DigiWise DigiWiseGet my free audit

Methodology

22 checks, and the rules every figure must pass.

Every figure the audit measures is read from your own Google Ads account and has to pass the rules below. When the data cannot prove a line, the line is left out. Here is each check, each rule, what the AI does, and what the audit cannot see.

For the audit, our software only reads the account. About 30 seconds, no credit card.

  • 1,000+

    accounts audited

    audits by Attilio Curcio

  • $1,000s

    of waste found

    audit reports

  • 4.6/5

    on Trustpilot

    18 reviews

  • 20

    years of Google Ads

    Attilio Curcio, founder

01The 22 checks

What runs on your account.

Each check and what it reads, across 7 categories. This list comes from the same registry the audit engine runs, so this page cannot promise more than the audit does.

Conversion tracking 7 checks

  • C-12 Does the account count sales at all? Whether sales tracking exists and counted anything in the last 30 days - the check everything else depends on.
  • C-13 Tags that went quiet Enabled conversion actions that have recorded nothing for 30+ days while spend kept running.
  • C-14 Sales arriving after the counting window closes How long buyers really take from click to sale, joined to your action's counting window - sales slower than the window never get counted.
  • C-15 Placeholder revenue values Whether every "sale" is recorded at the same flat default value - revenue that is a setting, not reality.
  • C-16 Counting the same sale twice A Google Ads tag AND an Analytics import both set as primary for the same sale type.
  • C-17 Auto-tagging switched off Without it Google cannot connect ad clicks to what happens on your site afterwards.
  • C-18 Last-click-only credit Whether your main action still credits every sale to the last click only - Smart Bidding learns from that partial picture.

Structure 3 checks

  • C-05 Paying twice for your own name Performance Max overlapping your own brand searches - clicks you would probably have won anyway, with the detected brand terms shown.
  • C-09 Campaigns switched on but not running Campaigns that look active in the list but are not actually serving - with the reason.
  • C-11 Ads silently blocked or rotating wrong Enabled ads disapproved on policy (they just stop showing - nobody warns you) and ad groups showing weak ads as often as winners.

Account care 1 check

  • C-22 Is anyone minding this account? Who last touched the account (human vs automation), what kind of work it was, Google's own unread suggestions triaged into act vs ignore, and Google's own account grade.

Keywords and search terms 4 checks

  • C-01 Searches that spend and never sell Every search term that took money in the last 30 days without producing a single sale - listed with its cost and clicks.
  • C-02 Blocked searches in place Whether the account excludes junk searches at all (Google calls these negative keywords), and how many are in place.
  • C-03 Broad-match bleed How much of the zero-sale spend came in through loose keyword matching.
  • C-04 Overpaying for the same click Keywords where a weak quality component makes Google charge you more than a competitor for the SAME click - with the weak part named and a fix hint.

Campaign settings 3 checks

  • C-06 Search Partners switched on Campaigns buying traffic on non-Google search sites - with the measured spend, clicks and sales it actually produced last month.
  • C-07 Display mixed into Search Search campaigns also buying banner placements - the textbook silent-waste setting, measured in dollars.
  • C-08 Targeting by "interest" instead of presence Campaigns reaching people merely interested in your area - tourists and out-of-region lookers included.

Budgets and visibility 3 checks

  • C-10 Winners capped by budget Campaigns that produce sales but hit their budget ceiling - with Google's own estimate of what a raise would add.
  • C-19 Visibility lost to budget How often your ads simply did not show because the budget ran out - per campaign.
  • C-20 Visibility lost to ad rank How often competitors outranked you out of the results - per campaign, with the click share you captured.

Schedule and devices 1 check

  • C-21 Hours and devices that buy nothing The hour-by-hour grid of spend vs sales in your account's timezone - the specific hours that spent real money for zero sales - plus cost-per-sale gaps between mobile, desktop and tablet.

Where the 22 come from: the checklist Attilio Curcio used by hand on full account audits. It has 55 questions to answer yourself, in your own account. The free audit automates 22 checks drawn from it, on your real data. Read the checklist.

02Honest figures

The rules a figure must pass.

An amount in the report has to pass every rule below. The report says what it measured, labels the one amount it estimates, and leaves out what it cannot prove.

Money only where it is measured
An amount appears only where your own numbers measure it: what searches with no recorded sale cost in the last 30 days, and the Search Partners and Display spend Google reports. Google's own budget estimate is named as Google's.
One estimated amount, labeled
What Performance Max may spend on clicks from people already searching your name. Google does not report it, so the report applies a rule of thumb to the measured Performance Max spend, calls it a rough estimate and shows how it is computed.
Never added up
There is no grand total of the fixes. Amounts of different kinds are never summed into one big number, and the estimate is never added to anything.
Broken tracking comes first
When the account does not count its sales correctly (no sales tracking, no sale recorded in the last 30 days, a main sales action silent for 30 days or more, or the same sale counted twice), the report shows no waste figure at all. Its first move is to fix the tracking.
A trend only when the volume holds it
A change in your cost per sale or per lead gets a percentage only when each of the two 30-day windows compared holds at least 30 sales and the change is larger than chance could explain (a Poisson test). Short of that, the report says it in words only: trending up, trending down or steady. Under 10 sales in a window, no sentence at all. The cost per sale of the last two counted months shows side by side once each month holds 10 sales.
The last 7 days left out of trends
Google keeps counting sales for days after the click, so the most recent days of any trend look worse than they are. Every trend in the report leaves out the most recent 7 days.
Unknown is never zero
When the data cannot prove a line, the line is left out and the report says what it could not read. A change history Google did not send reads as unknown, never as no changes.
Your figures, from your account
Every figure the report measures comes from your account, over real date windows, in its own currency. Nothing is typed in by hand. A comparison from outside your account says where it comes from: a range for businesses like yours, when the report shows one, is labeled a rough estimate from our past audits.
Sample data C-05

Paying twice for your own name

Rough estimate: about $821 a month may go to clicks from people already searching your name - 'PMax - Lead Gen' overlaps your own brand searches.

How we compute this: 12% of the $6,840 your Performance Max campaigns spent in the last 30 days. Google does not report which of those clicks were searches for your name, so this is a rule of thumb, not a measurement. It is never added to any other figure.

The report's own sentences, on the invented sample account.

Read the whole sample report: an invented account, every section.

03DigiWise AI

It measures first. Then it explains.

A model asked what is wrong with an account answers confidently either way. So ours is never asked. The 22 checks measure, and a model only puts what they found into plain English.

  1. 01

    It measures

    The 22 checks run as code on your account's data, read through the official Google Ads API. The same data always gives the same finding. No model is involved at this step.

  2. 02

    It explains

    Then our AI writes the short plain-English summary on your dashboard, from those findings. It is given your account's own figures and told to quote no other amount: never to invent, add up or annualize one.

  3. 03

    It is checked

    Before the summary is shown, every amount in it is compared with your account's figures. If one is not among them, a summary written without AI, from the same findings, is shown instead.

Google advises you. We check.

Google leaves suggested changes in every account, and many of them ask you to spend more. Check 22 sorts each one: worth applying, or mostly serving Google. A suggestion it cannot explain in one plain sentence is never marked worth applying. Google sells the ads; we do not.

Worth applying

  • Better ads and more ad text
  • Sitelinks, callouts and your phone number
  • Budget you already committed, moved to where it works

Mostly serving Google

  • Budget raises
  • Looser keyword matching
  • Search Partners and Display
  • Automated bidding and Performance Max, switched on

Sometimes right, never to apply blindly.

Sample data C-22

Is anyone minding this account?

Google suggests 9 changes to this account. Worth applying: 2. The other 7 mostly serve Google.

Worked out by the report itself, on the invented sample account. Yours comes from your own account.

See it in the sample report, in the account pulse.

Who acts on what it finds

  • See it free

    Free AI Audit

    For the audit, our software only reads the account. The audit finds the leaks, explains them and puts the most urgent first. What you change, and when, is up to you.

    See the free audit

  • Hand it to our team

    Managed by our team

    DigiWise AI re-reads the account every night and ranks what it finds. Attilio and the team he leads decide what is worth changing and apply it.

    $997/mo per account, up to $10,000/mo of ad spend.

    See the managed plan

  • Let AI prepareFree to prepare

    AI Autopilot

    Today, after your consent, it prepares the changes it would make on your Search campaigns, each with its reason. You apply the ones you want, in Google Ads. Automatic changes are not open yet: we will tell you before they open, and they will only start if you turn them on yourself, within your limits.

    See AI Autopilot

AI Autopilot prepares changes and applies none: a person makes every change in the account, you or the team on a managed plan. What the model receives, and what it keeps, is on the security page.

04The other half of honesty

What the audit cannot see, or do.

A tool that claims to see everything is lying about something.

Who outbids you, search by search
Google does not share it. The report shows how often your ads did not show because of the budget or because of rank, campaign by campaign (checks 19 and 20).
Change history beyond 28 days
Google's API only returns the last 30 days of it, and the audit reads 28 of them. Anything changed before that is out of its sight.
Anything outside Google Ads
Your CRM, your margins, your site analytics. The audit reads one system deeply rather than guessing across many.
Your product feed and Local Services Ads
The audit does not read your Merchant Center feed or your Local Services Ads (Google Guaranteed).
Any change to your account
Google's API has no permission limited to reading, so its consent screen looks broader than what we do. For the audit, our software only reads the account: no code in the audit can change, pause or spend anything, and you can revoke access in one click. On a managed plan you invite us as a manager, on purpose, and every change is logged in your monthly report.

05Free AI Audit

Run the 22 checks on your own account.

The full audit, free, in about 30 seconds: every leak, the cause and the fix for each one, with what it costs where we can measure it.

For the audit, our software only reads the account. No credit card, revoke anytime.

4.6/5 18 reviews (opens in a new tab)