- Jul 26, 2016
- 2,273
- 3,659
Quick one for anyone doing link building or digital PR.
I used this playbook in the past. It worked, but the development cost made it difficult to repeat.
That part has changed.
Lovable and Claude can now turn a government API into a working map, tracker or checker in an afternoon.
Why does that matter?
A journalist can rewrite your article without linking to it. They cannot replace a tool their readers need to use.
The example that put this back on my radar was a regional UK illness tracker built from UKHSA and NHS data.
It showed what was rising by region, alongside 12 week trends, regional comparisons and a weekly digest.
The team behind it said the first working build took under two hours. A comparable custom build had previously been quoted in the thousands with a timeline of a few months
That is build time, not the complete campaign. You still need to verify the data and find the story.
Proof that the format attracts links
The Rivers Trust built an interactive map using sewage discharge data:
https://theriverstrust.org/sewage-map
Semrush currently estimates the sewage map path has attracted 1.7K backlinks from 212 referring domains.
One useful map. Semrush estimates 1.7K backlinks from 212 referring domains.
Another example is reallymoving’s home affordability map:
https://www.reallymoving.com/property-market-insights-trends/home-affordability-map
Users enter their budget and required bedrooms. The map shows where they have the best chance of finding a suitable property.
The page even provides iframe code for publishers to embed the map and asks for a credit link.
Semrush currently shows 16 backlinks from 10 referring domains for that path.
A smaller example, but already ten separate domains linking to one interactive asset.
I am not claiming either tool was built with AI. They prove that the format earns links. AI simply makes the format much cheaper to reproduce.
The method
It needs fresh data and new angles every week or month. A one time dataset gives you a one time campaign.
The visitor should understand the answer within five seconds.
A starting prompt:
“Build a responsive [TOPIC] tracker using this official API: [LINK]. Show the latest result by [REGION], a colour coded map, 12 week trend and comparison table. Cache the data, display when it was last updated and show missing values as N/A. Include the original source and methodology.”
Expect to correct the first output. Check the geography, calculations and missing values yourself.
“Norovirus in the North West increased 40% this month” is.
Find the three or four journalists who covered the previous version of that story. Send them the relevant number and the tool link.
No mass press release needed.
The tool gives you something new to pitch every time the underlying data changes.
Three things not to screw up
The development barrier is no longer the hard part.
The hard part is choosing a dataset that keeps producing stories.
Worth testing if you have not revisited this since AI app builders became usable.
I used this playbook in the past. It worked, but the development cost made it difficult to repeat.
That part has changed.
Lovable and Claude can now turn a government API into a working map, tracker or checker in an afternoon.
Why does that matter?
A journalist can rewrite your article without linking to it. They cannot replace a tool their readers need to use.
The example that put this back on my radar was a regional UK illness tracker built from UKHSA and NHS data.
It showed what was rising by region, alongside 12 week trends, regional comparisons and a weekly digest.
The team behind it said the first working build took under two hours. A comparable custom build had previously been quoted in the thousands with a timeline of a few months
That is build time, not the complete campaign. You still need to verify the data and find the story.
Proof that the format attracts links
The Rivers Trust built an interactive map using sewage discharge data:
https://theriverstrust.org/sewage-map
Semrush currently estimates the sewage map path has attracted 1.7K backlinks from 212 referring domains.
One useful map. Semrush estimates 1.7K backlinks from 212 referring domains.
Another example is reallymoving’s home affordability map:
https://www.reallymoving.com/property-market-insights-trends/home-affordability-map
Users enter their budget and required bedrooms. The map shows where they have the best chance of finding a suitable property.
The page even provides iframe code for publishers to embed the map and asks for a credit link.
Semrush currently shows 16 backlinks from 10 referring domains for that path.
A smaller example, but already ten separate domains linking to one interactive asset.
I am not claiming either tool was built with AI. They prove that the format earns links. AI simply makes the format much cheaper to reproduce.
The method
- Pick a recurring news beat
It needs fresh data and new angles every week or month. A one time dataset gives you a one time campaign.
- Find the official data underneath it
- UKHSA for health
- Police.uk for crime
- Environment Agency for flooding and river levels
- ONS for wages, population and housing
- Land Registry for property sales
- NOAA, CDC, FEMA and Data.gov for US campaigns
- Build the smallest useful tool
- A map or postcode checker
- The latest result
- A trend chart
- A comparison table
- A visible source and update date
The visitor should understand the answer within five seconds.
A starting prompt:
“Build a responsive [TOPIC] tracker using this official API: [LINK]. Show the latest result by [REGION], a colour coded map, 12 week trend and comparison table. Cache the data, display when it was last updated and show missing values as N/A. Include the original source and methodology.”
Expect to correct the first output. Check the geography, calculations and missing values yourself.
- Pitch the number, not the tool
“Norovirus in the North West increased 40% this month” is.
Find the three or four journalists who covered the previous version of that story. Send them the relevant number and the tool link.
No mass press release needed.
The tool gives you something new to pitch every time the underlying data changes.
Three things not to screw up
- Cache the API response. Do not call the source on every page load.
- Compare rates rather than raw totals when population size matters.
- Never claim daily updates when the source publishes weekly or monthly.
The development barrier is no longer the hard part.
The hard part is choosing a dataset that keeps producing stories.
Worth testing if you have not revisited this since AI app builders became usable.
