Search for advice on keywords for photography and you will mostly find articles written for online shops, lightly repainted. They tell you to research search volume, check difficulty scores and target long-tail phrases. None of that is wrong, but it answers a question most photographers are not asking, and it ignores the fact that a photographer usually has three separate keyword problems running at once.
There is the portfolio site, where you want work found by people browsing or hiring. There is the stock portfolio, where an agency's internal search decides whether a buyer ever sees a file. And for many photographers there is a client-facing business, where the searches that matter are local and commercial. These use different vocabularies, different fields and different success measures. Treating them as one problem is why keyword effort so often produces nothing.
Problem one: keywords for a photography portfolio site
The honest starting point is that a portfolio of images with almost no text is a hard page to rank. Google needs language to work with, and a gallery gives it very little. The photographers who do rank are almost always the ones who write.
What actually moves the needle: a descriptive alt attribute on every image, written as a natural phrase rather than a keyword list; filenames that describe the subject rather than the camera's counter; and real prose on the page - a paragraph about the shoot, the location, the conditions, the client brief. A single well-written project page will out-rank a bare gallery of the same photographs almost every time.
Good site keywords for photography tend to be specific rather than broad. "Photographer" is unwinnable. "Documentary wedding photographer in the Scottish Highlands" is a phrase a real person types and a page can actually satisfy. The narrower the phrase, the smaller the audience and the higher the proportion of that audience who wanted exactly you.
One structural note: give each project its own URL. Sites that load an entire portfolio into a single scrolling page have one address for fifty subjects, and no amount of keyword work fixes that.
Problem two: keywords for a stock portfolio
This is a completely different exercise. There is no page, no prose, no context - only a title, a ranked keyword list and a category, read by an agency's own search engine. The rules are published and specific.
Adobe Stock accepts up to 49 keywords per submission and states that the first ten positions carry the greatest influence on ranking. Shutterstock requires a minimum of 7 keywords and a maximum of 50, with a title that must be a factual sentence about the subject. Both refuse metadata containing trademarks or brand names, and Adobe additionally instructs contributors to use each keyword only once and to keep a submission in a single language.
The single biggest improvement most stock portfolios can make is adding conceptual vocabulary. A photograph of an empty office at night is literally "office, desk, chairs, night, window". Buyers looking for that image are searching "overwork", "burnout", "remote work", "corporate isolation". Literal keywording describes the pixels; conceptual keywording describes the brief the buyer is trying to fill.
Adobe's own guidance points the same way, recommending contributors combine general and specific terms with conceptual ones, plus setting, viewpoint, number of people and demographic detail. Our deeper piece on the best keywords for stock photos works through how to build that list systematically rather than by intuition.
Problem three: keywords for a client-facing photography business
If you shoot weddings, portraits, products or events for paying clients, the keywords that matter are commercial and usually local. The pattern is service plus place: "newborn photographer Manchester", "product photography studio Austin", "corporate headshots Berlin". These phrases have modest search volume and very high intent, and they are frequently winnable by a small site because national competitors are not targeting them.
Two things matter more than keyword placement here. First, a business profile with consistent name, address and phone details across directories, because local results lean heavily on that. Second, pages that answer the questions a buyer asks before booking - what a session costs, how long it takes, what they receive, how far you travel. Those pages naturally contain the phrases people search, without anyone forcing them in.
Photographers frequently mix this vocabulary into stock keywords, which does nothing. No stock buyer searches for a photographer's city of business, and location keywords on stock files should describe where the image was taken, not where you are based.
A method for building a keyword list that is not guesswork
Start from the frame, not from a tool. Write one plain sentence describing what is in the image. Everything else derives from that sentence, and it doubles as your agency title and, trimmed, your alt text.
Then answer six questions. Who or what is the subject? What are they doing? Where is it, and what is the setting - indoors, outdoors, day, night, weather? How is it framed - aerial, close-up, high angle? How many people, and what can be accurately said about them? And finally, what is this image about, as opposed to what it shows? That last question produces the conceptual terms most contributors never write down.
Rank before you save. Because the leading positions carry the most weight on Adobe Stock, ordering is editorial work rather than housekeeping. Put subject, action and primary concept first; push viewpoint, demographics and setting detail into the tail.
Then cut. The caps are ceilings, not targets. Twenty-five accurate terms outperform fifty padded ones, and irrelevant filler is an explicit metadata rejection reason at both major agencies.
Where automation helps and where it does not
The six-question method takes a couple of minutes per image. At twenty images that is fine. At two thousand it is a week of work, and this is the point where photographers either stop uploading or start automating.
AI keywording is genuinely good at the literal half - subject, setting, colour, composition - and consistently weaker at the conceptual half, which is exactly the half that sells. The practical approach is to let a tool produce the base list and spend your attention on the first ten terms. Rastock AI generates the title, description and ranked keywords, embeds them as IPTC and XMP, and applies banned and mandatory keyword rules across a batch so trademarks and previously rejected terms never reach a submission.
Because the same output exports as CSV and embeds in the file, one generated description can serve the agency submission and the alt text on your own site without being rewritten. Plans are tiered by monthly upload volume - see current pricing - with no revenue share and no ownership claim on files.
What automation does not help with is the portfolio site and the local business. Those need writing, not tagging, and no tool substitutes for a page that answers a real question.
Common mistakes worth unlearning
Stuffing camera and technique terms. "Bokeh", "85mm", "golden hour", "long exposure" describe how you made the picture. Buyers search for what the picture is of. Adobe explicitly advises against gear-heavy titles.
Padding with singular and plural forms. "Child" and "children" count as the same keyword used twice, which Adobe lists among the mistakes leading to refusal - not as broader coverage.
Reusing one keyword set across a whole shoot. Fifty frames from the same session are not fifty copies of the same image, and identical metadata makes the agency's deduplication logic treat them as near-duplicates.
Leaving keywording to the end. It is far faster to write metadata while you still remember the shoot than to reconstruct it from a thumbnail eight months later. Our stock photography workflow guide covers where keywording sits in a repeatable pipeline from shoot to submission.
What to do first
If you sell stock, spend an hour adding conceptual terms to your twenty best-performing files and watch what happens over the next quarter. If you run a portfolio site, write three hundred words on your three strongest project pages. If you take client work, build one page per service and location and get your business details consistent across directories. Each of these is a different keyword problem, and doing all three badly at once is how most photographers spend a year without moving.
Frequently asked questions
What are good keywords for a photography website?
Specific phrases that match how clients actually search, usually combining a style or subject with a place - "documentary wedding photographer Edinburgh" rather than "photographer". Broad single words are effectively unwinnable for a small site. The bigger factor is having real text on the page: galleries with no prose give search engines almost nothing to match a query against.
How is stock keywording different from website SEO?
Website SEO is contextual - Google reads alt text, filenames, captions, headings and surrounding prose. Stock keywording is a ranked list read by an agency's internal search engine with no page around it. Adobe Stock weights the first ten keywords most heavily and caps at 49; Shutterstock requires 7 to 50. The two need different formats built from the same description.
Should I include camera settings or technique in keywords?
Rarely. Terms like bokeh, 85mm or long exposure describe how the image was made, not what it shows, and Adobe Stock advises against gear-heavy titles specifically. A small number of buyers do search technique terms for stylistic reasons, so one or two can be justified, but they should never displace subject, action or concept from the leading positions.
How many keywords should a photograph have?
Usually fewer than the limit. Adobe Stock permits 49 and Shutterstock allows up to 50, but both reject metadata that is padded with irrelevant terms. Twenty to thirty accurate, non-duplicated keywords generally performs better than a maximised list, because relevance scoring is diluted by terms that do not describe the image.
Can AI write photography keywords well enough to submit directly?
Usually accurate on the literal level and weak on the conceptual level. AI reliably identifies subject, setting, colour and composition, but tends to miss the abstract terms - ambition, isolation, sustainability - that buyers actually search when filling a brief. Treat generated lists as a strong first draft and edit the first ten positions yourself.
Do location keywords help on stock images?
Yes, when they describe where the photograph was taken. Adobe recommends location-based keywords and Shutterstock has a dedicated location field. What does not help is adding your own city of business to stock metadata - no buyer searches for the photographer's location, and irrelevant terms count against metadata relevance at review.