This article is the second part of the two-part series: "Time is Money in PPC advertising" (part one here)
Whether you’re paying an agency or internal staff to manage your search program it’s important that time is spent on the highest value projects available. Human resources are always finite; allocating those resources wisely is good business.
Often too much time is spent tweaking ad copy. Once the copy is well targeted and a few tests have been run to identify the strongest message it’s almost impossible to generate ROI for the time spent on testing further.
This begs the question: what should we spend our time on?
Last time I laid out the relative values of different broad pieces of the search advertising game: Term list 50%, Bid Management 30%, Landing Page/site layout 15%, Ad Copy 5%.
Clearly, if any one of these is really awful it can render the others irrelevant. If you sell thousands of products but only have 100 keywords there’s no point in testing landing pages. If you have tens of thousands of well-chosen keywords all landing on the homepage, bid management can’t save you. There is a certain level of triage that needs to go into deciding what projects need to be tackled next.
Assuming the program is reasonably healthy here’s how I’d recommend allocating human resources:
Term List Maintenance: 10% - 30%
Whether your business is closer to 10% or 30% depends on how thorough the list is already, how complex your business is as measured by the number of product categories, manufacturer brands, and SKUs you carry, and how much turn-over you have in your product offerings. In CE the term list will require more maintenance, in banking/finance somewhat less.
- Look for new and different combinations of words from product names, SKUs, manufacturers.
Focus on the products that are actually sold through search. Have smart people think of every different way to describe those products. It may seem like you should focus on the items that aren’t selling, but assuming you already have some terms for all your products remember: search responds to demand, it can’t create demand.
- Study the actual user searches that led to the clicks, particularly those that convert.
- Use product feed updates periodically to catch the new additions with particular attention to new manufacturers/lines.
- Drop terms that reference products no longer carried.
- Spend time on the site studying how products are described and presented. The smart human will recognize opportunities that no machine will.
- Don’t burn time with machine generated lists. Huge amounts of time wasted culling these lists of awful terms, terms you already have, and misleading terms that would have you competing for the wrong traffic.
- Don’t waste time studying Hitwise reports of terms that “are big” for your competitors. Hitwise makes no distinction between paid search and natural search, and they obviously have no idea what your competitors spend on their PPC ads. Moreover, most of your time will be spent tossing out terms that you already have in your list, or that reference categories and brands that you don’t carry. Focus on your business and what works for you.
- Keep an eye on the performance of these new terms. At some point you may get to a point where the new terms aren’t adding any tangible benefit. That’s the time to focus more time/attention elsewhere.
Landing Page Tests/Design/Research: 5% – 15%
Again, you’ll want to spend time wisely here. Frontload your efforts, spending more time on this early in the program than later.
- Test search results pages against sub-category pages early and pick the winner.
- If you have the ability to test different mock-ups of the winning template do so.
- Test dedicated landing pages for extraordinarily high traffic terms that perform reasonably well.
- Study what products are sold from the top keywords and consider constructing landing pages for those keywords that highlight those products.
- A/B test homepage versions for brand terms if you have sufficient traffic.
- Don’t test whispers, test shouts. Finding statistically significant differences is hard on a good site. Focus on testing big concepts, not nuances.
Ad Copy Testing/Maintenance: 5% – 10%
People spend too much time on copy, but that doesn’t mean it’s unwise to spend any time on it. Copy maintenance can be extremely time consuming, so make sure you don’t over invest.
- Initially, test which unique selling proposition works best for you as a tag line.
- Write modular copy, so that the line 2 and line 3 copy can be swapped out without breaking a sentence.
- Test offer copy vs standard copy at the first opportunity of a site wide sales event.
- If the previous test supports it, roll out relevant offer copy whenever the offer is wide enough to justify the time.
Data/ Bidding Analysis: 50% – 90%
There is far more gold to be found in the mountain of data than in any other area. The hours spent by a sharp, experienced analyst studying data, looking for trends, finding opportunities and refining the bidding logic will almost always produce more ROI than the other activities. This means other activities always incur opportunity costs associated with time not spent on analysis. See the results of our study on PPC management styles.
The list is endless, but here are some ideas:
- Study performance by product category, sub-category, manufacturer brand, etc.
- Study performance by landing page.
- Study the above in the context of seasons and changes in product mix.
- Assess the impact of time of day and day of week on performance. Fold findings into the bidding algorithm.
- Look for commonalities that don’t fit into the groupings above “terms with the word ‘foo’ in them do better/worse than their cousins”.
- Assess the keyword level data over longer and shorter time windows. Terms relating to higher ticket items may need to be assessed over a different window than those associated with “impulse buys”.
- How much spillover is there from the search program to the call center or the retail stores? You see all the costs of PPC advertising but lose tracking on some of the sales generated.
- What fraction of the buyers are new to file vs existing customers? Does that vary by product category? Particularly important to separate brand from non-brand here.
- How does traffic coming in on broad match perform compared to the traffic coming in on exact match?
- How does traffic from the Google syndication network perform vs traffic from Google.com?
- Does our bidding correctly reflect margin differences between different types of products? Can we tie that tighter? Do people searching for “V Neck Sweaters” actually buy sweaters?
- How do orders come in over time since the last click through? Does that order curve suggest a longer or shorter cookie duration?
- What kind of interaction is there between search and other online ads? How often do people buy with ads from multiple ad programs and what are the rules for assigning credit. Think about different scenarios: Competitive PPC click followed by an affiliate link (user searched for “[your brand] coupons”) – which program drove that sale?
These types of analyses produce much greater return on investment than most other projects. However, finding nuggets of gold takes hard work, and persistence. You won’t find gold every day or every week.
Sometimes it’s tempting to focus on the more easily demonstrated projects: “I added X keywords”, “I got our ads out on NobodySearchesHere.com”, “I changed the copy to reflect that granola bars are $9.95 this week”.
However, by putting more time into analysis and less time into low value projects, you’ll get more from your PPC program.