Les Binet Profile picture
Jan 6 9 tweets 3 min read
Why are Meta & Google moving away from digital attribution as a way of measuring marketing effectiveness?
Data issues are part of the story, but it’s bigger than that. This Google report gives some insight into their thinking. 1/9
thinkwithgoogle.com/intl/en-gb/fut…
#marketing
#advertising
The big problem is that digital attribution models struggle to measure “incrementality”. If 1m people see your ad, and 1k buy your product, how many sales are genuinely incremental? The answer might be anywhere between 0 and 1k. Attribution models on their own can’t tell. 2/9
As Google pointed out over 5 years ago, controlled experiments and rigorous econometrics are much better for measuring incrementality. And those techniques reveal that digital attribution models are often WILDLY inaccurate. 3/9
New analysis from Facebook confirms what many econometricians have been saying for years: attribution models overestimate the relative effectiveness of direct response channels, and underestimate the effect of brand building media. 4/9
Digital attribution models are poor at measuring offline media, for obvious reasons. This matters A LOT. Meta’s own econometrics shows that TV is still the medium generating the most incremental sales. Yet attribution models struggle to measure the effects of media like TV. 5/9
Another reason why digital attribution gets ROI wrong is that it’s short term. Meta’s new research confirms that ad effects can last for months or years, and most of the payback comes from these long-term effects. Attribution models rarely look beyond a few days or weeks. 6/9
Google’s report identifies short-termism & the failure to measure incrementality as challenges that must be solved if we are to measure effectiveness properly. Google suggest a way forward: supplement digital attribution with experiments & econometrics to “triangulate” ROI. 7/9
And that’s the direction that both Google and Meta have been moving in for several years now. It’s just taken a while for everyone else to catch up. 8/9
Matt Taylor, lead author of the Google paper is a smart cookie. Here he is at the IPA’s 2017 Eff Week conference, talking about some of these issues. 9/9

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Les Binet

Les Binet Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @BinetLes

Dec 12, 2022
New research from Meta highlights the importance of long term advertising effects. Econometric analysis of over 3,500 campaigns reveals that 60% of the payback comes from long term effects, and only 40% from the short term.

#brand
#advertising
#marketing
#econometrics
This means that short term measurement measurement systems (which include most digital attribution models) probably underestimate ROI by a factor of ~2.5.
Short term measurement approaches tend to favour direct response channels, but this study shows that brand-building channels actually generate more sales in the long term.
Read 6 tweets
Dec 9, 2022
Both Facebook and Google tell me that they are pivoting away from attribution modelling towards econometrics for evaluating effectiveness. Potentially, this has big implications.
Attribution models over-estimate the short-term, direct effects of marketing. So they tend to favour performance marketing, direct response, promotions, etc.
Econometrics is better able to measure medium to long term effects, and is able to pick up the indirect effects that attribution can’t measure at all. So econometrics is less skewed towards performance, and reveals the importance of brand activity.
Read 4 tweets
Apr 30, 2022
Peter Field & I recommend that marketers use a mix of “brand building” & “activation” comms. Our research suggests the optimum split is usually ~60% brand, ~40% activation, but varies by context.

But why split brand & activation at all? Why not do both at once?

1/16
Definitions first.

By “brand building” we mean building & maintaining memory structures that have a lasting influence on consumer behaviour.

By “activation” we mean activating existing memory structures to elicit immediate behavioural responses.

2/16
For example, a travel insurance firm selling to young people might sponsor a festival. If they do it well & get good exposure, they slightly increase the chance that youngsters think of them (& think well of them) when they need travel insurance. That’s brand building.

3/16
Read 16 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us on Twitter!

:(