• Richard Fergie

    “If time lag reports show that 98% of visitors will convert within 5 days, why set your remarketing cookie to target people on day 6?”

    This tells me that by delaying your remarketing for 5 days (i.e. don’t show adverts for 5 days, then hit them) then any results from the campaign will be truely incremental rather than canibalising sales that would have happened anyway.

  • matthopson

    Another great way to use the path length reports is to create a custom brand v non brand channel grouping so that you are able to see what part each type of query plays during the conversion path.

    You may find quite a few people make first contact via a more generic research type query before going on to finally convert against a brand KW.

  • Harris Neifield

    Great insight Richard. I agree – remarketing can cannibalize visitors who would’ve returned to your website anyhow. However, there are visitors who will only return to convert if prompted by an ad so if an advertiser doesn’t show ads during the first five days (the likely to convert time period) there is a likely to convert segment they’d miss.

    Delayed remarketing is a top tactic we’ve discussed in prior posts, but mostly to cross sell different products in different seasons. My concern with remarketing on day six was that if conversion rates are very low, CPA’s could be very high. Even though it might be almost entirely incremental improvement at day six (no cannibalization) it almost doesn’t matter if the CPA is very high.

    If the cookie pool is large enough, one could set up two custom combinations – one targeting the first five days, the second targeting day six and beyond. Then one could bid differently in each ad group.

  • Harris Neifield

    Thanks Matt. Custom channel groupings can definitely used to find good intelligence, like the brand v non-brand query progression towards conversion.

  • Richard Fergie

    “there are visitors who will only return to convert if prompted by an ad”
    I 100% agree with this. The difficulty is in tracking this and measuring the incremental uplift.

    “If the cookie pool is large enough, one could set up two custom combinations – one targeting the first five days, the second targeting day six and beyond.”
    This is the best approach in my opinion. If the 5 day audience is big enough then the 6+ day audience will be too because everyone from the 5 day audience ends up in the 6+ day audience eventually.

    I guess we must have had quite different experiences with this in the past. When I’ve used short term remarketing the results look great at first glance but once testing on incremental uplift is done they don’t look so great. Whereas everytime I’ve used a short delay results have been great.

  • Harris Neifield

    Great stuff! I agree the audience for six or more days will become much larger than the first five day audience. However if we’re bidding a lot more in the first five day audience ad group because of the much higher conversion rate, hopefully AdWords will show the first five day ad group more and the difference in the number of impressions for the two ad groups won’t be that substantial.

    You said it perfectly – measuring the incremental uplift is the difficult part. If an agency or the company is making other optimizations to their website and or advertising campaigns and launches remarketing at the same time, what caused the improvement in performance – remarketing or other factors? Remarketing could be just paying for conversions you would’ve gotten in lag anyhow so you’re nominal numbers in AdWords for remarketing could be misleading.

    Measuring the true incremental lift from remarketing is difficult.

  • Richard Fergie

    You can exclude the 5 day audience from the 6+ day audience using custom combinations so you don’t need to worry about these two audiences getting mixed up.

    I measure incremental uplift by running some adverts for a charity against the same audience. The difference between the number of view through conversions on the charity ads and the total (actual+view through) on the genuine ads is the uplift.

    Sometimes hard to get clients to allow budget for the charity ads and I’m sure Google doesn’t serve the ads randomly but this is the best I can think of at the moment.

  • Harris Neifield

    I’m not so much worried about getting the pools mixed up, you’re correct we can exclude via custom combinations (see my post here). If we’ve got a 30 day cookie pool, the day 1-5 cookie pool has 5 days of visitors and day 6-30 has 25 days of site visitors which is why the imbalance is there and is my concern.

    Charity ads would be a good test but you’re right budget is a concern. Anyhow, I’ll do some more thinking on measuring true remarketing lift as well. Thanks!

  • Tom

    Great post, mastering Google Analytics is very important. Thanks for sharing.