Airbnb Data Scientists May Have Saved Marketing

It used to be that data was useful to only a few people in marketing: the decision-makers. Senior-most marketers would receive business intelligence around sales growth in different markets or Nielsen survey data around the most-viewed broadcast channels. Everyone else carried out the resultant plans: get a 3-month TV commercial campaign up on these local networks, or buy some billboard space on this stretch of highway.

Today, it’s still true that marketing data and information is only useful to decision-makers. It’s just that now, all marketers are making decisions. We need historical and real-time data to improve the content we create at all levels, from ad and social media copy and distribution to events experiences and art direction.

But when it comes to actually sharing and using that information, we aren’t making any progress. Marketing has changed more in the last five years than in the previous 50, but our knowledge management and ability to share actionable data — to become more intelligent as a team — hasn’t:

Fixing the state of knowledge management between marketers is tied up in the issue of transparency and visibility of relevant information, which most marketing software hasn’t even tried to address.

But there are some examples out there that hint at the path to effective knowledge management and thus a truly informed marketing department — even if they aren’t necessarily from marketing teams.

The data science team at Airbnb, which exists to find and share knowledge for the purposes of critical decision making, found itself in an unproductive cycle. When someone on the team got a research request, they would naturally start by asking teammates for past, related reports.

The problem: that scientist often couldn’t trust that those past reports were accurate or up-to-date. So they would end up recreating content from scratch and then distributing it through Google Drive and e-mail — and, eventually, the cycle would recur.

In their words:

Low quality research manifests as an environment of knowledge cacophony, where teams only read and trust research that they themselves created.

Their eventual knowledge management solution was a combination of tech and process that accomplished five goals:

• Reproducibility: This refers to cataloging each stage of the report. Other scientists should be able to view exactly how someone went from initial question to the final visualizations and conclusions.
• Quality: This means reviews and approvals for accuracy and value.
• Consumability: Airbnb said it wanted content to be both accessible to lay people, as well as aesthetically on-brand.
• Discoverability: If someone wanted to learn about a topic, it should be easy to find it or navigate to content without asking around.
• Learning: By looking at a report, another team member should be able to discover new best practices.

It isn’t hard to see parallels between the Airbnb data science team and the knowledge management problems of virtually any marketing team today:

• A need for transparency around what related work the team has already done
• The potential for reluctance around leveraging another teammate’s work
• Uncertainty around what works and what doesn’t
• A reviews process for content
• The mandate to deliver quality, on-brand messaging
• The sore lack of a navigable repository
• No method for best-practices sharing
  The solution for marketers, then — not just to grow the team’s general intelligence, but also to use it to get the best content in front of audiences quickly — largely maps to those same five goals:
•  Reproducibility: Do you have a system for recording the journey from initial campaign ideation and briefing — including feedback and brainstorming — all the way to final production, distribution, and results?
• Quality: Do you have an approvals process in place (for briefs, for artwork, for content, for publishing) that elevates your marketing collateral and campaign without compromising efficiency?
• Consumability: Is there consistency around how you speak with customers or clients in your marketing content — as well as consistency around what’s visually on-brand?
• Discoverability: Is there a single, searchable source of truth for your marketing assets — like campaign briefs, images and content produced, and campaign results?
• Learning: Does your system accomplish all of the above, making it easy for a marketer to actually learn from past efforts?
  At the end of the day, marketing needs all of the above to make sure information gets to where it needs — whether that’s information around past or current campaigns, performance, audience data, briefing instructions, brand guidelines, or even materials like pre-approved visuals. And we know that the combination of tech and process works, even across departments with teams Marketing has to coordinate with — after all, that’s how General Electric was able to get 24,000 employees on the same page. Improving knowledge management and the flow of information  will help teams and departments move forward in an age defined by constant change.

https://blog.percolate.com