by James Lawson, consulting editor
Integration remains a key challenge for marketers in today’s digital world.
Raj Balasundaram | Global Head of Solutions, Emarsys
Robin Collyer | Leader in Next Best Action & Contextual Customer Engagement, Pegasystems
Tom Hutchings | Head of Insight & Analysis, Quant
Adam Oldfield | Managing Director, Force24
Kiyoto Tamura | VP of Marketing, Treasure Data
Nick Worth | CMO, Selligent
As digital continues its global takeover, automation is now vital just to cope with, let alone optimise, today’s complex marketing processes. However marketing still struggles to fuse disparate systems and data into a workable solution.
“It’s long past time for organisations to stop shouting and start listening,” says Robin Collyer, Leader in Next Best Action & Contextual Customer Engagement at Pegasystems. “To do that successfully, they need to get away from a channel-centric view and start thinking with one central brain.”
As Collyer’s title suggests, Pegasystems thinks handing off decisioning to a dedicated engine is the way to orchestrate the many different channel systems that large organisations employ. That kind of Marketing Automation (MA) solution demands a mix of static data and real-time feeds from critical sources like web sites or social, with links running in both directions to ensure the very latest information is always used as the basis of each decision.
“We build up a custom ‘movie’, an ongoing interaction picture, and use that to inform Next Best Action (NBA),” says Collyer. “We’re moving to a world when that NBA movie will populate at the time the customer, for example, opens an email.”
But when there are so many data sources and when marketing automation technology itself is so complicated – chiefmartec lists over 3800 different MA solutions – how can you procure and build an infrastructure that supports this kind of true hands-off, multi-channel operation? Even with less ambitious strategies, marketers must master data flows, pick out critical events and then deliver the appropriate reaction as swiftly as possible.
“Sending contextually relevant messages to consumers in the moment can be difficult when data and execution functions reside in different systems,” agrees Nick Worth, CMO at Selligent. “That challenge can be largely overcome by choosing single-vendor modular solutions that have native integrations for the core functionality.”
The traditional answer to integration woes, modular suites started with the big b2c vendors as they built or bought in new functionality. The newer, more b2b-focused vendors have also taken the modular route to add coverage for channels like social to their own core platforms.
As it often comes built into the initial software, adding extra native functionality to a suite is a matter of hours or even minutes. Configuration rather than customisation is also the aim for the likes of Apteco, well known for offering pre-built connectors from their own MA and analytics offerings to third party channel systems.
At MA specialist Force24, managing director Adam Oldfield claims his MA platform can connect and integrate seamlessly with over 700 different data points. “The most intuitive MA solutions are now enabling marketers to become ‘super-connected’, connecting to programmes like Slack, Skype, Google, CRM systems and more,” he says.
However, these integrations tend to be derived from previous bespoke client projects while there will always be new, unknown channel systems and of course data sources to deal with. Tom Hutchings, Head of Insight & Analysis at Quant Marketing, cites Email Service Providers (ESPs) as one example.
“ESPs can only be integrated if there is enough demand,” he says. “Developing new channel integrations can be time consuming and cause delays in the set-up because they are managed by third parties. Given the broad range of service providers in the market, it’s unlikely that your chosen MA tool will be integrated with all your client’s favoured options.”
As MA infrastructure complexity and heterogeneity increases, more vendors are positioning themselves as open, integration-capable hubs while dedicated integration middleware is also gaining more adherents for marketing applications. Mulesoft’s Anypoint Platform is one of the leaders in the latter, connecting on-premise and cloud apps, data and devices across SOA, SaaS, and APIs.
“Mobile companies in particular who are tech-heavy are tending to go towards Enterprise Bus-style architectures to orchestrate other standards,” notes Raj Balasundaram, Global Head of Solutions at Emarsys. “They often tend to use what they have in place already rather than buying in separate systems for marketing.”
The modern standards built into SaaS tools themselves are also making integration swifter and simpler. JSON is a good example, helping data ingestion and storage systems speak in a common format.
“It significantly reduces the cost of exchanging, parsing and communicating data,” says Kiyoto Tamura, VP of Marketing for Treasure Data. “This is why our open source data collector Fluentd, in use at companies like Atlassian, Microsoft and Nintendo, embraced JSON as the common data format from day one.”
However, these open IT standards work at a low level and only go so far in enabling commonality between disparate systems. For example, each company and its suppliers will still have to define for themselves how best to represent the many different variables used to define customers, campaigns and treatments that are exchanged between systems.
“My sense is that we need something layered on top of JSON because it only addresses syntax and format issues, not semantics,” says Tamura. “For example, which field should be interpreted as the vendor-specific customer ID? These things require standardisation and as far as I know, no open effort is in place yet.”
If a modern infrastructure is already in place, it will certainly facilitate MA integration, but simply plugging in any old data to disparate applications doesn’t in itself enable marketing. Wearable fitness vendor Runtastic is a case in point.
Though it had a bang-up-to-date, cloud-based set of IT platforms, it was still unable to form a coherent view of customer data and use it to inform its marketing actions. To trigger relevant and engaging messages to subscribers, it needed a system capable of processing huge amounts of activity data in real time.
“They had a customer database but it wasn’t properly structured for marketing, which led to massive amounts of attrition,” says Balasundaram. “You need marketing to lead integration efforts rather than IT. Because Runtastic now has our open marketing architecture, it makes it very easy for them to consume data.”
The latest buzz phrase on the data integration side of MA is the Customer Data Platform (CDP). Coined by industry guru David Raab a few years back, he defines a CDP as a marketer-controlled system that builds a multi-sourced customer database and exposes it to external execution systems. The last criterion is the vital one: systems only qualify if their data (or derived data such as model scores) is available to other systems for campaigns and messaging.
That still gives CDPs a pretty wide remit, running from B2B predictive lead scoring and customer success management to data management platforms (DMPs) that support online advertising. Marketing clouds with an integrated customer database are natural contenders too.
Selligent is one example, with its built-in single customer view functionality making customer profile information available to users of the different modules within its engagement platform. “Simplifying the marketers’ ability to use SCV data creates the real opportunity for winning action,” says worth.
But because the use cases for customer data are so broad, a single vendor is always going to struggle cover everything with enough flexibility, scale and ease of use. That native marketing cloud ability to push and pull data to and from the different channels it supports must always extend to other third party systems. Marketers need data repositories – or CDPs if you will – that impose as few limitations as possible.
“My big gripe with the Ensighten, Lytics and BlueVenns of the world is that they can end up becoming data silos themselves,” says Tamura. “They do have APIs, but it does not give nearly the flexibility or scale of cloud data warehouses or Hadoop-style analytics infrastructure. As marketers become more sophisticated, the level of flexibility and scalability typical in the “big data” systems and missing in MA will become an issue.”
Bringing MA customer data together with online display and so having to deal with non-PII records demonstrates this need for flexibility and scale clearly. Emarsys has deep experience here, building customer-specific data exchanges that work with both PII and non-PII data.
“If we are integrating with a DMP, you need a bespoke data exchange project but I expect this to be a standard production solution in the next 12 to 18 months,” says Balasundaram. “DMPs and MA will become one platform in future.”
Even when we do see vendors finally starting to converge the currently separate towolsets for MA, sales CRM and adtech into one suite, expert provider input will still be essential to build the database and put together the final system. Simply piping data into a warehouse is one thing but resolving identities across different systems like CRM, display ads or internal databases is far harder.
“In our experience, this is the most valuable and challenging phase in building a unified view of customers and also the area where we end up spending most of our time in onboarding our customers,” says Tamura.
Gaining this kind of knowledge through training or recruitment can very difficult and expensive for in-house client teams. They will also often have to move quickly to meet short project timescales.
The larger enterprise MA suites in particular are often sold via a network of partners and resellers who have already invested considerable time and effort in technical training to configure, integrate and optimise the functionality of the software. They will have hands-on experience gained through previous integrations with detailed understanding of releases, software quirks and bugs). However the same level of support may not be there for today’s midmarket SaaS tools.
“Often clients don’t have the appropriate technical skills or available resource within their existing IT teams,” says Hutchings. “Migrations also complicate things further with the potential need to run old and new systems in parallel.”
Turning to an MSP to build and run their marketing systems has long been a popular solution for marketing, solving many of those in-house integration and resource issues. Many suppliers now offer this kind of support for the latest generation of MA tools.
“Provided they have the experience and skill sets to deliver the project, then the agency should deliver a better solution in less time,” says Hutchings. “We’ve been told by software vendors that the biggest reason for client churn is under-use of the MA system.”
Following system installation, expert support helps train up new users rapidly, build out new interaction strategies and rapidly achieve ROI. There are also often other related services like email delivery and campaign management available.
“Ideally agency involvement should only be a short-term fix, perhaps to accelerate learning,” says Oldfield. “That said, an MA vendor that offers creative and strategic marketing support is the perfect hybrid.”
This is particularly the case with a US vendor that has a limited UK presence. Trying to obtain help from tech support in a different time zone is never ideal.
“The guidance of a UK-based partner that is entirely clued up on the goals of your specific organisation is often a huge help,” notes Oldfield.
Realising that deploying or extending MA systems will always throw up integration challenges is the first thing to accept. A new off-the-shelf platform might well have the potential to integrate with hundreds of other platforms and tools, but it won’t make those links by itself. With that in mind, marketing can plan ahead, set realistic lead times – and engage expert help if required.