This is part 4 in a series of blog posts capturing our SEO makeover for the Melete Foundation.
Keyword research is the foundation of good SEO. The goal of this step is to identify a list of 15-30 “target keywords” — phrases that people interested in your type of business are likely to search on and that you want to rank well on. It is not a step to take lightly. If you do mediocre keyword research, you’ll likely end up with a completely different list of target keywords than someone doing great keyword research. Since a number of future steps in the optimization process rely heavily on this list, everything else will suffer as a result.
In Melete’s SEO kickoff meeting we spent about a half hour brainstorming keyword ideas, but that was just to provide us with a starting point. Ben from Melete joined me at Affinity Lab for most of the official keyword research session, which lasted about 2 hours. Here’s what we did:
1. Review Competitors.
As is often the case with non-profits and foundations, Melete doesn’t have any organizations they consider true competitors. But there are groups doing similar things, and at the end of our kickoff meeting I asked the Melete team to come up with a list of them. They later sent me a list of 8 similar, yet more established, organizations: Projects Abroad, i-to-i, CDS, CIS, Global Volunteer Network, Cross Cultural Solutions, World Teach, and CIEE.
2. Learn from Competitors’ SEO.
I quickly looked at all 8 sites and identified 4 that seemed to have done some SEO, by looking for keyword-rich title tags. I focused in particular on 2 (Cross Cultural Solutions and CIEE). By looking at title tags and home page content, I was able to add a number of new potential keywords to our starter list, such as: international community service, teach overseas, international volunteer travel, teach in [country], volunteer abroad, and fellowship program. These were phrases that we’d missed with our initial brainstorm, and some of them, like “volunteer abroad,” opened up an entirely new set of words that would become part of our target list later on — and that we might have missed otherwise. Competitor sites are a great way to fill in the gaps in your brainstorming before you start deep diving into keyword types.
3. Bucket Keywords.
Next Ben and I took the full list of keywords so far (brainstorm + competitor research) and broke it into similar-themed groups. As is usually the case, there were multiple ways to slice and dice the list; for instance, “summer teacher exchange” could be in a “summer” group or in a “teacher” group. In cases like that, we put them in both and then later narrowed down the list. The final 5 groups were:
Each bucket had an average of 5 keywords.
4. Pull Volumes and Expand Buckets.
With some initial structure in place, it was now time to turn to data to expand, improve, and eventually narrow down our list. We went to Google’s keyword research tool and pasted in each bucket of keywords, one at a time. Each time, Google spit back monthly search volumes for our words as well as a long list of related words. We scanned through the keywords and, if there were obvious words that we should have included in the original buckets, we added those to the search box and reran the search.
This allowed us to pick up an even wider range of variations. Sometimes we redid the search a few times, sometimes only once. Once we felt like we’d caught all the major words for that bucket, we then went through the search results, checked the ones that seemed like good matches, and exported those to a spreadsheet. We did this for all 5 groups, using a separate tab for each. See these in the last 5 tabs of our Melete Keywords doc.
5. Combine Buckets.
We then copied all keywords from each of the 5 tabs and pasted them in one sheet. We ended up with about 80 keywords. The next few steps are time intensive per keyword, so often I will narrow the list down at this point by search volume if the combined list is 100 or more (e.g. remove all keywords with search volume under 200 per month). In this case, I decided to leave the list as it was because 80 is a manageable number and there weren’t obviously irrelevant words on the list. See the full list of keywords on the first tab of our keywords doc.
6. Add Relevance Scores.
Volume is a great way to evaluate keywords, but there are 2 other critical factors: relevance and likelihood. Of the 3 criteria, relevance is the only subjective one in the way we score it. I define it as the “fit between site content/service and likely searchers’ intentions, relative to other keywords on list.” Another way to think about it is, will most of the people doing this search stick around if they land on your site? Scoring this is mostly a gut reaction, but it’s good to run some of the main types of searches to see what’s currently popping up on Google. Sometimes we’re obvlious to completely different, equally popular meanings of search phrases (e.g. “paris hilton” the hotel vs. “paris hilton” the person).
Ben and I started by sorting our spreadsheet by keyword, to try to get similar keywords near each other. Then we flew down the list and assigned a relevance score of 1 to 5 to each phrase, with 5 being the most relevant. I entered a number and Ben shouted out if he disagreed. After we had scores for all 50 keywords, we sorted by score to see where the distribution was skewed. Since this is a relative score, and since we’re trying to prioritize keywords at the end of the process, we want a roughly even distribution. So we adjusted some scores to even it out. See our relevance scores in the first tab of our keywords doc.
7. Add Likelihood Data.
I define “likelihood” as “probability of achieving page 1 Google ranking after SEO, relative to other keywords on list.” You could also call it “competitiveness.” To score likelihood in an objective way, I use a great technique I learned from Jill Whalen: “allintitle” searches on Google, which tell you how many web pages have that phrase in their title tag. Jill’s argument for using this technique is this: “Because title tags are given so much weight by Google, any page that is using the phrase in their title tag is at least rudimentarily optimized for the phrase, and is therefore one of those that you’re competing against.”
For Melete, I took each of the keywords on our list and entered them into a Google search using this convention: allintitle: “[keyword]”. Then I took the number of results Google returned and entered that in my spreadsheet. See the data I entered here, under the “Title Pages” column in the first tab.
8. Convert Volume and Likelihood Data Into Scores.
At this point we had data for all 3 criteria, but we needed to get volume and likelihood on the same 1-5 scale as relevance. So we set up a legend for each on another tab, and then used a VLOOKUP function to convert them all into scores. As with relevance, volume and likelihood are relative scores. The scale doesn’t need to be a perfect reflection of actual values, and we want similar numbers in each group. See the scoring legend and conversion formula I used here, under the Legend tab and the Volume and Likelihood columns in the first tab.
9. Total Scores.
Whew. It’s been a lot of working getting here but now we enter the fun phase. We added a “total” column and summed the 3 scores, and then ordered in descending order.
10. Identify Targets.
With total scores in place, it was finally time to identify our target keywords. We added a “targets” column and assigned a 1, 2, or 3 to about 25 keywords total (5-10 of each number). We could have just gone by the “total” column, and made the top total scores 1s and so on. But it’s best to combine the scores with some human judgment, which we did. See our final list along with all scores and tabs here.
Keyword research can be a grueling process. The reward comes later when you’re optimizing the site and eventually analyzing your SEO results. When you’ve done a thorough job with keyword research, you don’t waste time second-guessing and rechecking your keyword list throughout these later steps. And of course the real reward comes when you see more high quality traffic coming to your site. Hopefully we’ll see that soon for Melete. But we’re not there yet.
Up next: content and tag optimization.