5 ms·
This was not what I was expecting. They're just using a small ad spend to measure CTR on the proposed article headlines/stubs instead of using some kind of ML
by mdorazio 10y ago
This was not what I was expecting. They're just using a small ad spend to measure CTR on the proposed article headlines/stubs instead of using some kind of ML to predict success like I originally thought. The approach is also not very cheap. At $40 in ad spend and at least a day to get the results, you could have just paid someone to write one or more blogspam articles, which is what this will optimize for.
- gk1 10y ago> The approach is also not very cheap. At $40 in ad spend and at least a day to get the results, you could have just paid someone to write one or more blogspam articles, which is what this will optimize for. I disagree. If you're one of the (few) companies that spend a significant amount of time planning and writing articles, then $40 is a very low amount to pay to learn which angle/theme/lede will get the best results, before your team spends time planning, researching, writing, and editing.
- sporkologist 10y agoSo I guess Priceonomics has to prove that their predictive ability is better than their client, who one would assume is already very familiar with their target market. That is a tough sell.
- brianwawok 10y agoAnd where does the click go? A page "Thanks for voting on this title!" If you do that for a few days, people are going to stop clicking on your clickbait titles with no content. I totally get rotating headlines to see which gets more clicks. I do not get rotating titles before the article is written.. that is going to make users quit clicking on your ads very quick ;)
- whatshisface 10y agoHow will they know which adds are yours? The CMS is off-brand and priceonomics says they will host it. If this tragedy of the commons really kicks in, you can bet advertisers will start calling for FCC intervention...
- ssharp 10y agoThat's a really cynical angle to look at this from. I'd say most places putting out good content are investing far more than $40 on each piece.
- mdorazio 10y agoThat depends entirely on how you define "good content" and who you think is creating it. The NYT interactive articles, for example, are obviously time consuming and expensive to create, but your average company using blog posts to boost search rank is usually paying ghost writers a small amount to crank out articles on a schedule, based on topics relevant to the brand's marketing. Personally, I think you're much better off doing some keyword research to find what people are interested in and creating articles based on that instead of trying to back into a headline that fits your intended content.
- swanson 10y agoJust a point to note: Pricenomics also recommends writing well-researched, data-driven content -- specifically spending up to 40 hours per blog post. I think the intention is that if you are willing to spend a week or two of effort on writing content, you could do some upfront validation to try to avoid duds.