Use Bayesian Reasoning When Statistical Significance is Lacking.
Let’s say you are trying to optimise a campaign for demographic information/bid adjustments. The campaign is based geographically in New Jersey and is advertising soccer boots.
You’re campaign has only been running for 2 months, so you only have a few conversions in regards to each age group.
To gain further insights – Get the overall age and cost/conv for the entire account.
If possible get the age and cost/conv data for any geographically similar campaigns. E.g. any campaigns in New Jersey
Gain the age and cost/conv data for all soccer boots campaigns, and any similar product campaigns.
Now, if every campaign plus the account shows that the age group 65+ is the most expensive, and so does the New Jersey Soccer Boots campaign, it’s a safe bet that you can reduce the bids for this age group for the New Jersey Soccer Boots campaign, even if your data for this specific campaign is lacking and only includes a couple of conversions.
Use a bit of guess-work and common sense if there is a mis-match between the age group and trend for cost/conv between the overall account and other campaign data.
2. Set KW Bids below First Page Bid if Campaign is Running Too Expensive per Conversion
Then increase bids on specific demographics and devices – the ones which provide the best cost/conv.
You can then, to a certain extent, dictate which specific users (in terms of age, gender, income, device etc) see your ads.