Steeve Vakeeswaran is the Head of Sales & Expansion at Zapier, the platform trusted by Spotify, Meta, Shopify, Dropbox, and more to automate workflows across 5,000+ app integrations. 

He recently sat down with us to jam on all things automation, covering topics like: 

1. How early-stage teams can start priming for automation and scale
2. KPIs for automated customer flows throughout your funnel
3. Windsor’s unique ability to automate with a human touch

“What makes Windsor magical is the ability to scale that personalized, human touch. It’s a naturally valuable product that’ll help you win more business.” 

‍

Why personalization & automation are more compatible than you think

When automation works well, it feels, as Steeve puts it, seamless and almost magical. 

Yet, we haven’t been able to scale automated systems while maintaining a human touchpoint or sensibility that feels natural. 

That is — until Steeve discovered Windsor, which emulated both sides of the coin: 

• The tech and strategy are grounded in foundations of systems thinking and automation
• At the same time, the product scales a very human element: personalized conversation

Perhaps what struck him most was how the end product of Windsor — a deepfaked video message — still felt organic in a way the industry hadn’t quite seen before. 

In his words: “Windsor was a no-brainer for me when I saw the power they have to replicate those one-on-one moments across countless unique users.” 

"We associate automation with robots and trained code. But Windsor nails the really tough task of automating while organically scaling a human aspect." 

‍

Early-stage teams should consider automation — sooner rather than later

For early-stage companies, especially at the pre-seed and seed stages, there’s usually an emphasis on the scrappy, DIY approach to piecing together your first product. 

As a founder with a small team and a minimal pool of users, you can get away with things that don't scale, like handwriting thank-you notes or handling every support ticket. 

In Steeve’s experience, as a two-time co-founder himself, the caveat is you’ll eventually have to ask, “If this effort works and drives returns, how will I get it to scale?” 

Not considering scale = stunting potential growth

He uses Windsor as an example. You don’t need deepfake videos at scale from day one. 

After all, as a young company, you could sit in your bedroom, record dozens of video messages in unique styles, and send them off to your small customer base. 

However, what happens if you find out one of those video styles performed notably well with users, and you’d like to keep those results going at a far greater scale? 

The answer isn’t to record more and more videos until you can’t bear it anymore. At that point, systems thinking and the option of automation become appealing. 

If you’ve been considering them from the get-go, that’s even more powerful. As such, Steeve advises early-stage teams to balance both: 

1. Be scrappy & experimental — Realistically, most early ideas or practices won’t pan out. So, feel free to constantly iterate and not get attached to things. 
2. Keep growth in mind — Simultaneously, you won't be early-stage forever. One of those ideas is likely to gain traction. When it does, you should be ready to grow it. 

Your users could love your service or solution. But, if you can’t scale it, there’ll quickly be a finite number of individuals who can enjoy your offerings. 

If you plan to scale today, you won’t accidentally stunt your potential growth tomorrow. 

“I’m a big believer in the scrappy, early stages of doing stuff that won’t scale. You have to balance that with: ‘If this works and gains traction, how would I grow it?’”

The GTM playbook: KPIs for gauging successful automation

When tracking the success of an automation tool, many teams tend to prioritize performance measures like, “Are my Windsor videos driving incremental sales or not?” 

In reality, says Steeve, that’s likely the last question you’d be able to answer when testing any new piece of software, new channel, etc. 

Instead, he recommends considering metrics through the lens of the user journey or funnel. 

Start by defining your current baseline plus the function or campaign you're measuring (i.e., testing for win-back email success). Then, break your funnel approach into: 

1. Opens — If you’re using Windsor, how many more email opens have you received since implementing video campaigns? 
2. Lower funnel — Are you seeing greater success with CTAs due to Windsor videos, whether that’s signups, clicks to your site, feedback submissions, etc.? 
3. Closing — At this point, you can finally gauge big-picture success with conversions and, as mentioned, incremental sales. 

“I find it valuable to approach metrics from a funnel or user journey standpoint. A lot of folks will skip key steps and jump to the very last question.” 

‍

Looking forward: Automation is soon to be ubiquitous

Below, Steeve leaves us with two of his predictions for the future of automation in eCommerce. 

1. Automated personalization won’t always be novel — and that’s okay

Windsor users today are some of the earliest adopters of AI-driven, personalized video marketing at scale. They won’t be the last. 

In the coming years, Steeve anticipates tools like Windsor following a similar trajectory to personalized email marketing over the last decade or so. 

Companies and consumers alike were excited by it, more and more brands caught on, and now, in 2022, it’s become table stakes — a given in every eCom team’s marketing playbook. 

The novelty value of personalized emails faded, and every modern marketer struggles to cut through the inbox noise of daily newsletters, dozens of brand sales, etc. 

But we all still use it, because the tech being table stakes doesn’t matter when: 

1. It still drives value, especially if you leverage it innovatively
2. No other channel has been able to totally replace email

Overall, industry standards exist for a reason: They just work, and there’s no need to mess with that. The average U.S. consumer will likely never turn off email, so neither will brands. 

2. Deepfake tech won’t just repurpose content — it’ll be generative

As for the tech itself, the main functionality currently offered by tools like Windsor is repurposing existing video by inserting changes like customers' names. 

Deepfakes in 2022 are impressive, Steeve explains, but still dependent on taking time to record and upload source footage. You can’t generate the content on demand. 

As these software capabilities inevitably expand over the years, he highlights the potential of creating Windsor videos (almost) entirely from scratch. 

Imagine selecting from a database of actor profiles and pre-written scripts to generate endless permutations of video campaigns. 

Or, beyond everyday stock actors and models, he floats the possibility of a Cameo marketplace model, where celebrities sign off on their deepfakes delivering scripted marketing messages. 

Ultimately, Steeve drives home that the many applications for Windsor don’t just signal growth for the personalized marketing space, but automation as a whole technology. 

“Down the line, I imagine every user could receive a hyper-personalized experience. That’s where brands will really win by automating at scale.”