How to keep your marketing automation campaigns from ruining your week

Safeguard your marketing automation campaigns and prevent embarrassing mishaps with these essential checks.

Marketing automation is a great tool. It allows companies to set up campaigns that run on their own without further human intervention. This saves time and greatly expands what your marketing team can accomplish with limited resources.

Unfortunately, marketing automation can also go wrong somewhat spectacularly. In this article I’ll provide a list of things to check to make sure your automated programs aren’t giving you or your company a big black eye.

‘Hi [First Name]’

This might be the classic example of marketing automation gone wrong. Personalizing an email can increase its effectiveness, so when you discover that you have first names in your email database, you change your subject line or salutation to include it.

But does every record have an appropriate first name for such a use? “Hi Mailstop 34” isn’t much better than “Hi [First Name].”

Before you insert a field into some marketing creative, make sure you have appropriate data in every record and make sure there’s a fallback such as “if the field is empty, insert ‘valued customer.’” Most systems can manage that.

Once you’ve accounted for every quirk and possible catastrophe from your list, you’re still not in the clear. Two months later, a salesperson will upload a new email list introducing new gotchas. By then, you’ve moved on to 17 other projects, and your simple personalization program is very far from the top of your mind.

Here are two ways to manage this. This particular example is about email, but you can apply the same principles to other campaigns.

Catch it before it happens. Create a requirements document for new data. For example, “every new email list must meet these standards.” Include the logic of your personalization campaign in that document. If people follow procedure (good luck on that), the errors will be caught and corrected before they go live.

Recover quickly. Make sure you’re on great terms with the customer service lead. Somebody who gets an email with the subject line “Hi Good prospect from the annual meeting” will forward it to customer service. Make sure they know they have your ear and that you will be very grateful when they clue you into this sort of mistake. For some reason, it can be like pulling teeth to get CSRs to tell you when something stupid happens. Encourage them regularly.

It might not be your fault

Some marketing automation problems result from bad planning, but often they’re a result of an unanticipated change in an underlying assumption.

• Step 1: What could go wrong?
• Step 2: How could we have known?

Let’s say you have a very small collection of webpages on a narrow topic. People who are interested in that topic are good prospects for a service you provide, so you set up a campaign to email everyone who visits one of those pages. (You can do that sort of thing with a customer data platform.)

These few pages are categorized a particular way in your content management system, so the setup is fairly easy. You rely on that category tag to identify the correct articles.

Two months later, the SEO guy finds out that this narrow topic is good for organic traffic, so he starts adding that category tag to many new pages. Now you’re bombarding your audience with emails.

How do you account for such things? The SEO guy is in another department. He doesn’t report to you or even tell you what he’s doing and he has no idea that his changes are turning your modest campaign into a nightmare.

Reporting is the answer. Keep track of the volume of each of your campaigns. If the volume suddenly changes, investigate.

Somebody ‘cleans up’ your data

You’ve set up a very efficient system where your lead funnel adds names and email addresses to your contact management software. Every new lead goes into your welcome series. They get a welcome email, a follow-up on Week 1, etc.

Then, some data scientist is told to clean up the sloppy records. Your new lead isn’t “new” at all. He was on the list 17 years ago. The data scientist merges the fresh record with the old record and that new / not-new contact falls out of the series.

The best solution here is to make friends. You won’t be able to find a report or a technological solution to every possible foul-up. You need to encourage an environment where people think, “This is in so and so’s domain and he’s a good egg. Maybe I should check with him before I mess with this.”

Keep a list of all running campaigns and review them

It’s simply too easy to fire and forget with marketing automation. You set up a great campaign and it serves you well for a time, but it just keeps running and running and nobody is watching it to make sure it’s still fit for purpose.

Keep a list, review it regularly and always put a sunset date on every campaign. You can extend the sunset date if it’s still working properly when you review it. Letting the campaign expire is better than allowing it to keep chugging along with something that doesn’t advance your goals.

Seed yourself

When you create a marketing automation campaign, find a way to seed yourself so you get the messages regularly. That won’t stop the “Hi [First Name]” problem because your record has a first name. But it can prevent some embarrassing problems.

When you get the seeds, check these things.

• Does it still use the right logo and brand markers?
• Do the links still work?
• Are the offers still valid?
• Is the contact information still correct?
• Does it still sound right? Something that sounded good when you wrote it last year might sound horrible today.

Read lots of reports

“Whatever can go wrong will go wrong” seems to be written into the fabric of creation. As a result, you can’t rely on your genius planning and operational excellence to prevent catastrophe.

Sometimes, your best recourse is to get an early warning when catastrophe strikes. If something slips and you catch it quickly, that’s far better than allowing it to linger for a month — or longer.

Create a special set of reports specifically related to marketing automation campaigns. One possible way to do this is with labels or a naming convention. For example, start the name of every marketing automation campaign with “MA-.”

Review those reports regularly. Look for peaks, valleys and weird deviations, such as …

• Are you getting more undeliverables than average in your MA campaigns? That might indicate you have a bad data source.
• Do your marketing automation campaigns have more unsubscribes than average or are they more regularly marked as spam. If so, you may be annoying your market.
• Are any individuals getting too many messages every week or month? Check with your email service provider to see if you can run such a report. When you see a blip (“John is getting 300 emails from us every month!”), start digging into the cause.
• Marketing automation isn’t just email. Check what you’re doing in all channels and before you set up any campaign make sure you have a method to monitor if, when and how it goes off the rails.

Monitor data connections

You don’t have to do this personally, but somebody has to. Marketing automation often relies on data from different systems.

For example, your order system might update the list that controls your cross-sell offers. That update relies on a connection between those systems, and sometimes connections fail – for various reasons. You need to make sure someone is notified when a connection fails.

Steer clear of marketing automation mishaps

Marketing automation relies on forethought, reports, monitoring and good operations. Don’t allow anybody to set up a marketing automation campaign without procedures to ensure you can be notified when it goes wrong.

• Is someone seeded?
• Does customer service know about this campaign?
• What back-end processes does this campaign rely on? Is someone monitoring those processes?
• Have you set a sunset date?
• Do you have a list of all your marketing automation campaigns and are you reviewing it regularly?

Time spent at the beginning thinking about what might go wrong and how to detect an error won’t catch every problem. But with luck, it will catch most of them.

____________________________________________________________

This article was written by MarTech and originally published here.

Author