By STEVE SHAW | DIGITAL Director | JuLY 2015
Following a talk I did earlier in the year about getting started with Marketing Automation, I’ve found that a lot of major brands are actually facing very similar problems to each other when getting some traction and results from their Marketing Automation strategies.
At various events, I’ve managed to speak to senior marketing representatives of some major UK brands. It appears that the barrier to delivering results from marketing automation are very common, some of which are easily solved, with others being a little more complex. In this article, I will take a look at some of the challenges and topics that have been resonating with the brands including:
- How to increase the volume of automation for marketing messages
- How to factor behavioural data into your plans
- Attribution and the effectiveness of automation
- Testing and optimisation
- Big data and a single customer view
- Getting the most from your technology
It’s worth mentioning at this point that Marketing Automation, whilst primarily coming in the form of emails at the moment, can and should include other channels and formats such as social media and mobile.
Increasing automation and reducing the manual process
The first topic I would like to look at is the brand experience. In moving from their manual process to more automated sending of marketing messages, most of the brands said they were currently running at an approximate split between 80% of manual and 20% of automated marketing emails. However, they were looking to move to a more balanced approach of around 50/50 as soon as possible, not only to ease the process but to make it more cost-effective.
What quickly became clear is that within large organisations, there are a lot of different teams carrying out different activities and, due to the way businesses grow, there are a lot of legacy situations where messages are being sent without the knowledge of the wider teams.
This is where the planning aspect needs to start. In order to be successful with any form of digital transformation, you need to know the scale of the task you are working on and what has been set as the foundation.
Start the process by documenting all of the teams that are currently sending marketing messages, be it the marketing, IT or customer service teams. ~Then, engage with these teams early on to understand the process they go through, what the messages are that they send and the KPIs that they work to, thereby giving you the basis for a plan. This early engagement can help develop advocates across the business, reduce the fear of change, outline the vision and value of the change and show that the returns are too strong to be held back.
The next step is to understand how you can drive consistency in the messages; the template, format and structure – these should all become one, which means that the attention can be focused on the content contained within.
With your marketing messages list and a template ready to go, you can then start to roadmap the approach to automation. Often, the brand teams looked for the quick wins, such as ‘which is the biggest email list or most difficult process we could automate to produce savings?’
Whilst this is a good idea, I recommend starting from a different perspective and looking for the message that can be automated which has enough volume to show improvements to process, but also not mission-critical that if it fails, your transformation process will be marked a failure from the start.
The next step is to start splitting each scheduled email into relevant target segments. This means emails will be generated with dynamic blocks of content based on profiling criteria rather than going for a ‘one size fits all’ approach.
For the brands that already have this process in action, they made a very interesting point that, in the change from mass marketing to a more controlled, targeted message, they delivered a reduction in the volume of emails being sent but a large increase in the level of engagement and traffic generation produced. This is one of those times where “less is more” certainly applied and has an effect on the level of investment required.
Overall, when starting from zero, the best approach to getting started with automation is to start small and grow the strategy. If you have, say, 1,000 customers in your database, then start with small groups of around 50 or so and run tests to see how effective they are.
How to factor behavioural data into your plans
Using behavioural data is crucial to creating effective marketing messages, as it reflects back to the customer the understanding of their requirements and represents an opportunity to display relevance.
If you are thinking about using behavioural data in your plans, firstly, you need to consider the number of personas and segments that you may end up with. This is an important factor, as the more segments you include in your messaging, the more content that will need to be produced to accommodate them and remain relevant.
The best behavioural data you can include represents an acknowledgement of the customer in a way that isn’t invasive, which could be the number of previous orders placed, an abandoned basket or something that isn’t too personal. Using highly detailed or personal information should only be used in account areas or something that is very transactional.
When you are creating your behavioural groups or customer profiling, look for familiarity in your data before forming profiles and trying to make the data fit. For example, if you are looking at segmenting by age, query your data to see what broad ranges you end up with and then target accordingly; you may find you need age brackets for every three years rather than using five-year steps.
A standard process most brands use is to score the customers on how much they match a particular profile by carrying out actions or showing a certain behaviour pattern. However, one thing that is often over looked is using negative profile scores to adjust trends and derive more accurate profile matching.
For example, we discussed one of the brands who provide courses; they have a lot of students carrying out research around particular topics, so when engaging with the site they appear as a buying profile but ultimately, they aren’t the same as a buying customer. So, we look for ways to exclude these profiles from our marketing activity by using the negative scoring on other implicit or explicit data.
Attribution and the effectiveness of automation
One of the main concerns for the brands is metrics and how to measure the effectiveness of the marketing messages being sent. Recency, Frequency and Monetary value (RFM) metrics appears to be the most consistent method that brands use to identify whether customers are engaging with the messaging.
A very interesting approach employed by one brand was to look at which product affected the response rather than merely the channel or the surrounding message. This approach is a method of looking for your flagship product; this then leads to focused thinking about which are the most popular, what drives cross-sell/up-sell and bestselling products, which was ultimately one of the business goals of driving sales.
Another metric discussed, despite being very difficult to balance with ROI and basic revenue generation, was the use of social media. At the moment, the measurement of success in this realm seems to be reach, much like that of television or traditional media-wide marketing that, in general, the more customers you can reach, the more likely they are to engage and transact with your brand.
The most effective way of providing attribution was to understand all the touch points and campaigns the customer was involved in. The single customer view is an essential part of getting the most accurate attribution model across multiple channels.
Testing and optimisation
One of my favourite topics is the testing and optimisation of marketing. In terms of the automation world, my main recommendation here is to use your time in creating manual messages for testing what resonates with your audience. When you are confident that you have derived a success against your KPI, apply this to your automated messages. Then, carry out the test cycle again.
It was interesting to hear a lot of brands in agreement about a “don’t worry if it doesn’t deliver the results you want first time” attitude. A ‘fail fast and learn’ approach can often yield the best results; in learning what doesn’t work, you can quickly fine-tune what does.
One of the brands pointed out that ultimately, humans do some strange things and often, they can be very random. One test was carried out for a particular product which, prior to Christmas, gave a leader in an A/B test of 40%. The same test was run after Christmas and the B test won by 24%, showing a swing in statistics of 60% between the two options!
It’s clear to see why people can get frustrated with optimisation and show the need for several controlled tests in order to derive quality data and clear strategies. Another element to this conversation discussed propensity models and how they can be used to derive the effectiveness and the required outcomes to campaigns. However, so many factors are involved that they can be difficult and time-consuming to manage.
Finally, we had another example of pitfalls that can be avoided when discussing your plans and the relevancy of testing and, more importantly, challenging the results. One of the brands had previously carried out a testing exercise on their website which, when looking at a single metric, had extremely positive results. So much so in fact, the change was rushed through to their production environment.
However, following the release, the brand noticed an immediate decline in other key metrics that could have serious issues for the business if left alone. The lesson learned was to understand all of the business metrics being monitored and measured in the relevant areas rather than a single one, hence the need for a business and commercial mind set.
“When it comes to customer data, I strongly believe marketing should own this area; they are the ones who should fully understand the purpose of how and why the data will be used.”
Big data and a single customer view
A very popular phrase at the moment is ‘Big Data and the Single Customer View’, so it was no surprise that in conversation with the brands, this topic came up a lot. One of the most voiced challenges was that the customer data that the brand holds was rarely centralised, was being used by multiple owners and how, more often than not, this data appears in silos across lots of different systems.
It was offered up that one of the key things that causes this is a lack of a single owner within the business; the persona with the vision and purpose of all this customer data. For me, this is often an output of large organisations and brands where there is a battle between Marketing and IT departments.
Having a background in development, I know all too well how tech teams want to provide the solutions and how it can be very easy to lead technology or platform first to get to the quickest solution. However, when it comes to customer data, I strongly believe marketing should own this area; they are the ones who should fully understand the purpose of how and why the data will be used. They should have the vision and be the drivers, whilst the IT team should become the facilitators, offering ways to integrate, share and shape this data to be even more effective.
The disparate systems tend to be borne out of frustration, with not making progress with the current processes or systems, so a view of starting from fresh is often the least path of resistance in the short term but the biggest failing in the long term.
With the latest advances in technology however, it’s becoming a lot easier to get systems to integrate with each other. APIs, plugins and common formats are becoming standard in any SaaS, so the requirement for IT to invent a new way of communicating isn’t required. This leaves them to focus on the two-way communication with these platforms and how to evolve the scale of the data they will be controlling.
If your brand is looking to create a single customer view, the best approach is to define your marketing strategy first and how you need to shape and work with the data. If you aren’t creating a new system, then find one of your existing solutions that meet your requirements closest and build from there.
Getting the most from your technology
We got to discuss some of the familiar names in Marketing Automation including IBM, Salesforce, Marketo, Adobe, Sitecore, SilverPop, MailChimp and ExactTarget.
An interesting story regarding the adoption of the technology was the use of automation by a major airline and their air miles program. The success of this came not from the experts in Marketing, CRM, data or insights, but from a junior member of the team that took the time to understand the technology that was being implemented.
The junior team member took it upon themselves to own the solution and understand the power that it provided. They quickly grew the respect and authority and position within the team as a result. However, it is certainly a cautionary tale about being reliant on a single individual team member; should this person leave, you could end up without any marketing at all.
We discussed some of the skills that were present in the teams that are implementing marketing automation successfully; it was very interesting to see how there were so many elements that we felt were needed to drive success of the strategy:
- Marketing experience
- Data and CRM experience
- Business and commerciality mind set
It’s clear to see why it’s difficult to find specialists in this field when such skills are required.
During these conversations with brands, I’ve come to identify that there are three core levels to getting to the holy grail of effective marketing automation:
- Mass market messaging
- Segmentation one-to-many messaging
- Personalised one-to-one messaging
It was clear from the discussion that Marketing Automation is very much on everyone’s radar and is forming more and more of the marketing strategy alongside personalisation for brands. The biggest barrier to getting it right seems to be the large number of systems in use by various teams and the fact that we are still some way from the idealistic Single Customer View.
The quicker we can get to one view of the truth with the customer (the single customer view), the quicker we can begin to engage with them on relevant topics.
Marketing automation is not something which, for a large brand, should be entered into lightly, it’s a long term investment that needs strategy and resource, so making sure you have the right business case first is essential.
Overall, marketing automation is something that can have real business changing results. You can reduce your outgoing messaging costs; have happier customers receiving relevant targeted messages and more valuable, more engaged and frequently converting customers.
Read more thought leadership
Ok the title is click baity – Google Data Studio is an awesome tool and many of you are probably already using it on…
Day one of SEO training and you come away learning one thing, Google own search… But what if it doesn’t? Well at least when…
Previously, we’ve seen how R can be used to retrieve data from APIs such as Google Analytics. Often with data, you’ll conduct the same…