Any good and advanced custom attribution model?

[Tormar]

Does anyone of you has some good and advanced custom attribution model for sharing or some advanced use case?

I have over 1000 campaigns per month from all sources and I don't know how optimise them?

Most of my adwords campaigns have impressions share close to 100%.

[phubui]

I actually haven't heard of that, but it might be referring to how conversions are counted after attribution modeling is applied? Example (with totally made-up numbers):

AdWords has 100 conversions when using a last-click model. When applying a first-click model, AdWords only has 25 "weighted conversions".

That's just a guess though. I also can't find any additional information on it. Do you have a source where you saw that term that I can read? That'd help.

[Tormar]

Thank you. I should ask you one more question, I can't find it.

Do you know meaning of "weighted number of conversions" in attribution models?

How analytics calculate it? Do you have any example?

[phubui]

A lot of people don't like last click because it gives all of the credit for a conversion to the most recent interaction before the conversion happened. While there's some merit to that, it completely ignores the many other touchpoints a customer interacted with before they converted. For example, a customer could visit your site, see an ad on Facebook, receive an e-mail, search for your brand on Google and click an ad there, and then convert. With last click, paid search would get 100% conversion credit in that case. This link helps explain the different models:

https://support.google.com/analytics/answer/1662518?hl=en

I may be misunderstanding the second question, so let me know if I'm off base on this next response. If you're seeing that you're getting a lot of conversions off of impressions (viewthrough conversions) over click conversions, it could be worth investing a bit more in those campaigns. You'd really need to look at your attribution model comparison to see how much those impression-based conversions are really impacting things. A lot of digital marketers prefer click conversions as they're easier to defend as legit than viewthrough conversions, so be a bit wary about those, but I wouldn't ignore them completely.

When comparing models, I'd start at a high level and drill down as necessary, so check things out by the MCF Channel Grouping dimension first.

Are you asking about the Last Non-Direct Click model? I believe direct traffic is excluded there so people can get a better idea of what channels typically lead people to the site before converting, but I'm not sure.

This might be some good reading for you if you want to learn more about attribution and pros and cons for the various models:

http://www.kaushik.net/avinash/multi-channel-attribution-modeling-good-bad-ugly-models/#lastnondirect

[Tormar]

Yes, I've been using model comparison tool. But I have few questions.

When I compare models, last click is always the best in my case, but recommendations are completely different. Should I ignore it?

If I've seen that Impressions are better then click like interaction type, should I create (or increase ratio) impressions campaign?

Also, when you compere some models, do you usually test models on type of traffic, or you compere some campaigns, or goals?

Also, if I exclude direct traffic from last step, I would automatically have better results from other sources, but in reality you would have the same ratio of direct traffic because you can't directly increase or decrease it. I don't understand that.

[phubui]

Hi Tormar,

Are you using GA? If so, have you used the model comparison tool? That might give you some ideas for what model makes the most sense for your needs. If that's not quite enough, I've heard good things about Interstate Analytics, which is a pretty low cost tool that you can use for attribution as well.

Hope that's helpful!

Phu