Also re AXA - I find it really shocking that a company with so much cash clearly tried to do this campaign on the cheap.
I’m on the way home so this is a tit scruffy write up sorry, but I once spent 3 months on a project for a bank (I literally hated this job so much but that’s another thing) which was commissioned to inform their brand comms/social/CRM/events work and it sounds like AXA really need this (call me huns xo) but the level of consideration that went into this project included:
* twitter data - you can create segments of users based upon certain criteria, so for example you’d query twitter bios to see if someone’s a parent or where they’re based etc and add them to your “target list”, then you’d write queries to scrape what they’ve said about... charity or donating, what links they’ve shared, etc. Or export all tweets and do some smarter language / theme / sentiment analysis on it. There’s a very manual element here as sentiment APIs are largely trash and don’t grasp swearing that well or local venecular or any slang at all. Loads of stuff on who they are following & clusters of interest where those people are following each other which can be quite good for themes or publishers, and brilliant for short listing a target list of influencers for PR to vet (which clearly didn’t happen with FOD). You can even see what time of day they’re tweeting/location/whatever. There’s a lot more than this and if u look at the documentation here https://developer.twitter.com/en/docs/tweets/data-dictionary/overview/intro-to-tweet-json you can see how twitter offers up ur tweets for the scrapinganything in this documentation can be quantified & analysed down to the individual. They recently capped it at around 3.5k tweets but there are workarounds for that.
* forum data - less easy & longer form content harder to do automated analysis on, but all of these pages can be scraped and pretty much the same as above can be done, you just need to see how the page has been marked up and write a script w selenium
* Instagram data - google image API can determine brands shared & type of place (landscape, faces, whatever)
* comscore data - people install a plugin on their browser and are compensated for selling their web browsing data basically (yes I used to look at the porn web page titles and lol it was my only respite) so it skews lower income but you can look at users who’ve used compare the market for insurance say and then see what else they’ve been looking at, or see how that journey goes for them? Did one for a £££ furniture store who saw that people use their site for inspo then go and find dupes on cheaper sites
* xAd data - this is tit but basically some apps have your location data and sell it on, it’s tit cos the app needs to be open at the time to capture your location so is patchy af. But say you wanna see where else your Barclays customer goes you could then see they also love a zizzis and a Homebase etc
* actual interview data with your target market but this is £££ and long af
I’ve got a neater write up on my CV looool under my ~ projects I’ll see if that’s non doxxing to share that.
But as much as influencers are trash, a brand like AXA has access to resources to help them create informed & good content and yet they choose not to
All of this. They also included Parti time Working Mummy who received loads of "you're amazing" comments on her ad. This is the woman who set up a monthly paypal donation account (to her personal paypal) to aid her work with DV victims. She still hasn't told anyone what she did with the money. AXA has a lot to answer for.
yes please - need some more FOD and less PTWM 
I'm here all week.....