Persistence, Expiry and Attribution – of eVars and events


I am rectifying a long-standing mistake, today, finally translating an article that I had posted absolutely ages ago on the German blog.

It’s about a real-life example that I often use to explain how eVars work. Since eVars are such an integral part of Analytics, a good analogy is important!

Unfortunately, I do not have a good analogy.

But I do have an over-the-top ridiculous one for you!

I believe that it being ridiculous makes it more memorable. And, I simply like going the whole 9 yards, even if it takes me into lala-land.

The example was passed onto me by a colleague early in my time at Omniture. I can for the life of me not remember who told it, but if you think it was you, let me know. Attribution where attribution is due.

Shopping with eVars

Right. Here we go!

Say I am the manager of a local store, part of a bigger chain, maybe. My store is a bit strange, in that my sales of any fresh produce are way below my peers.

So they make fun of me, call my customers unhealthy slobs, and it bothers me.

I sit down with my marketing manager, and she says she can help.

Week 1, she gives everyone who spends more than CHF 10 an apple.

Outside the store, she puts a sign telling people about the offer (“Spend CHF 10 and get a free apple!”), the point being to get people into the shop.

At the end of the week, we check our system: we gave away a lot of apples, but we also saw an increase in produce sales, overall. Good!

My marketing manager tells me that she can even tell me whether the sign was helpful. I’m impressed. How did she do it?

She hired a student and gave them a rubber stamp. The student stood outside the shop, next to the sign, and they stamped the forehead of every person who looked at the sign.

At the checkout, our clerks pushed a little button if someone had a stamp, and so now we can look at our system and separate out those sales we made with people who had a stamp.

We are comparing what part of the overall average customer’s shopping was produce, against how the people with a stamp on their forehead behaved.

Wow! They comparatively bought way more produce!

Seems like the sign really worked!

Jan? Seriously?

I told you the example was ridiculous. Nobody would stamp potential customers on their foreheads! You’d lose your job faster than you could say “but …”!

Some places can do it. Think about clubs! Or places that use little arm bands, cards, or lanyards, like big events.

On the web though, stamping is exactly what we do!

The forehead is the equivalent of an eVar. The checkout is the equivalent of an event.

Each visitor has up to 250 foreheads, in a big table that the system manages for each visitor ID.

There can also be a lot of different points at which those foreheads (and the stamps currently on them) matter: those are the events, and you can have 1000 of those.

Each time one of these events “happens”, the system takes into account all the foreheads (eVars) of the current visitor.

Think putting another student next to the produce aisle. They could increment a little clicker each time a customer looks at any of the produce. We could maybe call that a “Produce View”. At the end of the day, we would have two numbers: overall Produce Views, and Produce Views by people with stamp on forehead.

Would that help?

Sure would!

If a lot of people come to look, but then do not buy, maybe something is wrong in my shop? Could be the presentation, or even the products themselves!

I will have to go visit some other stores to compare! Maybe I can learn something…

Expiry & Attribution

Let’s take this further.

The pressing of the button at checkout works because I had assumed that the stamp hadn’t been rubbed off. That is called “persistence”.

My marketing manager is a pro, she used ink that would not immediately come off. In fact, the ink would stay for about 30 days if people would not make any special effort to get it off their foreheads.

With persistence comes what we call the “Expiry”: how long would a value, written into the eVar, stay there before it faded away.

In Adobe Analytics, you can set that to a lot of different things, like days, weeks, months, years, or until the end of a visit, or until some event happens, maybe a purchase.

You can also set an eVar to never expire, the equivalent of setting up a little tattoo parlour next to your sign. Probably wouldn’t do that in real life, but you can on the web.

Once the expiry happens, the eVar is empty again, the forehead is clean.

We used the “30 days ink”, right? So what happens the following week? My marketing manager has put a new sign up, simply reminding people to “eat 5 a day”, and the student is there with a different stamp (so we can see which sign someone saw).

If a person comes to the shop, they get a stamp, as before, but what happens if they already have last week’s stamp?

Well, we have a choice: we can replace the stamp with a bit of rubbing alcohol or soap, or we can put a second one next to the first, or we can leave the person alone with their first stamp.

In Analytics, we call this “Attribution”. Those choices above are “last touch”, “linear”, and “first touch”, and depending on which we choose, we later get different results.

I am 100% sure that we could take this even further. I’m thinking about having the student watch how people come to my market, and call that “Market Channels” or something like that.

But for now, we’ll leave it there. Persistence, expiry, and attribution, explained through a rubber stamp and some ink.

Simple and beautiful.


What about a stamp that has no persistence? Well, that’s a prop. Or, technically, it could be an eVar that is set to expire after the hit.

The equivalent of clearing all your cookies, BTW., is to don a new skin. Snakes do it all the time, which, technically, makes the normal analytics person an online snake.

Shops may want to use finger-printing, of course. That could be literally taking finger-prints, but it could be more subtle, like making sure that the student is always the same, and that they are good at remembering people.

If you live in a small place, or you frequent a small shop, all the persistence, expiry, and attribution is likely happening in the head of the shop keeper. Chances are that they have their expiry set to “never”, too.

That is one reason why living in a city feels so anonymous. For people in web analytics, I think it translates to some vague, cosy feeling of warmth… bliss in the knowledge that no one will recognize you, or remember what you do, or judge you because of it.

The web, in that light, is like a big city, or it can be. Don’t want one shop to know how many MTB tyres I buy? Just buy them from different places, no problem!

And don’t forget to delete your cookies!

I sincerely hope that some of you will flinch and look around for a person with a stamp next time you see a sign in front of a shop!

Or maybe think about how spoiled we are in digital analytics, because we can do these things, and we won’t be hit in the face for it.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.