Air Canada · Feb 2024
Its website chatbot told a customer about a refund rule the airline didn't have. A tribunal made the airline pay and rejected its claim that the chatbot was “a separate legal entity”.
AI consulting for marketing teams in Mexico
Don't hand your business to AI. Google's and Meta's ad automation and tools like ChatGPT get things wrong, and your company pays for it. We put an expert in charge: we decide where AI runs, check what it does and cut what doesn't pay.
Public cases
Every case has a date and a source. The last one is from our own account.
Its website chatbot told a customer about a refund rule the airline didn't have. A tribunal made the airline pay and rejected its claim that the chatbot was “a separate legal entity”.
Agreed to partly refund a government report it was paid AUD 440,000 for. The report had a made-up court quote and references to papers that don't exist.
Advertisers reported automated Advantage+ campaigns spending a day's budget within hours, with costs up to 10× higher. Meta said it fixed the problem.
In July, broad match became the default for new Search campaigns with Smart Bidding. In December, some accounts had keywords switched to broad match without being asked; Google said it was investigating.
In OpenAI's own test of questions about people, its o3 and o4-mini models made things up in 33% and 48% of answers.
In the Google Ads account of our martial arts academy in Mexico City, Google showed the search terms behind only 63 of 363 clicks in ten days. The ads Google wrote automatically had to be rewritten line by line.
Why it works
Four rules we apply in every engagement.
Ad platforms optimize for the conversions they're told to count. We set the goal from your margins and the leads that actually close.
No recommendations applied automatically and no machine-written ads going live unread. Every automation is tested on a test set before launch.
If a mistake is expensive or goes out in public, AI assists a person instead of deciding. It's a rule of our method.
Spend and results are reviewed against your own invoices and sales data, not the platform's report.
Performance marketing since 2012, for brands like Amazon, Walmart and Scotiabank. And we've trained more than 2,000 people in e-commerce and digital marketing at our school, LITS, and at Tec de Monterrey; former students lead e-commerce at Walmart, Elektra, Flexi, AXA and L'Oréal, among others.
Track record
~US$5M
in consulting and execution projects led, 2014–2022
10+
national and global brands served
6
countries in one Latin American SEO program
30+
people led across Latin America
2,000+
people trained in e-commerce and digital marketing, at LITS and Tec de Monterrey
Selected clients
Performance marketing, SEO, analytics and e-commerce work for national and global brands.
What we do
Four ways to work with us. Each one starts from what you already pay for and the metric it should move.
Every license, add-on, usage bill and pilot next to the metric it should move: cost per lead, conversion, hours saved. You get the cut, keep and fix list and the savings in pesos.
See the auditAn expert approves every change the platforms' AI makes or suggests, reviews search terms every week and checks ad spend against your own sales.
See oversightReporting pipelines, lead routing, feed and campaign automations, built with AI and running without it, with the monthly run cost stated before we start.
See automation buildsA four-hour executive AI workshop, or a four-session program where each marketer ships one working automation.
See AI trainingWhat the audit delivers
The spend map lists each AI and marketing-tech line with its cost, what it actually does and what to do with it.
| Line | MXN / month | Finding | Action | Saves / month |
|---|---|---|---|---|
| AI writing assistant, 40 seats | 20,000 | 14 seats in use | CutCut to 14 seats | 13,000 |
| Second ad-copy generator | 9,500 | Duplicates what the ad platforms already do | CutCancel | 9,500 |
| CRM AI add-on | 18,000 | No baseline for lead response time | FixSet the baseline, review in 90 days | — |
| Lead-routing pilot with an agency | 45,000 | No price per qualified lead | FixMove to a price per lead, or stop | — |
| Social listening, AI tier | 12,000 | One user, one report a month | CutDowngrade to the basic tier | 8,000 |
| Total | 104,500 | 30,500 |
Example with made-up figures. Your audit uses your own invoices and platform data.
The EDY method
List every AI and marketing-tech cost (seats, add-ons, usage bills, pilots, agencies) next to the metric it should move.
Run every line and every new idea through the scorecard. Each gets one verdict: cut, keep, fix, or build another way.
Build the lean version (rule, script, workflow or AI), with its monthly run cost stated before work starts.
Review spend, renewals and model prices every quarter, so the savings hold.
Free tool
When to use AI and when not to: score one task (reporting, content, ads, leads or service) on six questions and get one of four verdicts with the cost math, in about a minute.
How we work
No commission or referral fee from any AI or marketing-tech vendor. It's a clause in every proposal.
Prices in MXN plus IVA, invoiced with CFDI. Gain-share is an option on the audit, never a surprise.
You own the code, the accounts, the dashboards and the documentation. What we build runs without us.
Thirty minutes, no slides: what you pay for AI today, what it returns and what's up for renewal.