Omnichannel lead attribution for real estate marketing

[Morgan]

My good friend Ken Pozek @Pozek - reached out to me today and asked me to point my brain towards conversion tracking for real estate marketers that use multiple lead sources.

In his case, for example, he does a lot of PPC with REW, a bit of SEO, but he is also huge in the Youtube space. they work on Google My Business and all kinds of other marketing.

He’s in a lot of places

The challenge he’s looking to solve is not a new one, and it’s something that very sophisticated marketing companies look at often. What he’s really describing is “Omnichannel attribution” - there is a great article on different types at True Profits OmniChannel Vs Multie Channel Comparison

If you read the article (and you should) they describe several types of attribution.

From the very simplistic: First touch attribution (this is what we do now at REW, where the originating source of the lead is given credit) or “Last touch attribution” - where the last referrer is given credit for the sale

To more complex concepts like Linear, Time based or data modeling for attribution. It can get quite complex and very interesting.

Complex = expensive (And not always what is needed, it can often be overkill)

As I’m sitting and thinking this through, I think the linear approach is the best place to start.

It’s a good mix of being more complex than simply crediting the first or the last referrer but not so costly as having to spend millions on AI integrations and data scientists only to glean slightly better (or not better at all) results.

So what might that look like (and where might one build this) in the backend of REW?

Let’s brainstorm!

First, let’s help you understand what we’re talking about here:

When someone visits your website from a marketing channel, we capture that as what we call the “referrer” - here’s an example screenshot from our visits tab.

REW already tracks this (and from what I can tell, on each visit though I’ll have to confirm with @FernandoOrtiz and @Aayaam)

So if it is true that we are tracking each referrer for each visit - the “data” part is already taken care of. That’s a huge bonus as we can save a lot of time and cost.

Next, what are we trying to figure out here?

Simple really (reminder some things that are “simple” to understand are complex to code) but the idea is rather simple: Let me give you an example.

Let’s say someone comes to Ken’s site from Google PPC: https://www.theorlandoreal.com from a Real Estate Webmasters Real Estate PPC campaign but they don’t convert right away.

A few weeks/months later, they come into contact with Ken’s awesome Youtube Channel and they are driven back to the site. We would then have two referrers (PPC originally and Youtube).

And THEN perhaps Ken runs a Facebook campaign, and they once again end up back on his site, this time referred by Facebook.

And now they finally convert after at least 3 visits from at least 3 sources.

We now have 3 distinct referrers for that lead. (Google PPC, Youtube, Facebook)

The concept of linear omnichannel attribution would take each of these into account and give each of them (since there are 3) 1/3 credit for the conversion or .33%

That’s the easiest way to do this (we can get more complex after with ideas of weighted distribution etc) but for now stick with me on the linear model just so we can get through the math.

Now, let’s say the commission on the “sale” is $9,000.

All you have to do is credit each source with $3,000 (1/3) of the commission.

So, instead of the old model (First touch attribution) getting credit for a $9,000 GCI and 1 “deal” - that same source would get credit of 1/3 of a conversion and $3,000 GCI.

Does this all make sense so far?

This reporting would go directly into the REW Sources application, but we’d need to make a few changes (such as the ability to support fractions of a deal in the reporting)

A logical result of this reporting would be that the “true ROI” of a source would actually go down (since you’re still spending the same $ on the original source, but now it shares it’s credit with any follow up sources)

BUT (and this is the purpose here) - you would have a better sense of the “real” cost to land that deal from a marketing perspective, not just the cost to get the lead in the door in the first place.

Ideally, with observation and optimization based on the data - you would start making omnichannel marketing choices that would “increase” the conversion rate and thus actually reduce the “per source” cost per acquisition.

Do you get a lift on PPC lead conversions when you expose them to your Youtube channel afterward? How about if they also see you on LinkedIn? Wouldn’t that be great to know?

Now let’s geek out a bit on some numbers and see how this might work:

One of the coolest (and most under utilized) parts of the REW CRM is sources. The ability to literally track all of your marketing dollars and have your conversions, their rates and ROI’s automatically calculated for you. (Seriously if you’re not using this, you are missing out!)

So starting with the Google PPC example (real data from this YTD)

We’ve got 1,080 leads generated so far YTD (Cool beans), and we know the CPL is $14.75

Ideally those leads (over time in the database) will convert 2% to sale but let’s use 1.5% for more conservative math (my team does this, so I know it’s possible for any doubters)

That’s just over 16 deals.

If each deal were worth $10k in commissions, then the gross GCI generated (over time) would be $160,000 on a $16k spend. That’s a 10/1 ROI (we’ll take it!)

But what if we were to “also” spend another $5k promoting our Youtube Channel to these leads? And let’s say half of them came back via youtube (so 540 leads came back via youtube)

And then we spend another $6k retargetting on Facebook, and the other half came via facebook.

If the conversion rate didn’t increase, then your “actual” marketing spend for these deals would be:

$8,000 PPC + $5,000 Youtube = $13,000 for half of your leads

And

$8,000 PPC + $6,000 Facebook = $14,000 for the other half.

So instead of a $16k spend you’ve actually spent $27,000 to earn that same $160,000

So, instead of 10/,1 it’s more like 6/1

Are you with me so far?

And now for the credit calculations: (Omnichannel attribution)

Since we’re using “Linear” distribution in this example, then the various sources must “share” credit for each conversion.

Google PPC factored in each scenario but “shared” credit with one other source. So Google PPC would get 1/2 credit for 16 deals (or 8 full deals)

So the “ROI” calc on the PPC part would be the $16k Google PPC spend into $80,000 GCI - it’s now a 5-1 return (which makes sense it was originally 10-1 but now it’s halved as it’s sharing).

Now since Youtube and Facebook have different budgets, they would have different ROI numbers.

We spent $5,000 on Youtube and it gets credit for half of 8 deals (so 4 deals at $10k) thus it’s ROI calculation is $5,000 into $40,000 8-1

And we spent $6,000 on Facebook for 4 deals worth (when shared with PPC) and thus is $6k into $40k or 6.66-1 ROI.

But this is not what we would hope for: I mean, why spend money on Youtube or Facebook unless you expected to “increase” your conversion rate from your originally captured leads?

This is what we’re trying to get at in terms of an end result - does adding these extra channels increase our conversion rate (and thus overall ROI)

Imagine instead of the 1.5% conversion, you took it to 3% conversion?

That would mean all the spends are the same in the above example but the actual GCI doubles (so $320,000) and therefore the ROI calculation for each source doubles.

Google PPC would not have one down and yet by adding Facebook and Youtube our effective revenue generated and overall profits would be $160,000 higher.

OR (and this is just as valuable) - if you found you were implementing a channel (let’s say LinkedIn) and it showed that the profits you were “increasing by” were not even covering the spend on LinkedIn - you could make an easy determination to not advertise there.

Would love to hear others thoughts on this - post your comments below

[Morgan]

Incidentally, if you did want to get “slightly” more complex, you could look a the concept of “weighted attribution” whereby you could give more or less credit to a referrer based on where it landed in the chain of events.

So your “first touch” you might say deserves more credit than the subsequent touches so you could assign it 2x or 3x whatever weight you wanted as compared to the last touch. That sort of thing.

And of course if you really wanted to get tricky - you would include a halflife on that bonus so that as a set amount of time went by the bonus weight it go became less.

So let’s say day 1 it’s worth 3x, but per day that multiple drops by .001 - so after quite a bit of time it’s not worth more than the rest of the referrers.

There are lots of cool / fun ways you could play with this - and I’m sure there is a lot of data we could research from the likes of www.hubspot.com or www.salesforce.com on what the best mixes are for such things.

[Pozek]

This will also help us with our ISA department and agents as it relates to warming up leads. Currently we have tens of thousands of leads pseudo-dormant in a lead pool. The goal with retargeting old leads is to get a certain amount to raise their hand so the ISA can reach out to them and connect with an agent. Knowing which leads have been warmed up and reengaged would allow us to spend our focus on those people first vs. “cold calling” through our old leads aimlessly.

[Morgan]

Can you unpack this a bit more?

How would attribution help the ISA’s?

Today’s world “recently active” as a smart list. (Shows who has come back) and then they can click on the lead details and the visits tab tells us where they have come from.

I see how additional attribution helps the marketing team in terms of making decisions with their spend and measuring campaigns - but I’m not clear on how ISA’s would interact with this data.

Curious on your additional thoughts.

[Morgan]

While we’re waiting - just had an R & D meeting and discussed this at a bit more length

Here’s how I am proposing this could be done -

We’d use the word “channels” which would refer to any unique source we have tracking for (keep in mind, not all referrers can be tracked, but most paid marketing channels can)

So you would have “Channels,” - and they would be searchable in the lead filters and could also be displayed in columns on the view manager (similar to how groups are in the screenshot below)

Once we have channels mapped, we need to decide “what else” to do with the data.

So far we have our ISA’s able to see what channels a lead has interacted with (and I’d code it in which order) and also search them.

The next logical step is tying these channels to sources - we keep the primary source (first lead attribution) the way it is, but then add a secondary dimension called “channels” - we would use all the same calculations from the thread to allow you to see linear channel value.

How do you get your budget logged? You can already do that in the source functionality
How do you calculate ROI? Log your deals - the system will take care of the rest.

You would then have 2 metrics - first lead ROI (what we have today) as well as "per channel breakdown)

It’s a cool idea - the challenge really is, how many Realtors (on our roster) are doing enough omni channel marketing to make it worth it? And further to that, are they ready / currently adding their deals to the system? If you don’t log your spend and your deals, then the system would not bring the maximum value.