Where do we start if we don't trust any of our marketing numbers?
Start with the conversion signal, not the dashboard: confirm that each real outcome, like a purchase, qualified lead or booked call, is recorded once and in the right place before anyone touches bids or budgets.
The order matters. If you add negative keywords, shift budget or rewrite bid targets while a tag is missing or double counting, you are tuning the account to bad data. I've seen accounts cut keywords that looked like zero-conversion waste when the real problem was a conversion tag that never fired.
A check you can run today: place a test order or submit a test form, then confirm it shows up once in GA4, once in Google Ads, and once in your CRM or store. Then compare a week of totals across all three. Small gaps are normal. A gap of half, or a number that doubles, means the plumbing comes first. That is what our conversion tracking audit is for.
What is incrementality testing, and do we need it?
Incrementality testing measures how many conversions your ads actually caused by comparing a group that saw them with a similar group that didn't, and you need it once a channel takes enough budget that being wrong would hurt.
Attribution tells you which ad was in the path. It can't tell you whether that person would have bought anyway, which is why brand search and retargeting look so good in attribution reports. A holdout answers the harder question. Google Ads offers Conversion Lift studies split by users or by geography, though Google says it isn't available in every account and access goes through a Google representative. Meta offers lift studies too, and a simple version is pausing a channel in some regions and comparing.
Tests cost something: weeks of time and some lost sales in the group without ads. Save them for big calls, like doubling a channel or cutting one. Clean tracking comes first, or the result means nothing. Related reading: our attribution story.
Sources: Google Ads Help: About Conversion Lift, Google Ads Help: Set up Conversion Lift based on geography
Do we need marketing mix modeling?
Probably not yet, unless you spend across several channels, including offline ones, and have at least two years of weekly spend and sales data where spend actually varied.
Marketing mix modeling looks at spend and results in aggregate, week by week, and estimates what each channel contributed after accounting for seasonality and promotions. It uses no cookies or user-level data, which is why it is getting attention again. Google's open-source model, Meridian, recommends at least two years of weekly data for geo-level models and three years for national ones. If every channel always rose and fell together, the model can't tell them apart.
For most small and mid-sized advertisers, clean conversion tracking plus an occasional holdout test gives better answers for less effort. When MMM does fit, treat it as a budgeting tool and check it against a lift test, because different reasonable assumptions produce different answers. We wrote about Meta's take in Meta's MMM model API.
Sources: Google for Developers: An introduction to Meridian, Meridian: Collect and organize your data, google/meridian on GitHub
Which attribution model should we use?
Use data-driven attribution in Google Ads, which is the default for most conversion actions, and keep last click only if you have a specific reason; Google retired first click, linear, time decay and position-based models in 2023.
Data-driven attribution spreads credit across ad interactions based on your own account's conversion paths. Last click gives everything to the final click, which tends to over-credit brand search. The model also feeds Smart Bidding, so changing it changes how the account bids. After a switch, expect fractional conversions and a short dip in recent days while credit settles onto earlier dates.
The bigger point: no model tells you what caused a sale. Pick one, write it down, and stop switching models to make a channel look better. Each platform credits its own ads, so their totals will add up to more than your real sales. Decide which source settles budget arguments, usually your CRM or store, and treat platform numbers as directional. More on how this flows into bidding strategy.
Sources: Google Ads Help: About attribution models, Google Ads Help: First click, linear, time decay, and position-based attribution models are going away, Google Ads Help: Best practices for managing attribution model changes
What marketing numbers should we report to our CFO?
Report what finance already trusts: revenue or qualified pipeline, total marketing spend, and the ratio between them, with platform metrics as supporting detail rather than the headline.
Clicks, CTR and platform ROAS help the people running campaigns. A CFO wants to know what each marketing dollar returned in the company's own system of record. So we anchor reports to store or CRM data: new customers, cost to acquire one, the revenue or margin they bring, and the trend. Where we can, we also send value back to the ad platforms, like margin or customer lifetime value instead of raw revenue, so bidding chases the same number finance cares about.
One habit that builds trust fast: add a line to every report explaining the gap between platform-reported conversions and the CRM total. Finance stops arguing with marketing numbers once they see you reconcile them yourself. For a starting framework, read are you measuring what matters.
Can you track phone calls from our ads?
Yes. Google Ads can count calls from your ads and calls from your website as conversions, using a Google forwarding number that rings your real line and records which campaign drove the call.
There are two setups. Call reporting covers people who tap the call button on the ad itself. Website call tracking swaps the number on your site for a forwarding number when the visitor came from an ad, so you can see which keyword led to the call. Either way you set a minimum call length, so a 10-second wrong number doesn't count. Google says it can also use call recordings to judge lead quality when recording is on.
For service businesses, calls are often the main conversion and the one most often missing. If calls turn into booked jobs in a CRM, import those outcomes too, so bidding learns which calls became revenue. One warning from Google's docs: forwarding numbers belong to Google and can change, so never print one on a truck or flyer. Our CRM work connects the call to the sale.
Sources: Google Ads Help: About call reporting, Google Ads Help: Set up tracking calls to a phone number on a website, Google Ads Help: Measure calls from ads, Google Ads Help: Google forwarding number
