Let’s talk about one of our clients that owns a chain of pizza restaurants, but for the sake of privacy, we’ll call it “Sun Pizza”, or SP for short. They’re a small chain of specialty restaurants that use wholesome ingredients in their food, and prepare it fresh as you order it. SP is not like Domino or Pizza Hut. They are upscale, not fast food. They don’t have thousands of locations, but just a few hundred in various shopping malls. Yes, they serve pizza, but the audience, method of marketing, price points, and value propositions are quite different. First, ask yourself, What would make a consumer want to buy from SP instead of the other pizza places? We’ve described that Sun Pizza prepares their food fresh, so targeting the health-conscious and sticklers for flavor is recommended.
Location, Location, Location
Since Sun Pizza isn’t in every city, it doesn’t make sense to buy generic keywords everywhere. We’ll say there are locations in 33 out of 50 US states, so you’d want to exclude not only the 17 states that they don’t have, but not show ads in areas that are more than 10 miles from any other location. You could get really fancy about geo-targeting, but it is probably sufficient to start with just a few geo-targeted campaigns aimed at the top DMAs, or Direct Marketing Areas, which are major metropolitan regions, for example, when promoting a site that lists garage sales. If we were selling headstones and caskets, then we’d multiply the 300 DMAs in the United States against our set of generic terms, creating one master national campaign of 300 ad groups, then separate campaigns for each DMA that we wanted to geo-target. If you are doing this for a single location, then you just have two campaigns, one nation with geo-terms and one geo-targeted with generic terms. However, if you’re selling something like ringtones, then geo-multiplication might not make sense.. Once you learn what works in one master campaign, you can then propagate it to all the others. Also, don’t forget to create local business ads, if your client has a physical location.
Avoid Premature Optimization!
Among other things that are premature, scaling out campaigns too fast will cause you to be embarrassed with your PPC partners. Since geo-targeting is at the campaign level in Google (but can be done at the ad group level in Yahoo!), if you have 300 locations, that requires 300 campaigns. If each campaign is split by search versus content, then you now have 600 campaigns. With a limit of 25 active campaigns in Google per account, that would already be 24 accounts to manage. Spinning out that many accounts would be premature optimization, even if you have tuned your scripts to post these through the AdWords API in a snap. Make sure you learn what works in one or two prototype campaigns first, else you risk building combinations of keywords that have absolutely zero traffic, solving problems that don’t exist, and basically wasting your time. And if you have 300 ad groups that are essentially duplicates, when you change the master, say, add new keywords, change bids, put in new ads, you then have to propagate to the 299 other slaves. Even if automated, you still bear API usage costs, which can eat up your budget in a hurry. I’ve had days where I’ve spent a few hundred dollars alone on API usage charges.
The point here is to start with a couple of hand-built campaigns and tune until you have something that is reasonably good, then scale to the moon. If you’re doing geo-targeted campaigns, start with one or two test campaigns, don’t start in NYC or some metro that is a uniquely monstrous challenge.
Users tell you where they are
Depending on whose numbers you use, the percentage of searches that are local is between 20 and 30%. At the low end, the user explicitly includes the geo term in their query: “Omaha dentist” (by the way, I’m not counting “sun pizza” or “Boston market”, where the brand name has a geo term in it). Someone could be searching on “Toyota dealership” by itself, upon which you infer that it’s a local search and determine where they are based on geo-IP. So include the inferred local searches and almost a third of searches are for someone looking to buy a product or service in their neighborhood.
Thus, you can determine local via either the user’s search term or via their IP address:
- With the search term, you do geo-multiplication against your geo list and generic term list, 50 geo terms, and 100 keywords and you’re at a 5,000 keyword portfolio all in the same nationally targeted campaign. Be careful of cities and neighborhood names that are not unique. For instance, if are looking for a Louisville Roofing Company, you could be attracting traffic from both Louisville, CO, and Louisville, KY. There is also a major risk of multiplying too far, where you’re multiplying a low-population zip code against a low-volume search query. In geo-multiplying, our experience is that you only need to take the top 3 city names in the local area, multiplied against the top 10 things that you or your client do. The drop-off after is so steep that the long tail becomes non-existent.
- If going by IP address, you have to multiply campaigns, each with different geo targets. In our experience, the geo-targeted campaigns in aggregate should give you more traffic than your single nationally targeted geo-combo campaign. However, you will have many more geo-targeted campaigns. By the way, I highly recommend that you sign up for Shoemoney’s amazing local affiliate tools, which will simplify much of the drudgery I’ve laid out above.
Give them what they want
Now that you have your targeting in place for your PPC ads, you have to make sure you have the right message aligned with the right keywords. Organize your ad groups such that you could show the same ad on any keyword in the ad group and know it will perform just as well. For example, have one ad group for each product and service. But don’t create a zillion ad groups needlessly. For example, “pizza” and “pizzas” should not be in separate groups. However, “order pizza” and “pizza menu” should be separate. If you’ve got the keywords grouped together by similar intent, then it’s a simple matter to make sure that you have a relevant landing page to show them. Use Quality Score (Google) and Quality Index (Yahoo) as a double-check for relevancy. Use Google Website Optimizer if you want to test your landing pages.
Show up in the map and natural local search results
PPC is not the only game in town. There is a ton already written on Google Local Business Center, Yahoo Maps, and various other free and paid directory players. Google Local Business Center is perhaps the granddaddy of all of these guys because of the volume of traffic you can easily drive via the Local 10 pack results which sit above the regular organic results, it’s the map with the 10 listings, and the sophistication of the tools they provide. Even if you have 5,000 locations, and we have some clients who in this range, you can bulk upload all your listings. But optimizing each of these listings for each directory is something that will either take you a ton of time manually or for which you have to employ some clever automation. You didn’t hear me say to spam Craigslist, but I did say there are ethical, white-hat methods where you can get your raving customers to go in and stack the customer reviews, which will bump up your rankings.
One major error we commonly see with retail clients is to list all their locations on a single page that is a mile long, versus having one page per location with semi-unique content on each page. How do you have unique content on each page if you have 5,000 franchise locations that all sell the same thing? You use the “Mad Libs” approach with templated content. Welcome to {city} {category} for the best in {subcat1}, {subcat2}, and {subcat3}. We’ve been doing {category} in {city},{state}, and {region} since {founding year}. And so forth, making sure that the auto-generated content is not exactly identical. Then, allowing fans and employees to blog and comment will increase the level of unique content. Good SEO on these pages will help your listings show up on the map and regular organic results. Make sure that you don’t make any of the fatal SEO errors that we usually see:
- Check for canonical domain issue: Do the www and non-www versions of the homepage go to the same url? If not, then look at www.mattcutts.com/blog/seo-advice-url-canonicalization/ for an understanding of how duplicate content should resolve to a single page.
- Dynamic-looking urls: the site should not pass parameters in the url. For example, mysite.com/?page_id=34§ion=2&session_id=123abc is no good. Question marks and equal signs in the url mean that parameters are being passed. Instead, the url should look like mysite.com/large_blue_widgets
- Text vs. Images/flash: Search engines cannot read images or flash, they need text.
- Unique page titles: Do not start the page title with the name of the site unless it’s a strong brand. Make sure each page title is reflective of what that particular page is about.
- Sitemaps: There should be a link to “sitemap” on the home page—and that sitemap should have links with anchor text that are key search terms
- Run SEO reports: Put in the client’s url at SEOMOZ Page Strength. This site gives you a couple of dozen SEO metrics
- Run PPC competitor reports: Go to SpyFu and do a few searches. Put in a few keywords that you think are important to the site, plus the site url itself. If you’re looking at a keyword, it will come back with who is buying it on PPC, what it costs, how many clicks are available, and what terms are related. If you’re looking at a url, it will tell you what terms (if any) they are buying on Google, what terms they rank naturally on, and who are related competitors.
- Examine the Source code: Do a view page source from your browser, starting from the home page. Do a search for H1 and see if any come up. If so, are they using keywords that they want folks to find them on? H1 tags carry more weight than regular text but are not as strong as the page title.
- Look for Analytics code: While still on that view of the code, find instances of “Google”, “analytics”, or “urchin”. You may also see examples of other tracking software, too, towards the bottom of the body.
- Validate Your Source: Go to W3.org and put in the url. Like the other tools, it will return usually a few dozen errors with explanations.
- Use a Toolbar for “at a glance” SEO: SEOQuake is a great Firefox browser plug-in to show you Google PR, pages indexed by Google, pages indexed by Yahoo, backlinks, age of the site, and Alexa rank
Measure conversions
If you’re not tracking conversions, you are flushing money down the toilet, Period. There are several vendors that will give you hosted call tracking for a few dollars per number per month, allowing you to have a web-only number that forwards to the nearest branch location. And you should pixel email signups, shopping cart checkouts, and any other conversion event to which you can assign a value.
By the way, put branded terms in their own groups, since they will convert better and have a higher volume. Most users don’t know the difference between the search bar and the address bar, so they’ll navigate to your site via the search engine. Thus, you can’t really take credit for those sales, as most would have come to you anyway via natural search. That’s not saying you shouldn’t bid on your own name or that of competitors, You have to be smart about it and measure the level of cannibalization occurring. For a more sophisticated treatment on conversion optimization, check out my presentation on Advanced PPC tactics at SMX Sydney last week.
Optimizing your campaigns
Now that you’ve built your campaigns, tested them to the point of being stable, and then used programmatic methods to scale them out, you have to optimize budgets. Do you have any keywords that are still inactive? Are you more than 10% off where you should be pacing on your monthly budgets? Then either adjust your bids or change your budgets. Go back to the keywords tab and sort by minimum CPC bid descending. Look at anything that is $5 or higher, since that is a “Google slap”. Google thinks that your ads and landing pages have very little to do with that keyword, so they are forcing you to bid a certain amount to even be able to show your ad. This is not the same as your average CPC is high. You can be paying $10 a click because you’re bidding on a highly competitive keyword, but with a decent Quality Score, your minimum required bid can still be 50 cents. The latest version of AWE now shows Quality Score, else you can get it on the web. Look at any keyword that is not at least an 8. If anything less, then you need to re-group keywords into tighter themes, write new ads, and find more relevant landing pages.
Final Words
So there you have it. Whether you have a few locations or 5,000, these are relatively easy steps you can follow to drive measurable traffic to your office, car dealership, franchise, or other locations. If you want to get fancy, you could even do PPC via Facebook Self-Service, update a Twitter account, or build a full-blown Facebook game. I hope you found this helpful. If you have any questions, feel free to drop me a line at dennis@blitzlocal.com. If perhaps this type of work is something you don’t want to do in-house or you would prefer assistance from folks who do it professionally for a number of Fortune 500 clients, we are happy to talk to you about our service offerings.
