Automate Email Reply Management with Siftrock and Marketo

Since I started using Marketo, I’ve seen quite a few people ask if Marketo can automatically handle Out of Office replies to mass emails. Email replies can be mined for valuable list building details, including new leads, bad leads, or organizational changes. Which salesperson doesn’t want a “Let’s chat” reply? Wouldn’t Marketing love to record a reply as a Success in a Program?

With thousands of email replies each day or week, there’s no way to keep up. Salespeople filter them out, never bothering to look. Marketers have other things to do after filtering unsubscribes. Whether you use an email alias with a proxy box to manage “marketing@yourfirm.com” or you use Lead Owner Email Address, reply management can become a real problem for everyone.

Reply types often include:

• Out of Office (with or without Please contact…)
• Left Company or Retired
• Unsubscribe or hate replies.
• Human replies such as “I’d love to chat…” or “Let’s talk in a few months”
• Spam Filter – click here to verify you aren’t a bot/spammer.
• Soft Bounce – temporary error.
• Hard Bounce – permanent error such as User Not Found, etc.

There are undoubtedly other categories and if a human has to go through a few hundred, or a few thousand per day, there is no opportunity to scale your system.

Why look at replies in the first place?

It turns out automated replies from your audience can be helpful to your sales efforts. Each category can help generate a new lead, or help you clean your list. RingLead coined this process as “re-bound marketing.” Salespeople often use the replies in this way or to expand the buying team while someone is away. I know as a Salesperson, I would scramble if I suddenly saw my key buyer left the company!  I also received a lot of forwards from our marketing alias that turned into great opportunities.

The challenge, of course, is if you send thousands or millions of emails per day, or month, or per year, how can you reasonably manage the volume? Enter a tool called Siftrock.

[Disclosure: I was given a trial account for this How To and have purchased Siftrock for an employer.]

Reply Type Use Cases

Before we delve into the setup of Siftrock, let’s create a few common use cases for reply management.

Type of Reply	Action to Take	Notes & Considertions
Hard Bounce	If Hard Bounce, Email Invalid=T	Hard bounces can vary in type, such as User Not Found, Left Firm, or “Your domain is marked as a spammer.”
Soft Bounce	Marketing Suspend for 14 days	Usually a temporary error or OOO reply.
Out of Office	Marketing Suspend for 14 days Extract new leads	Would be great to identify date they return, but that’s in the future. Can we extract other contacts from their reply, and is that legal?
Vacation	Marketing Suspend for 14 days Extract new leads	Would be great to identify date they return, but that’s in the future. Can we extract other contacts from their reply, and is that legal?
Left Company or Retired	Email Invalid=T Lead Status=Left Company Revenue Stage=Junk Extract new leads	Can we find out who replaced them? Can we extract other contacts from their reply, and is that legal?
System or Unknown	Human to review	Check the Siftrock Inbox
Human	Human to review	Forward to Sales Rep or check Siftrock Inbox

Siftrock System Setup

When setting up Siftrock for the Trial or paid version, you need to ask IT to do several DNS changes. From experience, I recommend not starting your Trial until the DNS changes are fully verified. Siftrock did keep extending the trial until this part was setup. Siftrock now has a “quick start” option to receive forwards from an existing box, however, I did not test this feature.

Step 1: decide on a CNAME that will route replies back to Siftrock’s SMTP server. If your IT Security team has concerns about this, then discuss it with Siftrock’s team. As long as you carefully follow the DNS steps, you should be ok. Let’s say you use:

reply.marketingrockstarguides.com

Step 2: Technical Setup – go through the steps on your DNS record.

Step 3: DKIM/SPF – yes, you should do this. While this isn’t 100% necessary to start, you will experience the same deliverability pitfalls you would when asking Marketo to send on your domain’s behalf. Setup DKIM/SPF for your Siftrock CNAME and test it with a Gmail account and other tools.

Step 4: Marketo Integration – Siftrock does require API access to at least the key email management fields like Unsubscribe. However, if you decide to allow Siftrock to add new Leads based on the reply parsing, Siftrock will need full API access. Remember, Siftrock won’t take actions you don’t ask it to. Siftrock also integrates directly with Act On, HubSpot, and Eloqua.

Deliverability Questions

When testing, I did not see any impact to deliverability, with several small tests from 500 to 2000, then to 100,000 per send. Sub-domains typically take on the reputation of the main domain or the actual sender IP. The sender IP is still your Marketo (or Eloqua) instance IP, however, the Reply To IP will be Siftrock. This should work as intended as long as DKIM/SPF are properly handled inside Marketo.

Updating an Email Asset in Marketo

As part of the setup process, I recommend creating a test program and to clone a real email asset which can be tested in the wild.

Modify the Email From and Email Reply To to use something like

hello@reply.marketingrockstarguides.com

in both fields. Any replies will go directly to Siftrock’s server for processing. You will not see a reply to your corporate inbox. There is no need to create such a box at your firm’s email server.  I felt a bit weird not having a box on my server, so think of Siftrock as your new email box for these aliases. You will have the opportunity to review all Human or Unknown replies within Siftrock.

Siftrock Workflows and Types

Using our Use Case table above, we’ll create a few workflows to tell Siftrock what to do with each reply type. In this post, we’ll discuss the Marketo version, although other connectors may have similar functions. It’s important for you to review Siftrock’s reply type list because it constrains the workflow creation even though you can create branching and filtering logic once you choose a Reply Type.

Before you Begin

I recommend using the above Reply Type Table to map against Siftrock’s types as well as to determine how to treat each Type. In Marketo, you will want to create a Program to contain all of the Static Lists to use for each Scenario. You may also need to create Siftrock mirror fields or even trigger flows as you decide how to handle incoming data. Do this first.

For example, Siftrock won’t assign a lead to the Sales Rep, so if you create a New Contact from a Reply, you’ll need a workflow in Marketo to pass this lead back to SFDC. It is possible to leverage a mapped “Lead Owner” field that exists in Marketo to Siftrock, then create a forwarding rule in Siftrock to route the email to the correct person, or a shared Sales Alias.

Create a New Contact from a Reply

This use case can be a powerful aid in building a list. There are some concerns I personally have with how legal it is to just add new people to your list. I’d consider simply forwarding such leads to Sales, or at least Suspending them and sending to Sales. Siftrock walks you through this case, but here’s a summary.

Keep in mind you can go to the Inbox or New Contact list to review flagged leads for export or other needs.

Unsubscribe a Lead from a Reply (Human)

This was my original reason to get Siftrock up and running. I am a bit surprised there isn’t a default Type for this. It’s easy enough to setup.

Other Human or Unknown Replies

Also a powerful aid in list management. The replies that fall in here will be ones that aren’t clear to the machine, or are clearly human replies. You will also find non English languages in this inbox. If you are heavily non-English, you may consider creating a lot of custom filters or branching based on the email alias used.

Handling Lead Owner Email Address Replies

I haven’t tested this myself yet. A lot of firms rely on the Lead Owner Email Address that’s tied to Lead/Contact Ownership in SFDC. This allows marketing to send out emails that look like they come from Sales Reps. With the advent of Account Based Marketing (think tools like Yesware, Outreach, Engagio, and Salesloft), some reps already have some level of reply management. For marketers who still own this nurture, however, it’s important to still have the leads matched to the right rep, yet be able to send personalized emails at scale. To take the reply burden from Sales, Siftrock can be used.

From SalesOps’s point of view, it isn’t practical to reconfigure SFDC to have the User’s email address changed to be “josh@reply.company.com”. Instead, use a formula field in SFDC to pass back the Siftrock friendly email that you can use as a token. Roughly, here’s what to do:

1. Create a new field in SFDC on Lead and Contact, and make sure it is visible to Marketo.
2. In SFDC, add the the Formula:

if Lead Owner Email Address is josh@company.com, then field value=josh@reply.company.com

then take that token and replace it inside each Email Asset. Then test it before going at scale. In some cases, you may need to tweak the process to ensure Marketo sees the field or values properly.

You will go through all of this setup to ensure you can automatically take any Human reply to forward directly to the Salesperson. Both Marketing and Sales benefit tremendously from the automation.

Drawbacks of the System

The only concerns I had when using Siftrock is the Workflow interface is not immediately obvious. It’s not hard per se, but it’s not organized in the same way Marketo flows are or in a clear “If Reply is of Type X, then do this.” Yes, it does work like that, but interface doesn’t make it clear. The confusion is related to the use of colors and icons. For example, I accidentally cloned or edited workflows without intending to. It’s not immediately clear which workflows are On or Off, or if there are errors with them. When creating Workflows, Siftrock automatically turns them On, which may not be desirable with a fully live system.

There are a few use cases Siftrock is working on because of natural language processing complexities. While their processing is good, the reality is no computer is 100% at identifying subtleties in language the way a human would. This is more about setting your expectations if you are thinking of using Siftrock or a competitor. If your company works with a lot of non-English email, this tool will have limits. I’m pretty sure it will be easy to setup a lot of non-English rules using the Human/Unknown Reply Types. I don’t know if Siftrock plans to add other languages in the near future.

Terminology is a little muddled with Siftrock. For example, they say a Reply Type of “Bounce” means “soft bounce,” while  Marketo uses “Bounce” as “Hard Bounce.” The word “General” means “Any Out of Office not otherwise classified.” There are a few other examples where we had to clarify precisely what was happening.

None of these drawbacks were critical to deploying the tool at all. Siftrock has been very helpful throughout the process.

Conclusions

Tools like Siftrock are a solid martech stack add on that helps marketing ops scale systems for just a few hours of setup work. Pricing is based on reply volume, which may not be a typical metric, so Siftrock estimates 2-3% of total send volume will have a reply. Those with very high email volumes may want to start with 1.5% as a minimum volume estimate. I recently read that human replies for B2B are at .02%.

Depending on your email volume and internal need for testing, you may want to ramp up slowly using certain types of emails. Unlike some vendors, Siftrock will warn you of overages before making price adjustments.

I recommend Siftrock to any marketing ops team. Humans should not spend hours filtering email replies – it’s not scalable. Siftrock’s support is solid, they are open to suggestions, and the product works.

[Updated Feb 3 for minor grammar changes; updated reply data]