The Problem of EO Data Pricing

(first published in the UK Remote Sensing and Photogrammetry Society (RSPSoc) "Sensed" journal, April 2024: https://www.rspsoc.org.uk/index.php/publications/newsletter.html)

Going back to when I started working in Earth observation at the end of the 90s, if you wanted to understand data or imagery pricing, things were pretty straightforward. If you could get by with archive Landsat data at $500 a scene then that’s probably what you would go for. If you needed a bit more ground resolution, you would probably fork out for some SPOT data (which I think back then was a couple of thousand dollars a scene.) Pricing was relatively simple (buying was a bit more of a process, but that isn’t the topic of the day!)

Then, more commercial satellites started to be launched and it got a little more complex. One of the problems was selling “by the scene”. When there were just a couple of options, we knew (and could remember) that a Landsat scene was 184km x 184km, and a SPOT scene was 60km x 60km. The scene size being a function of sensor swath. But with more satellites being launched, and more commercial options, the “scene size” became something more arbitrary. There needed to be a common denominator, and so pricing became more based on $ per km2.  SAR data made things even more complicated as scene size varies depending on imaging mode with higher resolution modes being priced higher.

From what I remember (and I stand to be corrected) the first time I saw standard data pricing in km2 was the IKONOS satellite. The satellite was launched in 1999, becoming operational in 2000, and from that point on, high resolution imaging satellites were (mostly) priced in km2.

Using km2 as a unit of price, on the face of it, was easy to understand. For customers wanting wide areas of high-resolution imagery, pricing was easy to digest: the price given depending on the coverage you wanted. This was especially the case for defense procurers in the U.S., who were, and remain, the largest customer for high resolution commercial imagery. The highest resolution data also came with the highest price tag – IKONOS data at 80cm back in the early 2000s was priced at >$100/km2. Defense customers were willing to pay this to add a robustness to their Image Intelligence capabilities.

Even then though, the data price wasn’t as straightforward. Prices would be different for archive versus tasking, or for levels of data processing (raw to analysis ready data [ARD]). Things then become more complex with the arrival of lower-cost satellite solutions delivering similar resolution imagery. The smallsat designs of (then) Skybox Imaging and Planet really started to challenge the pricing status quo. The lower cost satellites measured in the hundreds of thousands as opposed to hundreds of millions meant that data could be priced much lower.

To outsiders, the arrival of new solutions which offered similar ground resolutions could seem confusing. Sub-metric data prices from companies such as DigitalGlobe and Geoeye (the two later becoming Maxar Technologies) had come down into the $25 per km2 range, but were still priced a lot more than what you could get from a smallsat sensor. Why would one pay so much more DigitalGlobe data? I have been asked this question many times by investors in the sector. The EO community knows that there are other parameters in which data is priced, especially geometric and radiometric resolutions… but to the outsider these concepts are more difficult to grasp.

I started to use car analogies. Ferraris, Fords and Tractors are all vehicles, but you use them for different tasks in the same way one would use satellite imagery. But thinking of data in terms of “high-cost” and “low-cost” stuck. It wasn’t necessarily the best for your company to be viewed as “expensive” despite offering a different capability. Price lists then started to become less obvious to find. For “low-cost” data it was also not so obvious how to price – most would require a minimum data order. Selling exactly 1 km2 of data would not be good for business. And so again, price lists were not advertised on websites, rather you would need to enquire as to the price: archive, tasking, level of processing, repeat coverage, size of AOI etc. would all factor into the actual price. Going back about five years, it was actually very hard to find any readily available public information of satellite imagery prices.

Now some service providers and operators are starting to buck this trend to bring some transparency to imagery pricing. Umbra is one such company; the SAR operator lists its prices depending on ground resolution and tasking. SkyWatch sells data at standard pricing depending on resolution, optical or SAR, and from multisource satellites. The open policies of these companies are encouraged and does help to bring a layer of transparency to the industry – an important consideration as commercial EO matures and new users try to understand what EO can do for them, and at what costs.

However, this only purports to address part of the problem: data is still priced in km2, and for most markets, this still doesn’t make a lot of sense. If you need repeat information over a specific AOI why would you need to buy kilometers squared of data that you don’t need? Or, what if you don’t want imagery altogether, and just need the information derived from the data?

Buying in km2 works if you do need a lot of data which could be further processed in-house. Procuring this way is somewhat of a remnant of the defense sector and other established – mostly governmental – user markets, but less of interest to user groups which do not have significant data processing capabilities and/or only require the derived product or information. One way to try and overcome this is to offer solutions through a platform-as-a-service. The service price could be based on km2 of imagery acquired and utilized. However, a simpler process whereby an end-user procures the solutions for a flat annual fee could emerge.

Another way to sell data could come by selling “coverage”. For certain new applications such as in environmental societal governance [ESG], emissions monitoring, carbon markets, even precision agriculture etc. vast quantities of data are needed with repeat coverage. Boiling this down to a price per km2 just wouldn’t make sense (for the areas of imagery required, at say 5m ground resolution, the actual cost for one revisit would be a fraction of a dollar per km2). What is sold to the user, reseller or analytics/service company could be access to all imagery from the satellite constellation over a given time. Obviously, this price would need to make sense based on the investment into the satellite constellation required to support its business, whilst being able to generate value to the client. 

That may be easier said than done, it also doesn’t take into account the preferred business model for the services/analytics companies which leverage the EO data to provide solutions to their clients – the actual final end-user of the EO data derived product or service. But, in order to expand EO services offering and increase demand, a more flexible, open, accessible approach to data sales is needed rather than selling by the Km2 approach with stricter licensing agreements.