps created. 

I realised quickly that I didn’t have this luxury. 

For example, something performing quite well for others as a part of their sequence is cold calling. However, I didn’t have any SDRs to do the dialling. I would love to add in a customised Loom video on data we could supply a prospect -  however I didn’t have anyone to create the content. 

I quickly recognised the cheapest, most scalable way I could run outbound was exclusively via email & LinkedIn touchpoints.

These are channels I can utilise, without needing humans to drive the activities. They’re cheap, and I can scale them without humans. Largely because there’s tooling out there to automate messaging on these channels (with some constraints).

Continuing to pull on the thread of performance vs scalability. 

Historically BDRs/BDR managers have been writing these sequences. We target them on ops created. As a result, it’s in their interest to have highest conversion from contacted > opp created

Rarely do they think about the cost of any activity they’re doing…they just want as many as possible to convert. Failing to recognise  every dollar spent creating a Loom is a dollar spent, and that dollar could be utilised enrolling more folks … or doing other things. Every phone call takes time, this costs money, or has opportunity cost.

As a CEO I don't want 10% more opps / meetings booked from the audience (as result of these phone calls / loom videos) for 50% higher cost, or 50% lower throughput. 

My suggestion for anyone designing outbound moving forward is to start with the cheapest most scalable touch points (for us this is LI/Email), and see what baseline performance they give you. 

“Start with the cheapest, most scalable touchpoints you can in order to establish your baseline performance. Layer-in more expensive & time consuming touchpoints from there and determine ROI”. 

From there you can add in more expensive touchpoints having established your baseline. When I say “expensive” I mean that to cover the actual cost of the touchpoint and time taken to execute. 

Having seen the lift the extra costly touch gives you, you can accurately determine whether that extra touchpoint is worth it.a

Lookalike: Sequence Example

Moving out of the theory and to the practical. We wrote a lookalike sequence made up of 6 touches over 11 days, made up of 4 emails and 2 LinkedIn messages.

Because our audiences were so tight, the messaging was easy. 

• You are in X industry
• You probably need Y data points
• Can I enrich an account with them ?

We ran the same structure for every campaign albeit with different data points.

In this case because the audience was so tight, I could deliver the same message to everyone. 

There were no dynamic snippets, or dynamic examples in there. You could try and automatically compute every company’s vertical, and pre-fill with dynamic examples based on the computed vertical – but that level of automation can break. Making your brand look silly in the process! 

The principles by which we wrote the message were covered in a previous Notion webinar, called “hyper-relevant” email. I’d recommend this for anyone building these out for the first time. 

‍

‍

Tooling & CRM Sync 

Having established your audiences & messaging, we need to ensure these accounts are synced to the CRM if they aren’t already in there. 

• Ensure any account you’re working is in your CRM
• Get each company tagged with the campaign that you want to run or include them in.

• At GoodFit we use Hubspot
• All 3600 accounts live in HubSpot - our entire serviceable market.

• We use HubSpot’s list builder to build a list of the folks I want in a campaign

In your CRM you want to build your list at a contact level, i.e we want a list of the contacts working at the companies we have tagged in a specific audience/campaign. 

When building your list filter by the tagged companies in your audience. Here you can also filter out folks who are already being worked, have an open opportunity, or those who have already been contacted recently. You can also filter by contact if you’ve got different sequences targeted at different buyer personas.

You then want to export this list of contacts to wherever you’re running your sequences from.

At GoodFit we upload our lists of contacts to Lemlist, a tool we use to automate both our LinkedIn and email touchpoints. Our sequences are configured in their platform with accompanying logic. 

TL:DR - Build list of tagged companies & contacts, and upload them to the tool you run automation from. 

Constraints & Risk 

As much as we’re aiming for infinitely scalable outbound with as little human interaction as possible - there are some constraints. 

Running outbound programmatically has risks, and you should scale it based on your risk appetite.  

In terms of constraints, LinkedIn only allows a single person to connect with 100 net new contacts per week. If you abuse LinkedIn’s connections and messaging capability you’re at risk of having your posts shadow-banned, or worse your profile deactivated. 

Email is more scalable than your LinkedIn touchpoints, although when sending significant email volume you should definitely seek advice. 

I’m no expert, I’d encourage you to reach out to folks like Allegrow, your automation tools themselves and more for advice and to establish your risk appetite. 

I would also suggest running your outbound email from a separate warmed-up domain and workspace than your primary. 

Results & Performance 

We’ve mapped our market, built our audiences, written sequences for each and put the tooling to work on outbound activities. 

Now we need to measure our results and iterate. 

‍

• In month #1 we enrolled a cohort of 250 companies, ~480 contacts.
• Our LinkedIn connection rate across members of the team was consistent. 
• We saw a 6% actionable response rate ( i.e all responses we could do something with)

My assumption going into month #1 was we’d get up to 3% response rate and we’d need to layer in less scalable, human touchpoints to reach +5% response rate. So we were pleasantly surprised! 

For context back in 2017, running completely manual outreach with templated sequences we were seeing highs of 20-30% response rates. Those days are gone, and the process was significantly more expensive and less scalable than the above. 

Across most of the businesses I advise (after some coaching) I typically see a 40% meeting booked rate from an actionable response. Performance here was inline with expectations. 

When setting out assumptions for our meeting > opp creation rate, we had less data to go on. Having analysed the results we’re actually outperforming conversion from our inbound funnel - currently at 80% opp creation rates from outbound calls booked. 

Given we qualified the accounts when mapping our market, zero’d in on lookalikes that we had reference customers and great data for, and having highlighted this data in our outreach. We recognised by the time these accounts were on a call with our rep they were super qualified, de-risked in terms of needs, and easy to convert to opps. 

The above is the benefit of an outbound approach vs inbound where you have less control over those who submit the demo requests., 

The above was our first attempt at programmatic outreach - the numbers were small but funnel performance very strong. We’ve since continued enrolling folks and are seeing consistent results while experimenting with new campaign types. 

In terms of cost, we didn’t need to pay for the data (as we were using our own GoodFit product), and thus only paid for a number of seats with Lemlist costing in the low thousands p/a. 

If paying for a premium data-provider, I would still expect for the typical cost of an SDR you can run the above stack without issue. In most cases matching response rates, increasing productivity / throughput, and significantly reducing overhead. 

From your SDR/BDRs perspective, they then can focus their attention on higher value activities like responding to interested, qualified accounts and/or chatting with prospects over video. 

The above will be a critical part of our ongoing TOFU efforts at GoodFit moving forward - super excited to hear if you’re able to replicate this in your own business context.