A minor disaster occurred. Our insurance industry client’s paid search campaign was in full swing. Then, its site experienced a critical failure. The client had made what it thought was a routine change. However, the code that told the server which toll-free number to display on the paid search landing pages had failed.
The site’s functionality worked as it should earlier. It detected our bidding agent’s query string, which revealed each visitor’s specific search engine and keyword phrase. It then served the entire site to that visitor with a unique toll-free number that could be tracked back to the paid search campaign, keyword by keyword. Visitors from natural search were served a generic toll-free number.
Suddenly, site visitors who had clicked on paid search ads weren’t seeing a unique toll-free number. Instead, they saw the same general number as those viewing natural search results. Associating the source of paid search clicks to the insurance applications that came in over the phone became impossible. As a result, the client asked us to halt the paid search campaign while it worked on correcting the problem.
It took nearly six weeks for the client to fix the problem. No conversion tracking to a unique toll-free number, no marketing data on which to optimize the pay-per-click (PPC) campaign, and, therefore, no PPC campaign running while repairs were made. In hindsight, this cloud had a silver lining.
Why Are There Still Conversions?
Our client should have seen zero conversions while it was fixing the problem. The phone should have stopped ringing. After all, no searchers were finding the site via paid search. No search engine visitors were seeing specific toll-free numbers tied to the paid search campaign. The campaign couldn’t possibly drive calls to a phone number that wasn’t displayed, right?
Oddly, the client saw its application volume continue at roughly 20 percent of normal. There they were: conversions from a halted campaign. An impossible result?
The Conversion Lag
The client discovered its site experiences a conversion lag. For the next 45 days or so, sales came in on toll-free numbers that hadn’t been shown to visitors in over a month. Visitors who had clicked a search ad in Google or Yahoo had apparently written the number down and called weeks, sometimes months, later to fill out insurance applications.
The Reporting Disconnect
Prior to the site’s breakdown, we prepared a weekly report on how many applications were generated, keyword by keyword, engine by engine, and at what cost. We raised or lowered the client’s spending based on how well the campaign performed the week prior, sometimes raising the budget by as much as 100 percent or lowering it as much as 50 percent.
But some people took their time to mull over the decision before buying. Some, we learned, shopped around before pulling the trigger.
Until that revelation, the client made decisions based on the weekly numbers; increasing spending after “good” weeks, decreasing it after “bad.”
Problem was, the prior week’s conversions weren’t reflecting just the prior week’s paid search spend. They were also capturing conversions from people who visited the site weeks before on keywords that were being targeted month after month.
The Strategy Change: Waiting
In some verticals, if we adjust a client’s budget weekly based on the previous week’s results, we adjust the campaign too quickly. We don’t allow all results to come in. We fix things that aren’t broken and break things that work properly.
Today, we wait at least a month, sometimes two, in many verticals before changing PPC ad campaigns. If we increase campaign spending today, we must wait until all potential conversions generated by that adjustment are reflected in the campaign’s performance. Depending on the vertical, this can take as long as 90 days.
If you’re engaged in PPC advertising and adjusting spending or optimizing your campaign based on week-old data, you may be leaving money on the table. The higher the consideration for your product, the larger the conversion lag is likely to be.
The Phone Number Swap Test
An easy way to learn if there’s a conversions lag is to occasionally change the toll-free numbers associated with groups of keywords. Note any conversion drop-off, and pay special attention to how many weeks you still see conversions to the “expired” phone number.
When you know your conversion lag time, the adjustment is simple. Alter your campaign based on lag-time data rather than the previous week’s data. You’ll be surprised how many products have conversion lags. Overanxious marketers make course corrections so frequently, they instead make course collisions.
Join us for Search Engine Strategies in Toronto, Canada, May 4-5, 2005.
Want more search information? ClickZ SEM Archives contain all our search columns, organized by topic.