How to Scale #Personalization, with Guillaume Cabane (Drift)

This article includes key takeaways from Guillaume Cabane’s talk on how to scale personalization. Check out the full video on Youtube or get the latest from our Scale series.

4 key mindsets

Mindset #1: People don’t like buying from you

In 2015, 53% of B2B buyers stated that they preferred researching solutions on their own. In 2017, that figure rose to 68%. The dislike and mistrust buyers feel towards vendors is become more and more widespread. And this is due in no small part to marketing’s tendency to spam as well as to overpromise and underdeliver. What that means for marketers is that your website and your marketing are becoming the last place people turn to in their buying process.

Mindset #2: Stop copying

When it comes to product, everyone knows that copying a competitor doesn’t work. Instead, investing in your differentiation and focusing on the angle of innovation you bring to the market are the surest ways to build a moat. The same holds true for marketing. The lack of diversity in the channels used by marketers translated in a 70% rise in average CAC in recent years, a timely reminder of Andrew Chen’s “law of shitty clickthroughs”:

Over time, all marketing strategies result in shitty clickthrough rates.

For Guillaume, this as well as increased skepticism from B2B buyers (Mindset #1) means that marketers need to innovate. Simply adopting existing tactics will only return a fraction of the impact that the tactic initially yielded. And therefore lose marketers’ competitive edge.

Mindset #3: How growth teams fail

Most growth teams fail because they aren’t able to rely on good KPIs, they aren’t able to present a prioritized roadmap, and they can’t offer a reliable forecast of their impact on revenue. Point by point, here’s why growth teams are coming short.

First, KPIs: most of the time growth focuses on traffic or leads, which is too loosely correlated to revenue and uncorrelated to the quantity of work an the degree of innovation supplied by the team. Metrics further down the funnel, such as qualified leads and demos, are more telling metrics but are more impractical since there is a lagging effect (especially in SaaS).

Second, prioritization: many growth teams rely on arbitrary frameworks (like an ICE score) to determine their roadmap. These tools don’t allow for a clear picture of why the team is committing its efforts to certain ideas rather than others.

Third, revenue forecasting: with a lack of adequate KPIs directly correlated to revenue and no sure way to link roadmap priorities to revenue, growth teams are failing at tracking their efforts, focusing their efforts, and showing their results in actual dollars.

Mindset #4: How growth teams should fail

Whereas product teams rely on data and customer feedback to prioritize their roadmap, growth teams are inherently experimental. If you know an idea is going to work then you aren’t going to test it, right? Guillaume recommends that growth teams adopt a mindset of expecting to fail and define success instead on the number of experiments performed.

The average success rate of growth teams in the Bay area is 20% at the experiment stage.

The way Guillaume recommends going about prioritizing work is to focus efforts on experiments that hold the potential for the highest lift in revenue. To determine this, Guillaume’s team considers each idea and lists: hypothesis, days of work, metric impacted by test, dollar value per metric, odds of success, and weighted value (i.e: dollar value*odds/days to build). By doing this, he can create an accurate picture of total pipeline currently in the growth team’s backlog and prioritize tests that yield the most revenue per day of work.

Note: This pipeline is incredibly valuable for growth teams as it means you can commit to clear revenue targets, prioritize your backlog accordingly, and hire against that pipeline by measuring how much extra pipeline you can handle with an additional resource on your team.

4 key best practices

Best practice #1: G’s Reveal Loop (and Private Reveal Loop)

Guillaume reminded us of his setup to target anonymous website visits with outbound emails, and added a new sequence which is derived from the first. The ‘classic’ Reveal Loop used by Guillaume looks something like this:

Get IP address of website visits → Ping Clearbit for domain → Score lead with MadKudu → Prospect with outbound email campaign

The reworked version requires a bit more effort. For starters, since B2B products have a finite TAM (total addressable market), you can create a database of potential customers, lookup their HQ address and collect their GPS coordinates. Then use it in G’s ‘private’ Reveal Loop:

Get IP address of website visits → Ping DB IP for geolocalization → Search your database for a match → Check if distance is small enough to be considered a match → Prospect with outbound email campaign

Note: to determine a fuzzy match with an appropriate degree of certainty, G recommends checking your potential customer database for the minimum distance between 2 coordinates and using that value to confirm a match.

Best practice #2: Out-bidding the competition for high-value leads

The problem a lot of growth teams encounter is that the cost of selling to certain leads is higher than the value of those leads. The solution consists in being able to distinguish between high and low value leads, and splitting the funnel early. Once you split your funnel into buckets each with a personalized experience derived from the value of the lead, you can effectively out-bid your competitors on high-value leads. And under-bid on leads that have a low value. Doing so forces your competitors to either a) overspend or b) retreat from the market.

Best practice #3: Look outside the funnel

The funnel as it is mapped tends to ignore a large pool of potential leads. The reason? The funnel starts BEFORE a website visit. On average, a B2B buyer does 12 searches before engaging with a brand. The challenge for Guillaume has been to try to identify and reach out to leads before the competition despite having no data to work on from those leads. Here are some of the tools he’s put in place to get around that roadblock:

• Bombora: identify leads based on current research topics
• SEMrush: identify leads based on competitor’s campaigns
• PredictLeads: identify leads based on who they are hiring
• SifData: identify leads based on champion moving companies

Best practice #4: Getting hyper-personal

One of the goals Guillaume sets for his team is to create memorable experiences. To do so, G uses several levers. One is to create a wow effect: once a visitor has been identified as a high-value lead, a bot asks which we we likes his coffee. An order is then created and sent to the HQ (address is collected via Clearbit) to that person’s name. Another lever is to create the perception of humanness. G automated the creation of a custom screenshot of the customer’s website including the Drift chat, then engages the reciprocity rule (cf. check Robert Cialdini’s Influence) by starting an email with “I built this for you”. Simple. Personal. And effective.

Guillaume was our guest for a previous talk on How to Scale Marketing. Check the takeaways and full video recap of his talk.

Subscribe to get notified of the latest Scale talks.

About Guillaume Cabane:

“G” was VP Growth at Drift and formerly VP Growth at Segment, putting both companies on track towards hypergrowth through aggressive experimentation and extensive personalization. In this episode, he shares some of his new experiments as well as the playbook that helped him build — and scale — some of the top-performing growth teams in SaaS today.