Naked Marketing: The 5 Foundations for Big Data Marketing
In the first post of the Naked Marketing series, I confessed that our 60-day big data marketing implementation was only possible because of the work we’d done in the previous 18-24 months.
In this post, I want to give you an overview of that foundational work and explain why it’s been so important to our success with this new way of marketing.
To do that, I’ll discuss the Five Not-So-Easy-Pieces that we put together as our big data marketing foundation layer. But keep in mind: the things we needed to assemble and the order in which we tackled them reflected our own unique situation and our legacy stack. You may need to start in a different place.
Eyes on the prize: total alignment
The big win we were aiming for in transforming our marketing operations foundations was a total partnership between Marketing and Sales. Not the lip-service kind, the real kind.
In the past, we had the typical ‘throw it over the wall’ model, in which Marketing created leads and opportunities in a vacuum (our ‘pipeline’) and threw them over to Sales to convert into revenue.
That wasn’t working. Too few of the marketing-sourced opportunities were converting to actual revenue. So Sales stopped trusting our marketing machine. That meant they wouldn’t jump on everything we threw over the wall—so we stopped trusting them. Lose-Lose.
We knew that the new model had to be a complete and open collaboration between Marketing and Sales. After all, we really are in this together. And that meant a totally transparent operational model: where Sales could see everything we did to create leads and opportunities—and we could see everything they did to turn these into wins and losses.
At the start of this journey, we were nowhere near that ideal. And the state of our marketing operations—our data, applications, integrations, and processes—was a big reason.
So here’s what we did:
The 5 Foundations of Big Data Marketing
1: Getting our data — and our marketing automation — in order
The first thing we needed to fix was our data.
We’d spent a lot of time building a marketing database and it just wasn’t being used everywhere around the globe. That’s usually a sign that it’s not doing what it was meant to do.
We had rolled out a central service model, building a database that integrated our Salesforce and Eloqua data—but we’d run up against some license limitations (per record pricing) and found the size of the database was becoming really expensive.
So we knew we needed to re-negotiate and re-organize the data, and re-think our central warehouse.
Back in March 2014, our Eloqua contract was up for renewal and we had to move from v9 to v10. Since that would mean a pretty major migration anyway, we used it as an opportunity to re-evaluate our marketing automation platform and our data strategy.
Pretty quickly, we ended up with a shortlist of two: Eloqua and Marketo.
In the end, Marketo won because of ease of use (our non-technical marketers needed to be able to use it) and because we felt the integration with Salesforce would be easier and more robust.
The change also gave us an opportunity to clean up our processes and practices—to start with a clean slate, using all we’d learned with Eloqua—and set up a cleaner, simpler, better-integrated marketing automation platform.
What our new Marketo instance and database provides us with:
- A more usable and approachable database with better quality data (while we were at it we of course applied our Informatica Data Quality magic including Master Data Management (MDM)).
- Better and lower cost integration of Salesforce and Marketo data (complete two-way sync, with custom fields synced using our Informatica Data Integration product).
- A single source of truth for Sales and Marketing.
- Mutual accountability!
2: Getting our website and web analytics in order
Next up was our website and analytics. We’d previously built on SDL Tridium, which was a great CMS for localizing our content all over the world but had some serious limitations (chiefly around things like personalization, ease of use, and mobile-friendliness).
We were also using Adobe Analytics but a very vanilla instance of it, which was pretty bare bones—not much better than Google Analytics (which is fine for the basics but doesn’t let you do the kind of modeling and reporting we’d be needing).
We switched to the Adobe Experience Manager for our whole web stack and ramped up the analytics to where we needed them to be. We looked at lots of web platforms and ‘marketing clouds’ and felt that Adobe was way out ahead in integrating the stack—with strong personalization, responsive/mobile capabilities, and all the stuff we’d need (metadata capabilities for a robust taxonomy, tag management, etc.)
For web data geeks like me, it’s a great platform (and all Java-based and scalable).
Adobe also has a healthy developer and integrator ecosystem that would keep the platform growing and innovating. That was important to us. We don’t want to find ourselves in a dead end ever again.
Then we ramped up the analytics to do great visitor tracking, advanced analysis and really track down conversions and affinities. We can now, ultimately, map these to actual product interests on the web, leveraging our product taxonomy in the digital asset management system behind Adobe Experience Manager (visiting web pages and downloading assets for a product like MDM accrues to MDM product interest in the affinity matrix).
Without a tight, well-run, well-instrumented website, we just couldn’t have pulled off the big data marketing operations machine we knew we needed.
What our new Adobe Analytics did for us:
- Tight tracking of all (anonymous and known) visitor journeys.
- Hard-core web analytics to see page flows, referrers, repeat traffic, conversions, etc.
- Integration with Demandbase for reverse IP mapping, tracking of firmographics like industry, company size etc.
3: A scalable paid media and SEO program
Okay, the machine was getting into shape. Now we needed some traffic to pour into it.
We ramped up our content marketing program (with our partners Velocity in the U.K.), and used it to fuel a modern digital marketing program to get people into the top of our funnel.
The strategy was pretty simple: create compelling content on issues we knew our prospects cared about, then offer that content in an ongoing, multi-channel advertising and social media program that we optimized over time.
We tried all sorts of tactics—from paid search, content display, remarketing, and paid marketing programs on social channels, such as LinkedIn Lead Accelerator.
The result: pretty good traffic (about 30 percent of total web traffic) that we could then track as it moved around our world.
A word on campaign codes
The religious and consistent use of campaign codes for all marketing activities was absolutely critical to our marketing operations success. In the past, we’d let the campaign code hygiene slip a bit and our analytics suffered. Now we have a locked down campaign code protocol and if anyone breaks it, we take them out back and … Tough but fair.
What our media program did for us:
- Generated high volume of incremental traffic.
- Provided consistent campaign codes so we knew which marketing channels and programs net-new names, pipeline, and revenue came from.
- Optimized over time to get more and more efficient.
4: Predictive analytics to score leads better and a nurture approach
To ensure we were passing quality leads to sales, we needed to layer on the ability to score and nurture all leads.
Some of this we could do in Marketo, but to do a really good job, we needed some predictive analytics. For that, we turned to Lattice Engines—a predictive scoring platform that has really paid dividends.
What Lattice does is to analyze people/companies who actually bought our products, looking at many hundreds of data points for each one—everything from job title, company data including credit ratings, hiring profile, technology profiles, and location, to behavioral data, and shoe size (kidding about the last one, but it’s only a slight exaggeration).
Lattice then looks at all leads to score them (and re-score based on any additional response data) according to how similar they are to people who actually went on to buy. Leads are scored as A, B, C and D, with A being the highest propensity to buy.
In practice, the leads that scored ‘A’ turned out to be 6 times more likely to buy than the average leads we had previously passed over to Sales.
When that happens, the dynamic completely changes: Sales starts looking forward to leads from Marketing—to trust the marketing machine. And rightly so!
Of course, you can never simply “set and forget” predictive models. As we evolve and change, our marketing and sales methodology and practices, we have to verify that the model is still accurate or retrain aspects of it. Hot stuff.
What our predictive analytics provides us with:
- High-quality, high-fidelity lead scoring—trained on conversion to revenue (not just pipeline).
- Far higher conversion rates.
- Boost of sales productivity, because we are no longer asking for follow up on D leads (they go into nurture).
- A major credibility boost with Sales.
- Faster revenue!
5: Data integration and governance
The fifth piece of our marketing operations foundation was the ability to connect up all our data in a reliable way. This includes data integration between the various applications—but it also means taking all steps necessary to be able to join data from different systems down the road.
For example: We made sure that for every visitor their Marketo munchkin (cookie) was communicated to our Adobe Analytics system using Marketo API integration with Adobe Analytics. So we could have full confidence that we could directly tie our Adobe Analytics data to each prospect and their sessions.
A key part of all of this—and the entire big data marketing operations journey—is data hygiene and governance. To get the most out of big data, data quality is just as important as it is for traditional data warehouses.
When your entire revenue machine is built on data, you need to know the data is clean and accurate. The rigorous campaign codes discussed above (and intelligent campaign naming) are a great example of that, but it is only one of the new processes and policies we needed to implement in order to keep our data quality high.
If you take nothing else from this entire Naked Marketing series, I hope you take this:
Without consistent data rules and tags, you cannot do the analysis and tracking that big data marketing depends on, because you will not be able to connect the dots all the way to revenue!
And because we are in B2B, data mastering and hierarchies are another important component of governance. We’re using our own MDM platform so that leads and contacts are de-duplicated and all relationships are perfectly documented, e.g., which site does a contact belong to, which company does the site roll up to. For the latter, we use data from our partner Dun and Bradstreet (everyone knows DUNS numbers, right?) to help with that. It integrates natively via Web Services with our MDM, see Business Data Enrichment.
With data governance and data integration you can keep the campaign data in sync with the opportunity created in Salesforce even after you pass the lead over to Salesforce (the essential piece of the attribution puzzle).
Without it, the world goes dark.
What our data governance did for us:
- Keeps our data clean, accurate, and trustworthy.
- Let us track all the way to conversion (campaign codes!).
- Establish the first and last touch (and everything in between) for every sale.
So there you have it: our Five Big Marketing Operations Foundations.
We had all this in place before the 60-day sprint that I talked about in the first post, so now you can see that our starting line was already well down the track.
There’s no short cut to this kind of thing. The engine is only as good as each moving part and—above all—the quality of the data you pour in. In our next post, coming soon, we’ll look at why we took a big data approach to solve our challenge.