One age-old question that often comes up when I chat with new prospects or new clients is, “Should I bid on my brand terms or generic terms where I rank in the top three organically?”
This question is hardly trivial. It matters a great deal because PPC search budgets are generally constrained — and now, with other enticing options available from Facebook and other programmatic channels, we need to demonstrate that our search advertising budgets are being allocated optimally.
The best way to answer this question is by using a multi-stage process of evaluation, followed by experimentation. Let’s start by evaluating brand keywords. Before you start your evaluation or experiment, you need to answer six important questions:
Now it’s time to introduce a few economic concepts that directly apply to the “Should I bid?” problem. One is “opportunity cost” — the cost/loss of not doing something. The second is the idea of marginal net profit. That’s about understanding the relative profit of all the search and paid media you are engaged in. (Another way of thinking about marginal net profit is by asking yourself this: if you were to invest another $100 in a particular channel within search, keywords, social, display and so on, which investment would deliver the highest return?)
Finally, there’s the concept of “an inelastic auction.” Inelastic auctions are said to exist when obtaining more volume results in dramatically escalated costs as you battle others for top positions.
That’s it for the economic concepts (whew!). But there’s another factor that applies here: the concept of cannibalization — the primary driver of the “Should I bid?” problem. Cannibalization, in the context of PPC, refers to situations in which marketers’ paid and organic listings compete with each other for search clicks. When this happens, marketers often attempt to pare back PPC spend, reasoning that “they typed in my brand and want to visit my site, so we’ll get them anyway, eventually.” Your own “cannibalization index” can be determined by evaluating your answers to questions 1–6 above.
Okay, you’ve got the concepts. Let’s move on to solving the question. (Note: the methods below aren’t perfect, but chances are that using them can move you closer to a solution.)
Step 1: Calculate CTR (click-through rte) on Organic. Search Console can be used to determine this.
Step 2: Turn on or off paid search nationwide/globally, and then compare organic CTR pre- and post-test.
Step 3: Calculate total cumulative CTR when the paid and organic listings coexist in the SERP. (Cumulative CTR lets us understand the incrementality on total clicks and CTR of paid.)
Step 4: Compare conversion rate on paid vs. organic, including taking PPC and organic sitelinks into account. (Often marketers use different landing pages for their paid primary and sitelink visitors, and those tuned landing pages out-convert the organic pages. If this isn’t true in your case, then your calculations are made easier.)
Step 5: Take the drop in organic clicks as a result of your paid listings getting the clicks, subtract that number from the paid click count, and re-calculate the ROI of the paid advertising, based purely on the incremental clicks that the paid campaign delivered. (You may need to use a “fudge factor” if conversion rates on the organic listings were lower.)
“But wait,” you protest, “I can’t just turn off brand keywords during a test. My boss will kill me if sales drop! Besides, where else am I going to put that budget? Most of the other keywords are already close to minimum acceptable ROI, and I can’t use the brand dollars to bid those up!”
OK, fair point. In many cases, it still makes sense to bid on branded keywords, even after taking cannibalization into account. If you don’t want to send too many ripples through your current campaigns at one time, limit the scope of your experiment, for example, by testing results in five mid-sized cities. (You’ll have to use location bid modifiers/adjustments to do this. See
About bid adjustments from the AdWords Help documentation.)
If you choose to run your test in specific geographies instead of nationally or globally, you’ll have the added advantage of conversion data that isn’t subject to seasonality or other factors that might show up in a purely longitudinal test (test over time with one time period being the control group).
Internal company politics also play a role. You may want to add a fudge factor to your calculations based on a Marketing Mix model calculation, or if you are still building a brand and want to be recognized as a market leader.
If you’re like most marketers who run this test, you likely will find that bidding on your brand keywords still makes sense from an ROI, net search profit and budget allocation analysis. Others, however, may find that paring back spend — or increasing it — increases net search profit (profit after deducting all costs). For high organic positions, the answers are more dependent on landing page variations and head or tail terms.
One thing is for sure: if your boss is questioning brand spend or spend on high-ranking organic keywords, run a test like the one above. You’ll gain the confidence that you’re doing the right thing or, at the very least, are on the right track.