How can we weigh post-view conversions in a retargeting campaign?

It's really up to you, but it could be a good idea to discount each view-through conversion by time. For example: within 24 hours - 80%, 48 hours - 60%, 72 hours - 40%, and 96 hours - 20% (again, just an example). Figure out what works for you. Does your vendor give you the ability to place a dynamic conversion dollar value into the code? This could help. We actually just put an article up on view-through conversions: http://www.retargeter.com/strategy-2/view-through-conversions-don%E2%80%99t-discount-the-%E2%80%98billboard-effect%E2%80%99 Perhaps this can clear up some of your questions or hesitations about view-through attribution.

Keith's warnings about low-cost CPA campaigns are valid, and combined with Greg's very valid suggestion to do A/B testing, you can probably weed out most of the stuff that isn't actually adding incremental sales to your bottom line. I think Greg's estimate of 3-6% is a bit low, at least based on our experience, but a lot depends on how your remarketing campaign is set up, whether you are using product-level, dynamic ads, and your vendor is doing a good job setting frequency caps and optimizing the campaign to account for recency of site visit and other factors.  All that said, run valid tests where you're only changing one variable at a time, and test one vendor at a time, don't bid against yourself and complicate your channels.

Here's how I typically tackle this problem: Create a new conversion that only attributes a conversion within a 1-2 day window of someone seeing the ad.  It won't be perfect, but it can help get you credit for the people that don't click on ads. Use viewthrough conversion data as a directional data point that lets you AB test the effectiveness of different audience segments or different contextual segments.  So, seeing 100 viewthrough conversions won't necessarily tell you that your ad drove 100 additional conversions, but if you see 75 from whitelist A and 25 from whitelist B, then whitelist A is better for your ad. This also assume that the offer on your ad is the same offer that people have seen on your website.  If the offer is different or if the offer is for a different company, then I usually attribute most of the viewthrough conversions to the person seeing the ad.

Without investment in the proper tools to statistically analyze your media mix and its impact on view attribution, there are some simple techniques you can use. First and foremost, make sure you are managing your buys on equal footing: all of those bonus links, tiny bonus ads and other freebies for CPA campaigns are designed to bomb view cookies out there and snag as many view conversions from others on the plan as possible. I would pass on the free stuff and go for a lower CPM or no view conversion on the free stuff. Now that everyone is on equal footing, Set your campaign view window to something you think is reasonable liberal for your product's conversion cycle: software download might be 30 days whereas auto leads might be 120 days. Run your campaign. As your campaign concludes, analyze your data: look at the date of ad impression and the resulting date of view conversion. You should be able to determine the appropriate window where most of your view conversions happen. If 80% complete within 45 days, set that as your window. This approach looks at each buy and media channel independently. A more accurate method would be to analyze how search, display and other tactics ultimately should be weighted in addition to how wqide open your window should be. But this would require another layer of analysis, tracking and sophistication. Whatever method you employ, make sure you share it with your media vendor to ensure they are optimizing on the same set of assumptions for maximum performance. Final note: this method should be used across all tactics, not just retargeting.

I'm not sure whether this applies to retargeting but what we do to address similar concerns is to buy publisher inventory for a week prior to any campaign and advertise on a charities behalf, sending traffic to their site but importantly, measuring impression view exposure off site AND then post view landing/conversion rates on our site. This gives us a control view through rate sans our advertising. Then when we run our actual campaign the following week we continue to track the same exposure and view through, measuring the uplift in conversion and using that as the true percentage of post view conversion attributable to the campaign. Obviously this can cost money so we don't do it every campaign but we do it regularly enough to maintain a general benchmark to be more accurate in ROI reporting. Measuring real and total dedupolication is very hard without proper tools so this is a way of doing the due diligence on your ad performance management responsibilities, crucial to ensuring actual incremental sales over 'tracked' sales.

The answer needs to be tested or an attribution platform should be used to come to a reasonable answer. However, I would say 20% would be a reasonable percentage.