Fully agree on this. When I was at Google we built a Chrome extension that would insert a customer flag directly in the support tool (i.e. high potential customer, try upselling). Bringing this data from a database or dashboard directly into the tool that they use at just the right time makes all the difference.
Really enjoyed the post Benn, as many previous ones. One thing that comes to mind about the experience though, is that sometimes building data as an experience embedded on the operational tool is not possible.
Marketing teams comes to mind when optimizing or building new campaigns or ads. They usually always like to have all the data from all their ad platforms in one place, yes, but regardless of that they either have different agencies who then use each ad platform to do the operational task, or they have to go not each ad platform for each change.
So what I’m trying to get to is, given that most companies now rely on different “apps”, “vendor software”, “SaaS” , for each specific task, and sometimes even for the same task, how would one approach to embed data I the way you describe here?
Just to make sure I'm following, I assume the concern is something like this: Marketing teams are running ads on Facebook, Google, and Twitter, and don't want to manage each campaign in each tool because you'd have to jump from the Facebook ad platform to Adwords and so on. That means embedding data in those tools is still incomplete, because it's a bad experience to move between them?
Assuming that's the question, I don't know exactly how we solve that, though I'd think there'd be services that aggregate all of that up currently (I'm assuming moving between those tools is a problem regardless of how much data is or isn't in them, so I'd guess people are already trying to solve that problem).
If that's true, then I'd also think those services are probably follow the Yelp model pretty well. Given how data oriented ad performance is, I'd think any kind of ad management software would already incorporate it. If all that's true, then ad platforms might already be living in this future.
With one caveat though: I suspect they mostly draw data from the ad providers. That could surely be enriched with other information, like traffic to different pages on your site, conversion rates on those pages, etc. So, for ads, it might be less about building a Yelp-like experience, and instead thinking about what a Yelp-like experience on top of a more complete data OS that has access to all of your data might look like.
Great post Benn. How do you think about addressing the nuance of business meaning associated with enterprise data as it pertains to the buffet? I think you start to touch on this, with diners mixing their peas in their pudding, so how do we balance a crowded buffet line with 4 types of puddings and limitless toppings + veggies (i.e the rise of citizen data scientists getting the right data)?
I've got two answers to that. First, I think a lot of questions don't actually require the buffet. Most non-analysts aren't actually trying to be analysts, and don't actually want to "explore" data all that much. People do this today because those are the tools we've given them. Instead, I think most people are looking to solve a problem (eg, choose a restaurant). With more disciplined and targeted tools (eg, Yelp), they could solve it without having to wade through a bunch of data that makes it hard to figure out what's right.
Second, in the cases when people do need to do real analysis, they should be able to more easily work alongside experts to do it. That's a big reason why I think generic consumption tools should be combined to include both self-serve and advanced tooling. Rather than pushing "non-technical" people to their own tools, everyone should work in one place, and work on these open-ended problems together.
Finally - and more importantly - I didn't get ay emails about us going to the ship...tell me Des jus left me off the email list.
Another great post Benn. I really enjoy your writing. The Yelp analogy is great.
I find I'm often asking teams and leaders the same question: "what is the product the team is offering?" I think this fits along the same lines as your point in this post. I receive a lot of funny looks when I ask this while leading teams that were labeled data platform teams. People think that is a ridiculous question, but I believe you need to crawl before you walk. You need to understand what you are providing and who your customer is. Offer your data as a singular product first. Then if users want to build something off of that data you iterate and expand to being a platform. I see many teams that are associated with data head down the path of providing the data with no context, shape or reason other than offering it. Whether that is access to the date lake, an API, etc. To your point, that is not the Yelp experience. That is Yelp existing as a list of restaurant information that you have to slice and dice to make useful. I believe many teams should shift their mindset and ask themselves what product they are offering and stop trying to sound more grandiose by being a platform.
Your Yelp analogy explains that really well. I'm going to borrow that!
Thanks! There are a few other "data as a product" posts out there as well, in case you haven't seen those (this one seems to be the most commonly cited: https://locallyoptimistic.com/post/run-your-data-team-like-a-product-team/). They take a more expansive view of what it means to be a product, but there are some similar ideas.
Fully agree on this. When I was at Google we built a Chrome extension that would insert a customer flag directly in the support tool (i.e. high potential customer, try upselling). Bringing this data from a database or dashboard directly into the tool that they use at just the right time makes all the difference.
i love this
Really enjoyed the post Benn, as many previous ones. One thing that comes to mind about the experience though, is that sometimes building data as an experience embedded on the operational tool is not possible.
Marketing teams comes to mind when optimizing or building new campaigns or ads. They usually always like to have all the data from all their ad platforms in one place, yes, but regardless of that they either have different agencies who then use each ad platform to do the operational task, or they have to go not each ad platform for each change.
So what I’m trying to get to is, given that most companies now rely on different “apps”, “vendor software”, “SaaS” , for each specific task, and sometimes even for the same task, how would one approach to embed data I the way you describe here?
Just to make sure I'm following, I assume the concern is something like this: Marketing teams are running ads on Facebook, Google, and Twitter, and don't want to manage each campaign in each tool because you'd have to jump from the Facebook ad platform to Adwords and so on. That means embedding data in those tools is still incomplete, because it's a bad experience to move between them?
Assuming that's the question, I don't know exactly how we solve that, though I'd think there'd be services that aggregate all of that up currently (I'm assuming moving between those tools is a problem regardless of how much data is or isn't in them, so I'd guess people are already trying to solve that problem).
If that's true, then I'd also think those services are probably follow the Yelp model pretty well. Given how data oriented ad performance is, I'd think any kind of ad management software would already incorporate it. If all that's true, then ad platforms might already be living in this future.
With one caveat though: I suspect they mostly draw data from the ad providers. That could surely be enriched with other information, like traffic to different pages on your site, conversion rates on those pages, etc. So, for ads, it might be less about building a Yelp-like experience, and instead thinking about what a Yelp-like experience on top of a more complete data OS that has access to all of your data might look like.
Great post Benn. How do you think about addressing the nuance of business meaning associated with enterprise data as it pertains to the buffet? I think you start to touch on this, with diners mixing their peas in their pudding, so how do we balance a crowded buffet line with 4 types of puddings and limitless toppings + veggies (i.e the rise of citizen data scientists getting the right data)?
I've got two answers to that. First, I think a lot of questions don't actually require the buffet. Most non-analysts aren't actually trying to be analysts, and don't actually want to "explore" data all that much. People do this today because those are the tools we've given them. Instead, I think most people are looking to solve a problem (eg, choose a restaurant). With more disciplined and targeted tools (eg, Yelp), they could solve it without having to wade through a bunch of data that makes it hard to figure out what's right.
Second, in the cases when people do need to do real analysis, they should be able to more easily work alongside experts to do it. That's a big reason why I think generic consumption tools should be combined to include both self-serve and advanced tooling. Rather than pushing "non-technical" people to their own tools, everyone should work in one place, and work on these open-ended problems together.
Finally - and more importantly - I didn't get ay emails about us going to the ship...tell me Des jus left me off the email list.
Another great post Benn. I really enjoy your writing. The Yelp analogy is great.
I find I'm often asking teams and leaders the same question: "what is the product the team is offering?" I think this fits along the same lines as your point in this post. I receive a lot of funny looks when I ask this while leading teams that were labeled data platform teams. People think that is a ridiculous question, but I believe you need to crawl before you walk. You need to understand what you are providing and who your customer is. Offer your data as a singular product first. Then if users want to build something off of that data you iterate and expand to being a platform. I see many teams that are associated with data head down the path of providing the data with no context, shape or reason other than offering it. Whether that is access to the date lake, an API, etc. To your point, that is not the Yelp experience. That is Yelp existing as a list of restaurant information that you have to slice and dice to make useful. I believe many teams should shift their mindset and ask themselves what product they are offering and stop trying to sound more grandiose by being a platform.
Your Yelp analogy explains that really well. I'm going to borrow that!
Thanks! There are a few other "data as a product" posts out there as well, in case you haven't seen those (this one seems to be the most commonly cited: https://locallyoptimistic.com/post/run-your-data-team-like-a-product-team/). They take a more expansive view of what it means to be a product, but there are some similar ideas.
As usual a thought provoking piece, Benn. Great work!