Keywords – what is the best approach?
There were multiple questions about keywords in my recent post asking for questions for me to mull over. I’ve thought about how best to do this – copy the questions into this post, or just try to answer as many as I can and you can trust me that I took your questions into account! I think the latter is the best approach, but let me know if you disagree. To complete the circle, I’ll reply to each of the comments that I think I have addressed in this post.
Before I start on this tricky subject, I need to mention that I have a plan, with my friend Brutally Honest Alex, to collaborate with Clemency Wright, an expert in keywording techniques on a new book specifically on that subject. We have a structure in mind, but, to be honest, we are a ways away from actually writing it.
My approach to keywording is to always start with my current favorite keywording tool – IMS Keyworder. I’ve tried the various keywording tools and this one is both fast and easy to use. I search for the words that I think best describe the image, select the examples that best match my image and then choose the keywords that I think fit my particular needs. I probably end up with 20-30 keywords through that process, which I copy into the keyword section in Lightroom. I keep the original files in one massive catalog. Currently I am at 104,000 and this library goes back to 2006 and also contains every picture I have taken – both stock and family type images. I have a simple system of identifying good images – they are flagged. Commercial images have no rating (stars), editorial ones have one star, family images have 2 stars. Because I often want to find an image many years after I have taken it, having the detailed keywords in Lightroom makes this a breeze. I can’t imagine how many individual keywords are in the database now – travel shots in particular need very specific keywords – for instance I was working on some images from Portugal today and this one of the hotel we stayed at for a couple of nights obviously needed very specific keywords that are now saved somewhere in the database:
My keywords for this didn’t need to be too complex. It is obviously an editorial image and so people looking for this image would almost certainly search for the name of the hotel or perhaps the group who own it. So I simply used:
“Alexandre de Almeida, Anadia, Curia, Curia Palace hotel, dramatic, facade, famous, gardens, golden twenties, ground, historic, hotel, lake, magnificent, parkland, portugal, residential, restored, sunset, swan, tourism, traditional, travel, vacation”
Now I look at them, “swan” is not really accurate (it appeared in the previous shot in that pond), but there is a limit to the amount of customization I can do on each specific image. Incidentally, this is a shot where the plain blueish sky was replaced using Luminar4 for a sky that I thought fitted the mood better. No problems so far in getting this modified image accepted.
As you can see, I have both compound words – “Curia Palace Hotel” and also Curia (the town) plus Hotel. I could have added a few more keywords – such as “holiday” – the English version of Vacation, but in general I think buyers will be after this specific hotel rather than a generic holiday shot.
In Lightroom, I simply copy the keywords I get from the Microstock Keyword tool and then add new ones as needed. I then give it a title: “Gardens in front of the Curia Palace Hotel near Anadia in Portugal” and description: “Anadia, Portugal – 20 August 2019: Ornate gardens in the grounds of the historic Curia Palace Hotel” and in these I try to repeat the key words I want to come up in searches. Finally, I use the Map section of LR to add the GPS coordinates of the location as my camera doesn’t do that.
One question asked why I don’t assume that the search engines are clever enough to recognize words like “hotel” that are already in “Curia Palace Hotel” or plurals, or synonyms (similar words). If this was the case, then using 15 keywords would be more than enough for most images. This question illustrates a real conundrum of microstock photography. If I was submitting to just one site, then I could customize the keywords for the capabilities (as best we understand them) of that one site. If the site had a clever search system, then you could keyword to take that into account. But, the reality of the business is that sticking with one agency can be business suicide as it is impossible to know where the buyers for your image are searching. The old days where an image was sufficiently great or rare to cause buyers to search for it across multiple agencies are long gone, I think, and so I upload all my images to all the sites I think are worthwhile. Hence, to be practical, I need to keyword for the lowest common denominator, and so I always try to add words that I think a buyer might use.
This focus on volume and on multiple sites means that I think you can only be reasonably accurate on keywords. For instance, this one:
I keyworded this with: “above, agriculture, beautiful, country, countryside, cultivate, cultivation, douro, douro valley, europe, field, glass, grape, growth, harvest, high above, hill, landscape, nature, outdoor, outside, Plant, plantation, port, portugal, portuguese, red, red wine, river, rural, scenic, stone wall, tasting, terrace, terraced, travel, valley, vine, vineyard, viniculture, viticulture, wine, winemaking, winery”
Most of these are OK. But when I looked at my similar images and I try to minimize the work involved in keywording, then I ended up copying all details across to this image:
Now where is the “river” in this image? Of course, it isn’t there. But, as I say, there is a balance between every image being perfect with its keywords and the amount of time available to spend on this task. I try my best to avoid obviously incorrect keywords, but I am sure I have inappropriate ones on many files. Apart from Alamy and its Click Through Rate, I’m not aware of any negative issues arising from having a few irrelevant keywords in your images.
Finally, in this section, why 40 keywords and not 15? There was an interesting article from Adobe’s technical search team recently where they said that the first image returned in a search had 10 times the number of downloads as the image in position 30, and that half of all downloads occurred in the first 50 images displayed. So getting to that first page is critical to a sale. So if I happen to rank well for “viticulture” in the example above, then there is a chance that my image will appear high for that specific term and get a sale. A “clever” search engine would know that winery was close to viticulture and so not require both, but how would I get that high position if I just use the limited number of keywords? To some extent, adding a keyword that has a common mis-spelling, might also work wonders! I don’t normally do that but worth thinking about!
iStock is an outlier in all this with their controlled vocabulary. And Adobe Stock makes things difficult, by prioritizing the first 10 keywords or so (as does Alamy). Lightroom forces everything into alphabetic order. So my process here is to keyword in LR as discussed. Then export all the chosen images as JPEGs to a new folder that is used by Stock Submitter for uploads. I then use the Stock Submitter keyword editor to select a group of similar images and then move some of the important keywords to the head of the list. It is still possible to use Stock Submitter for iStock (ask if this is possible in the Telegram group supporting the application) and so I then select all the images in my upload group at once and go to the ESP keyword tool in Stock Submitter to choose the relevant controlled vocabulary for each word. Why select the whole set of images at once? I might end up with 250 or so distinct keywords, but I can assign them to the controlled vocabulary all at once, and this is saved in the file or in the App ready for upload.
So in summary – I like to have both compound and single keywords when appropriate. I use synonyms when I think the buyer might use them, or they come up in multiple similar images in the keyword tool, and I like to have reasonably comprehensive titles and descriptions so that where those are used in search engines, I have more chances of my image being found in one way or another. I doubt there is a perfect way, these days, of keywording if you are trying to support multiple agencies. As always, ask more questions on this in the comments below!