Three things we learned from a failed product launch
We spent a few months of time and resources developing our ‘search terms optimiser’. It shows the user the search terms that were matched to broad keywords or phrase matches to keywords on Google Ads and Bing. These are generally much more expensive. An example in the difference in CPC’s below.
In this example below the difference in CPC between exact and phrase is almost 100%.
Then it shows the user how much he can save by adding them as exact and also the button to add them as exact. Sometimes the savings can be in the range of 10% of the entire budget but this depends a lot on how well maintained the account is.
Savings and increased ROAS also comes in the form of increased relevance scores and better Click troughs (CTR’s).
However the product never really took off as a standalone product.
Many of our clients like the feature and continue to use it but as part of the windsor.ai – attribution Insights platform. The product just was not strong enough to live on its own. Here below I try to summarise some of the learnings.
1. Understand the user
This is kind of a no-brainer and everyone knows one should do this. However the depth of understanding is easily mistaken. We saw that our users liked a feature and requested improvements to it and suggested many themselves. So we understood how they use a feature.
However what we did not understand deep enough was how the users buy tools. For success this obviously also has to be understood. Its obvious but was missed, part of the learnings come in learning number 3.
2. Test distribution and sales in parallel with product development
Somehow for an engineer its easy to focus on solving a problem and building a product. Its easy to neglect testing the sales and distribution, especially if you already have some users. Having a few users does not yet mean there would be a scalable sales and distribution channel already in place. This is of course better than nothing but still not quite what is needed.
3. Dont think: ‘It’s different this time’
Of course the first two lessons are lessons that have been learnt again and again. Don’t start developing a product without finding someone who can commit to being a customer. Because the first assumptions about user needs usually turn out to be wrong.
However this time we thought it was different. This was because users were already using a simpler version of the feature as part of the larger product. So we thought there was a clear need for it and users were requesting new features.
Well, turned out the users of the marketing attribution platform use and buy products in a very different way.
Anyways, now the search terms optimiser is integrated into the attribution insights platform and users like and use the feature. We also just added some functionality to it so one can setup automated rules for where the search-terms gets added.
Example of table to add search-terms as exact keyword matches
Rules to automate search-terms to be added to campaigns
Broad keyword: insurance, CPC 4.5
Broad keyword campaign names contain the capital B to symbolise that they are broad.
Search term: Health Insurance, CPC 3.2
Gets added to the exact campaign and adgroup. Exact campaigns and adgroups contain the capital E to symbolize they are exact.
In some cases the exact campaigns can benefit from using keyword insertion rules so that the relevance scores gets up and there are even more savings and better CTR.