LinkedIn is one of the few social networks out there that has managed to convince investors and market watchers of its worth, thanks to a strong value proposition that is focused on selling into a B2B audience, rather than attempting to generate revenues around an engaged B2C one. Facebook and Twitter are great tools, but it’s not as easy to understand how they generate money. However, LinkedIn’s proposition relies on understanding the vast amount of data it has on 200m+ employees using the service and then providing insight to companies and recruiters that are looking to fill talent gaps.
Andres Bang, LinkedIn’s head of sales and operations systems, was speaking at Salesforce’s World Tour in London last week, where he was discussing the challenge of getting the right actionable data from the petabytes of ‘noise’ it owns, to LinkedIn’s own sales people. Although admitting it was a challenge, he was keen to emphasise that the data that LinkedIn holds is invaluable and unlike data held by any other company. He said:
“[We have] macro level company information, as well as micro level professional information. When someone gets promoted, we see that, when someone is a customer of ours and moves to a prospect of ours, we see that. When companies are hiring into their sales organisations, we see that. When someone comes on our website and downloads a whitepaper, we see that. How senior are they? Are they connected with other decision makers?
“We have some of the most amazing data scientists in the world. We also tap into public data stores out there, so we can calculate the health of the company”
But, there is a challenge for LinkedIn (not necessarily a bad one) whereby they don’t operate on a traditional model of having a consolidated IT department that is headed up by a CIO/CTO type. Instead, the company operates on siloed business systems that are dedicated towards driving revenue e.g. a data analytics team running on Teradata, the sales and operations team running on Salesforce. These are separate teams that need to integrate in order to get full insight from any data collected. Bang said:
“What do sales and marketing care about? Do they really care about the systems infrastructure? Do they care about the development language that they are building on top of? They don’t. They care about the tools to drive revenue. Traditionally CRM has been seen as a sales management tool. At LinkedIn, we try to get Salesforce to be a sales tool, get them to drive revenue with it.”
A single pane of glass
Bang gave delegates quite a bizarre comparison for how LinkedIn’s sales team operates. He said that instead of his sales reps being ‘cops on the beat’ searching for opportunities on the ground, he would rather they worked like Batman in Gotham City. By this he means that they are given the Bat signal (an opportunity that is highlighted thanks to clever data analysis), where they then go and target and execute, but can quickly come back out of Gotham (or in this case the sales system) and get back to their day jobs. This is done by providing LinkedIn sales reps with a simple, single “pane of glass” that operates on Salesforce.com’s platform with a custom UI, which is deeply integrated to a variety of other systems including Teradata for analytics, Oracle for financials etc. The integration is carried out by Dell Boomi. He said:
“You can basically get access to all the information that you need to do your job in what we call our single plane of glass. No more toggling between your lead gen system, to Outlook, to salesforce opps, etc. It’s all there.
“The day from the wire frame to the day we flipped the switch and went live, with 2000 sales people using it, was 4 months. Not only is that incredibly fast on the Salesforce side, but the amazing thing is that all the different data systems that we interface with, we got that up and running in four months as well. I have never seen an ETL project with that level of complexity get up and running that quickly.
“What Salesforce did for CRM world, I believe Dell Boomi is doing for the ETL world as well. You can hire someone that doesn’t have 10 years of engineering experience, can come in, learn the platform, integrate the front end to the back office in a number of weeks.”
LinkedIn gave one example of how when it began using Dell Boomi for its Oracle integration into Salesforce for close orders, which was previously being done manually by full time employees literally taking the data from one system and plugging it into the other. After experiencing rapid growth during 2010/2011 it was planning to hire eight more people to handle this, but after implementing Dell Boom this plan was scrapped and the new hires were repurposed for more “important stuff”. Dell Boomi integrations at LinkedIn have now been extended and Bang went on to say that the tool is so simple to use, that they only have one full time person managing it, which covers over 30 different integration points. Quite impressive.
Bang said that although Salesforce.com has been a great tool for LinkedIn and it is used heavily, it does have some limitations. For instance, LinkedIn uses Teradata and Hadoop for analytics and through integration surfaces the most important data to sales employees, rather than using native tools in Salesforce. However, Bang also wanted to analyse how the tool was being used by LinkedIn’s sales reps in order to understand how he could make sure they spend as little time in there as possible (the Batman theory), as he felt there was too much admin going on in Salesforce. He said:
“One of the limitations of Salesforce, is that for us we have a very complex sales order management process, so to fulfil that order we had to collect a lot of data points. It was just a very cumbersome process. Every time we asked a sales person to fill in a field on CRM, we are taking away time from them in the field selling. My colleagues and peers would boast about the time they were spending in CRM and Salesforce. It’s a stupid metric, the only thing that we value is how much you are selling and is the data we are providing you converting into sales.”
In order to get to this point, LinkedIn put Google Analytics on top of the ‘single pane of glass’ and analysed every click by sales reps, where they were spending time, the conversion metrics for every data point. This allowed Bang to strip away 80% of the data points in Salesforce and put a lot of the process behind the single pane of glass into the integration points. He said:
“Think about the average sales person on LinkedIn, they’ve got 2,000 accounts that they are covering, much smaller deals, much higher volume of transactions. 100 different opportunities at a time. What did I do? What’s my next step? How do I close this deal? It’s impossible.
“The reason we are able to do this is that we took all the complexity out of the UI, put it into the integration data. Behind this is a huge if-then statement: ‘if’ you put in this type of data ‘then’ we need to do this next, ‘if’ it’s this type of deal, ‘then’ this is what’s most likely to get you the win.
“We can see historically what type of industry you are selling to, what type of insight this originates from, what your peers are doing, which then tells you to take a certain action. You can stop, go away and the system will tell you what you should do. Custom UI, super fast integration, extremely powerful use of data.”
LinkedIn’s key lessons
Bang finished by explaining what his key lessons from the project were, they are pretty self explanatory so I’m going to list them here as some useful takeaways:
- Measure everything. If it can’t be measured, it can’t be managed. If it can’t be managed, it can’t be improved. LinkedIn was able to measure everything because we put Google analytics on top of it.
- Sewer and aqua ducts – Rome wasn’t great because of the cathedrals and coliseums, they got the basic infrastructure right so that the common person could get basic hygiene. For LinkedIn, before it did one thing on the Salesforce.com platform, it made sure that it got the data right. LinkedIn made sure it invested in an integration platform that was going to keep it up to speed and keep it agile.
- Gett users to get in and out of your systems as fast as possible.