Subscribe
Sign in
p/experimental-methodRepeatable Business Experiments 💌 -> ⚗️ -> 📈
Start new thread
Dan Pantelo

Marpipe 1.0 - The first multivariate testing platform for ad creative

by•

Testing your ad creative is the most important thing for performance, but actually doing it sucks. Marpipe automates the entire process.
From startups to enterprises, use your assets to build/launch/analyze perfect multivariate tests & discover ads that win.

Add a comment

Replies

Best
Dan Pantelo
Maker
Hunters, I’ll be honest with you - we had an awesome creative team but no idea what we were doing when making ads. We kinda just made stuff that looked cool, and if everyone else agreed it looked good we would launch it. 🤷‍♂️ Sound familiar? Thing is, creative is (by far) the most important element to ad success. Statistically, most ads we launch don't work well. So why are we still totally shooting in the dark with NO efficient processes? Why does it feel like data science/automation has benefited every part of the marketing process EXCEPT for creative? Marpipe is 2 years in the making. We originally built it to solve our own problem as an agency, but after showing it to folks at Adobe, MediaMath, and others, we raised enough money to spend the past year building it so we can deliver it to you as a simple freemium platform (right in time for Q4!!); ✅ Using just a few creative assets, build hundreds of creatives in a multivariate grid. ✅ Launch them all to your desired level of spend (works for small OR big spenders). ✅ Find positive-outlier ads to scale. 🎉 Discover WHY your ads work at the asset-level so you can build better creative in the future. 🎉 Never make creative guesses again & stop getting crushed by creative fatigue Now trusted by creatives & marketers at companies like Segment, Tubi, and Mars, we hope you’ll take it for a spin as well and let us know what you think. Happy testing!
Dan Pantelo
Maker
@phattanle thanks Phat!
wata. hia
How would i connect all these to ad account?
Dan Pantelo
Maker
@wata_hia Super simple! Once you make an account, go to Settings and click 'Connect Facebook.' From there, all the ad accounts, pages, and IG accounts you have access to will be able to be used for launching your creative tests
Pierce Porterfield
Maker
@wata_hia Great question! Once you connect Facebook to Marpipe, you are able to create separate accounts for each brand you manage. Each account can then be linked to it's respective FB ad account, page, and instagram for launching!
wata. hia
@pierce_engineer How would the launching process be like?
Pierce Porterfield
Maker
@wata_hia It's really easy, all in all it only takes about ten minutes to get hundreds of ads live in Facebook Ads Manager. This video shows the whole process:
Dan Pantelo
Maker
@pierce_engineer @wata_hia let us know if you'd like to demo the product with us soon! Always happy to walk folks through it.
Shani Majer
Looks very cool. How is this different than responsive ad campaigns?
Pierce Porterfield
Maker
@shani_majer Thanks Shani! Good question - Marpipe does not try to auto-optimize your campaigns. Instead it provides you with the tools to build, launch, and analyze creative tests at a much larger scale than what has previously been attainable. When you create an experiment in Marpipe you may end up with potentially hundreds of creative variants. Each creative variant is incrementally different from the next. Once you automatically launch the experiment from the platform, you start to collect data on each isolated component (the benefit of MVT experimental design). Not only is it easier to find winners when you cast a wider net - but you also collect valuable insight that build upon itself.
Shani Majer
@pierce_engineer Thank you for the in-depth response. Very interesting stuff and curious to learn more!
Dan Pantelo
Maker
@pierce_engineer @shani_majer I'd be eager to demo you whenever you're ready to check it out!
Dan Pantelo
Maker
@pierce_engineer @shani_majer awesome just ping me and let me know when you're free to jump on a demo :)
Andrew Gluck
Ability to create many different variants and launch them easily
Dan Pantelo
Maker
@irrvrntvc Good TL;DR!
Dan Pantelo
Maker
@irrvrntvc @pascal_wagner same!! Could've just made that the first comment and save everyone some reading time lol.
Bgf419
I've watched this video like 14 times on repeat, absolute fire 🔥 Quick question: is there a list of use cases for multivariate testing?
Pierce Porterfield
Maker
@brett_friedman1 Good question - Multivariate testing is useful for anybody who is spending money to run ads. It allows you to uncover insights in much higher resolution compared to A/B testing. Many of our customers use multivariate methodology to open up Facebook and Instagram as viable channel. Others use MVT as a focus group - figuring out how consumers will respond to different messages, positioning, and visualizations.
Bgf419
@pierce_engineer ohh so can I do audience testing with this? like run product-market fit tests basically?
Pierce Porterfield
Maker
@brett_friedman1 Yes, when you're launching your experiment you can choose to target multiple of your saved audiences. Once you start collecting data our analytics will show you how different audiences react to the various elements of your experiment. We usually see big differences in audience preference!
Dan Pantelo
Maker
@pierce_engineer @brett_friedman1 let us know if you want to walk through a demo with us!
Dan Pantelo
Maker
@pierce_engineer @brett_friedman1 I'm going to reach out and schedule! what's a good email to reach ya at?
Andrew Gluck
How do you ensure statistical significance for these creative tests? Don't you need that to figure out winners?
Dan Pantelo
Maker
@andrew_gluck1 Good question! The amount you should spend on a test to get significant results depends on two factors: the # of ads in your test, and existing average cost for your KPI (or, Cost Per Result). Based on hundreds of tests that we've seen over time, we recommend that you spend 75% of your average CPR per ad. SO, Example: Your average CPR is $10 in your ad account. We recommend you spend $7.50 per ad in an experiment. So you can run 100 ads for just $750. We've seen time and again that this is enough spend to accumulate more than enough results to yield clear performance outliers and make decisions that'll hold at scale. That being said, I've also seen that most marketers have their own way of thinking about what they consider 'significant' or 'enough data', and a lot of folks will make decisions well before stat sig is achieved in any sense, which is why we allow you to pre-set your own spend when launching experiments (in case you don't want to follow our reccs)!
Olivia Wu
Congrats on the launch. Looks incredible!
James Goldman
I haven't seen anything like this! It seems like the data you gather will be super powerful. But rendering so many ads sounds hard, is there a limit to how many of these "variants" you can create? I imagine even rendering a few hundred could take several minutes if this is all browser based. Regardless, congrats, and I'm still going to try it out.
Pierce Porterfield
Maker
@james_goldman Thanks James, we haven't seen anything like this either! We actually allow up to 1500 creative variants to be rendered for any given experiment. Even when maxing out your creative variants limit our rendering process will only take 1-2 minutes max. It's pretty overpowered if I'm being honest :P
James Goldman
@pierce_engineer that sounds awesome! I played around with it and it is even snappier than you described, don't sell yourself short!
Kendrick Nguyen
Marpipe to a marketing team is like Gatorade to the Florida Gators. Love!
Pierce Porterfield
Maker
@kendrick_nguyen Haha I have yet to hear that one but yes essentially!
Bogdan Ionita
Just wow, it could help us save weeks in display a/b testing and should put all designers on the same page.
Dan Pantelo
Maker
@bogdan_ionita incredible, can't wait for you to get into it Bogdan - let me know if you'd like a demo!
Bogdan Ionita
@danpantelo Marketing team is giving it a look, let's see what they come up with. Keep in touch!
Dean Shapero
Love the product, we're about to use it for one of our campaign launches. Keep crushing it!
Dan Pantelo
Maker
@dean_shapero Incredible! Thanks Dean - so excited for you guys to discover those winning ads for Loginhood. Ping me if you have any questions along the way.
adam tropp
Wow, this sounds super cool. Can I view my ad performance broken down by subcategories? I've always dreamed of having a convenient way of seeing results per audience, out-of-image-copy, etc. Would be a big boost to my creative intelligence gathering process if I could do that in a less cumbersome way...
Pierce Porterfield
Maker
@glsshsthrwngstn Yes, we give users the ability to view their performance results: In-image-copy, out-of-image copy, images, graphics, backgrounds, audiences, etc
adam tropp
@pierce_engineer WOWIIIIIIIEEEEEEEEE! That is great news. I will definitely have to check out Marpipe!
Sachit Gupta
There's moments as a marketer when you look at a product and go, whoa this is going to the change the game and create a new category. That's exactly how I felt when @danpantelo showed me Marpipe for the first time! Congrats on the launch - this is going to go into the toolbox of every marketer!
Dan Pantelo
Maker
@sachitgupta Sachit - an endorsement like this from you is so particularly special to us and our team. Thank you for the incredible words here.
Shannon S
Not quite the first fyi, https://adoptics.com. Nice looking product though.
Dan Pantelo
Maker
@sprfrkr Thanks Shannon! Adoptics is awesome and we're fans of their work in the space. But Adoptics is an A/B testing platform, not multivariate testing. When I was running an agency in my past life, I found that A/B testing would only tell me which ad was the best, but not WHY. With multivariate testing, which is much more annoying to do, we can actually understand WHY our best ads work well. Would you be open to getting a personal demo of Marpipe from me to see the difference?
Shannon S
@danpantelo great insight! Though, multivariate with ads is tough to reach statistical significance of 90%+, anything less is not a data driven decision. Still, it looks like an amazing app!
Dan Pantelo
Maker
@sprfrkr totally agreed! That's why we let users set their desired level of statistical significance in the launch settings, but we can also test a surprisingly large amount of variables for cheap. For instance, we usually hit 90%+ if we spend 75% of the avg CPA per ad. SO, if your avg CPA is $10, then you should spend $7.5 per ad in a test, so you can test 100 ads for just $750 in ad spend! I would really love to show you how it works myself if you're open to it.
Andrew Przybylski
Couldn't be more proud to see this launch. You and the team have worked so hard. Really great to see this in creators hands and what they make of it 🔥
Dan Pantelo
Maker
@andrewprzy1 thanks so much Andrew - couldn't have done it without your help along the way, seriously..
Sam Russo
This idea is awesome, excited to try it out!
Dan Pantelo
Maker
@sam_russo Thanks Sam, I'll see you there!
Color
Great team and product! Congratulations on the launch!
Pierce Porterfield
Maker
@colorinthesnow Appreciate the love!
Scott Friedman
It's about time someone figured how to maximize marketing campaign spends and find a way to leverage AI to ensure ads are presented to its audience in a way that resonates the most. Love it!
Pierce Porterfield
Maker
@scottfriedman6 Theres an advertising renaissance on the horizon
Wills Hapworth
This is a no brainer for anyone who values evidence and data and truth to drive decision making. This is about the difference between 'I think it looks good and will work' and 'I know it works'. this is the new standard in developing creative that drives real results across ROI that matters: sales, engagement, other conversion metrics, etc. stop guessing and winging it!
Pierce Porterfield
Maker
@wills_hapworth1 Couldn't have said it better myself, thanks Will!
Andrew Przybylski
@danpantelo Techstars is partnered with some large corporates, do you have an enterprise model or is that on the roadmap?
Dan Pantelo
Maker
@andrew_przybylski2 absolutely! We're already working with large enterprises like Segment, Tubi, Mars, and PepsiCo - I'd love to connect with you to see how we can bring enable the Techstars portfolio to achieve breakthrough performance.