A homeowner finds ants in the kitchen on a Tuesday, searches "ant exterminator near me," clicks your ad, and calls. The CSR quotes the quarterly plan. The homeowner says they will talk to their spouse.

Friday they search your company name, tap the phone number on your Google Business Profile, and book.

Your tracking platform has both calls on file. Ask almost any report where that customer came from and it says Google Business Profile, because that is the last thing they touched before they booked. The ad that found them on Tuesday gets nothing.

That is last-click attribution, and in a pest shop it does something specific: it hands the credit to whatever surface sits closest to your phone number. Usually that is your profile listing or a search for your own name. Last click is grading your brand name and calling it a marketing report.

This is about customers who touched you twice

Start from the measured pile only: new customers you can tie to a tracked call or form by an exact match on phone or email. If you have not split measured from estimated yet, the lead tracking post walks through it. Everything below happens inside the measured pile, so none of it is a guess about who called the main office line.

Inside that pile, most customers have one tracked contact. They called once and booked. For them, first and last are the same thing and the attribution question does not exist.

The question only exists for the customers with two or more tracked contacts before they booked. That group is usually small. It is also where last click quietly moves money around, because every one of them gets credited to the second surface and never the first.

A worked month

Take the same illustrative shop from the lead tracking post: 120 new recurring customers in May, 63 of them measured. Its paid search spend is $6,000, split $5,100 on non-brand terms ("termite inspection," "ant exterminator near me") and $900 on its brand campaign, the ads that show when someone searches the company's name.

Pull every tracked call and form for those 63 customers in the 14 days before each one booked. 46 had a single contact. 17 had two or more, and they fall into four paths:

  • 9 clicked a non-brand ad, then called the Google Business Profile tracking line and booked.
  • 4 called from a Local Services Ads listing, then searched the company name, clicked the brand ad and booked.
  • 3 clicked a non-brand ad, then came back through an organic search and filled out the web form.
  • 1 called the profile line, then booked through an LSA listing.

Now credit the 63 customers two ways. Both columns sum to 63, because each customer is counted once in each.

  • Local Services Ads: 24 on last contact, 27 on first
  • Google Business Profile: 18 on last, 10 on first
  • Paid search, non-brand: 6 on last, 18 on first
  • Paid search, brand: 8 on last, 4 on first
  • Organic web: 7 on last, 4 on first

Seventeen customers changed hands, which is 27% of the measured pile. Look where they went.

Last click says to cut the campaign that found the customers

Run cost per customer on the paid search line, split by campaign:

  • Non-brand, last contact: $5,100 for 6 customers, or $850 each.
  • Non-brand, first contact: $5,100 for 18 customers, or $283 each.
  • Brand, last contact: $900 for 8 customers, or $112.50 each.
  • Brand, first contact: $900 for 4 customers, or $225 each.

At a $115 quarterly ticket and a 55% gross margin, the top of the healthy range for a pest operation per NPMA industry data and PCO Bookkeepers guidance, an illustrative recurring account produces $253 of gross profit a year, or $63.25 a visit. At $850 a customer, a non-brand click pays for itself on the fourteenth visit, more than three years after the initial. At $283 it pays back on the fifth visit, about a year in. Same spend, same customers, same month.

The last-click report tells this owner that the brand campaign is the best ad in the account at $112.50 a customer and that non-brand is a disaster at $850. The obvious move is to cut non-brand and put the money into brand. That move would cut the campaign that made the first contact with 12 of the customers credited elsewhere, and feed the campaign that mostly answers people who already know the name.

The paid search line as a whole reads $429 per customer on last contact and $273 on first. The total moves. The split inside it moves a lot more, and the split is where budget gets reassigned.

Your brand campaign is a receptionist with an ad budget

Someone who types your company name into Google already decided to call you. Something else introduced them: a truck, a neighbor, a yard sign, or in four of the cases above, an LSA listing they called earlier that week. The brand ad answers the phone for that decision. It rarely makes it.

That does not mean you turn it off. If competitors bid on your name, a brand campaign keeps your own searchers from landing on someone else's ad, and at $900 a month that can be cheap insurance. Budget it that way: as rent on your own name, graded on whether you still own the top of your brand search. Never let its cost per customer become the bar every other campaign has to clear. It will clear it every month, because it is grading the customers your other marketing already produced. The acquisition cost post makes the same point from the other side: strip brand out before you compare anything.

The second-order cost shows up weeks later, on a different line

Suppose the owner acts on the last-click report and cuts non-brand. Last click predicts a loss of 6 customers a month.

The first-contact column says the exposure is larger. Nine of the profile's 18 credited customers and three of organic's seven had already reached the shop through a non-brand ad that month. No one can prove the ad caused the second call; at local volume that proof does not exist. What the records do show is that a dozen customers credited to free surfaces started with a paid one, which is enough to know the cut will not be free for those surfaces.

If those customers stop arriving, the drop shows up on the profile and organic lines, a few weeks after the cut, in the busiest part of the season. The owner reads it as "the map pack got worse" and goes looking for an SEO problem. The ad report, meanwhile, looks fine, because the campaign it flagged is gone.

First contact is not the answer either

First contact has its own blind spot. It credits the first contact you recorded, not the first time the homeowner heard of you. The ad they saw and did not click, the truck parked two doors down, the neighbor at the mailbox: none of those leave a record, so none of them ever show up as first.

So read the two columns as a range, and treat the gap between them as the finding. A surface whose first-contact count runs well above its last-contact count is introducing customers that something else closes. Here that is non-brand search, and to a lesser degree LSA. A surface whose last-contact count runs above its first-contact count is closing what something else started. Here that is the profile, the brand campaign and organic.

Neither kind is good or bad. Closers are worth having. What goes wrong is grading an introducer on the closer's column, because it will fail that test every month while it does its job.

The fractional models that split one customer across several touches, 40% to the first, 40% to the last, 20% to the middle, do not fix this. The weights are someone's opinion typed into a setting. A model that learns its own weights needs far more conversions than a pest shop produces; 17 multi-contact customers a month is not enough to fit anything. Two plain columns you can audit beat a weighted one you cannot.

The window is a decision, so make it on purpose

The 14 days above is a choice, and the choice changes the answer. Stretch the window to 90 days and this shop picks up six more two-contact customers, most of them people who called last summer about wasps and signed a quarterly plan this May through the profile line. Last summer's call did not produce this May's plan. It was a different problem and a different decision.

Set the window from your own records. For your two-contact customers, count the days between the first and last contact. Most pest decisions are made in days, because the problem is in the kitchen now, and you will usually see a tight cluster followed by a long thin tail. Cut the window where the tail starts, write the number down, and keep it fixed so this May can be compared with last May. A window you change every month is a dial for producing the answer you wanted.

Add up what every surface claims and you get more customers than you have

Each platform's own report credits every customer it touched. In the worked month, LSA touched 28 of the measured customers, paid search 26, the profile 19, and organic 7. That adds to 80, from a month that produced 63 measured customers.

Every surface that touched a two-contact customer claims all of that customer. Each claim is defensible on its own, and the sum is impossible. The check is the same one that runs through all of this: count each customer once per column, and make the column sum to the customer file. A source report that adds up to more customers than you signed is counting touches, and touches do not pay an invoice.

There is no benchmark for this worth having

I won't give you an industry figure for how many pest customers touch you twice, or how many days they take to book. The numbers that circulate do not transfer.

Your share of two-contact customers depends on how many surfaces you track separately (a shop with one tracking number for everything will show almost none), on your service mix (a bed bug or wasp call books the same day, a termite inspection or a mosquito program gets shopped), and on how clean the phone field is in your CRM. Most published path-length figures come from online retail, where a purchase can take dozens of visits over weeks. A homeowner with ants on the counter does not behave like someone researching a couch.

The number to watch is your own share and your own window, read each month against the same month last year. A rising share of two-contact customers is worth knowing about on its own: it usually means more of your customers are shopping you before they book, and that the first call is where you lose them.

How to run it on one closed month

  1. Start from the measured customers for one finished month: customers matched to a tracked call or form by exact phone or email.
  2. Pull every tracked contact for each one in the window before their start date. Use 14 days until your own records tell you otherwise.
  3. Tag paid search contacts brand or non-brand from the campaign or keyword your tracking platform recorded.
  4. Build two columns, first contact and last contact. Each customer appears once in each. Confirm both sum to the measured count.
  5. Compute cost per customer by surface and by campaign in both columns.
  6. Flag any line where the two columns differ by more than a third. Those are the budget decisions last click would get wrong.
  7. Before you cut an introducer, look at which closer's customers it touched first, and expect that line to move too.

Every step after the first depends on matching real customers to real calls and forms. That matching is what the Almanac does. It reads the customer report your CRM already emails, matches new customers to calls and forms deterministically on phone and email, and keeps measured and estimated in separate columns that sum to your actual customer count. It never claims to know which click drove a sale, because at local volume no one can, and it never touches your bids or connects to your bank.