Mohamed Wagdy Bondok

Marketing usually breaks in the handoffs. I work on the handoffs.

I run marketing as one system, from the first touch to the right team.

Send a lead of your own and watch where it goes:

Expos and events Paid search Paid social Listing platforms Referral partners

Explore the map with the keyboard

What the spend returned, by kind of product.

High-ticket products: 20x+ return on ad spend; Long, considered purchases sold by a sales team. Measured to the closed sale, over one year.
Education and services: 5.6x revenue to acquisition cost; Drasco Educational Services, which I co-founded. What each enrolment brought in against what it cost to win, across three to four years of paid media.
DTC e-commerce: Launch in progress; A direct-to-consumer brand I'm helping launch. No results yet.

Three calls I made, and one I'd make differently.

I let cost per lead rise about 8x.
I wrote to universities one by one, the slow way, and let campaigns recruit the agencies.
I stopped trying to get the signal out of WhatsApp and sent Meta the qualified leads instead.
A call I'd make differentlyI trusted Click-to-WhatsApp to tell Meta which chats became leads, because the setup matched the documentation. It never did. Now I prove a signal arrives end to end before any budget depends on it.

Strategy

One brand message, or a different argument for each buyer?

I chose different arguments.

The band above the map is the planning: what each audience is told, and what a customer is worth before any money is spent. The brand system is mine. I model the budget and make the case for it to the people who sign.

Demand generation

Where does a buyer first hear the name?

At a physical event or in a search result, and I judge both by what they do further down the map.

Search and social ads, placements with international listing platforms, partner agencies for markets the ads can't reach, and international magazines read by high-net-worth buyers, including issues carried on Emirates flights, which we place with the team. Physical channels wear their own shapes on the map.

A call I made I let cost per lead rise about 8x. Full reasoning on letting cost per lead rise

Partners and channels

Some buyers are easier to reach through someone they already trust.

Marketing recruits partner agencies for markets that direct acquisition doesn't reach, through LinkedIn and business-targeted campaigns. I set the terms on those deals, and a colleague fronts the relationship day to day.

The expo organiser and the listing platforms are partners too, and I run those negotiations. At Drasco I built the partner channel two ways. For TAGA I write the presentations and proposals, among them an outdoor media proposal for an international prospect.

A call I made I wrote to universities one by one, the slow way, and let campaigns recruit the agencies. Full reasoning on writing to universities one by one

Capture and qualify

Most of the middle is finding out which leads are real.

Before a salesperson spends an afternoon on them.

Pages I built, WhatsApp, a voice agent and the phone line all arrive through one intake path. Each lead's source is locked there, the click behind it is traced back to the campaign, ad and keyword overnight, and AI agents ask the qualifying questions.

Routing to teams

A market wasn't converting. I traced it to the speed of the first call.

It looked like a creative problem. The cause was good leads waiting too long for that call.

Routing is the hub. It sends each lead to the team it belongs with: lead qualification, sales, customer care, the partner channel, account growth, or suppliers and HR. Then it finds someone who speaks the lead's language. A lead turns hot when it reaches the right team.

How it works: routing and the phone line

Underneath all of it

Under the funnel there's a second map.

Tracking, data, automation and reports. If any of it is wrong, every decision above it is made on a wrong number, and nothing raises an error.

I built the measurement layer with AI help: server-side Tag Manager, Google Analytics, the Meta Conversions API and the Zapier feed that sends qualified leads back to Google Ads. Under it all sits the CRM, which I designed and built using AI, with its reporting on top.

A call I'd make differently I trusted Click-to-WhatsApp to tell Meta which chats became leads, because the setup matched the documentation. It never did. Now I prove a signal arrives end to end before any budget depends on it. Full reasoning on trusting Click-to-WhatsApp

How it works: tracking, the CRM, AI

The loop

Outcomes travel back up and change what happens at the top.

A lead that reaches its team lands on one timeline in the CRM. For a lead that came from an ad, what happened next goes back to the ad platforms, so bidding learns from qualified pipeline instead of form fills.

A call I made I stopped trying to get the signal out of WhatsApp and sent Meta the qualified leads instead. Full reasoning on sending Meta the qualified leads

People

Marketing sits in the middle of many teams. The handoffs between them are the work.

I manage the call-centre team. Sales, operations and leadership don't report to me, and most of what I build only works if they use it.

Teams I've led: my own team at Drasco, first in an office and then remotely from Cyprus, and the call-centre team I manage now. Teams I work across: the teams where leads land; leadership, through decision papers and monthly reports; a copy and SEO colleague, with designers and developers when the work needs them; agencies and vendors for events, media, listings and print; partner universities and education agencies at Drasco; and TAGA's leadership, on its move from traditional to digital. On the map, a ring marks a team I've led and a diamond marks where a team I work across meets the work.

How it works: who signs, and who does the work

Ventures and consulting

I learned budgeting with my own money.

At Drasco Educational Services, the study-abroad company I co-founded, every bit of acquisition spend was ours, and we were profitable from day one.

I grew its web app to 3,000+ users in six months and turned them into 150+ enrolments, with revenue about 5.6x acquisition cost. Sign-ups became consultations and consultations became enrolments, fed by paid media, webinars and a partner channel I built.

Two businesses I consult for.

TAGA is a family business in outdoor advertising in Saudi Arabia. I handled its move from traditional to digital, and built its website, taga.com.sa, in English and Arabic.

For a direct-to-consumer e-commerce brand, the launch is in progress. The unit economics come before any ad: one hero product, with campaigns run by rules written before launch. Neither is wired into the engine above, because both are separate businesses.

The map, as a list

Every point on the map, stage by stage, with what I do there.

Strategy

Positioning & messaging. I set the angle each audience is approached on, and what the brand’s authority rests on. Shapes Demand generation, through the Brand system and Paid media.
North-star metrics. Before any spend, I agree the measure that counts: cost per qualified lead or cost per sale, rarely cost per click. Cost per lead went up on purpose once the measure became qualified leads instead of form fills. Fed by Reporting. Sets the bar for every stage below. How I decide
Budget and channel mix. I model the budget and the channel mix, and make the case to the people who approve spend. Each channel’s share is argued on cost per qualified lead and cost per sale, the measures I set up where none existed. Shapes Demand generation. Fed by Reporting and Business cases.
Break-even targets. Before a campaign starts, I set the most a qualified lead or a sale can cost while the spend still pays for itself. At my own company I weighed acquisition cost against customer value from the first campaign, and we were profitable from day one. Shapes Budget and channel mix, and the DTC E-Commerce launch.
Partner market entry. For markets our ads can’t reach, I set how we recruit partners and the terms of those deals, and I built how their leads are tagged, scored and routed. Partner agencies recruited through LinkedIn and business-targeted campaigns, with scoring and routing already running in the map below. Shapes Referral partners. Built on Source attribution, Lead scoring and Lead routing.
Five-language funnel. Buyers meet the funnel in their own language, from the first page to the person who calls them back. The group site runs in five languages, and the phone menu greets callers in theirs before routing finds a rep who speaks it. Shapes Websites, Landing pages, Inbound calls and Lead routing.

Demand generation

Paid search. I run search campaigns on Google Search and Performance Max, plus Yandex, and judge them on qualified leads. A search click lands on a page in the buyer’s language, and the keyword behind it is attached to the lead overnight. Feeds Landing pages. Gets outcomes back from Offline conversions. A call I made here
Paid social and video. I run Meta and LinkedIn campaigns, and YouTube through Demand Gen. A Meta ad that opens a WhatsApp chat, where an AI agent asks the first questions before a person picks up. Feeds Landing pages and WhatsApp. Learns from Offline conversions and the Meta pixel and CAPI.
Expos and events. I plan and run our events abroad with the expo organiser, from negotiating the package to how leads from the stand reach the intake. A visitor who stops at the stand goes into the same intake as an ad lead, tagged to the event so it can be judged on what it produced. Feeds Landing pages and the event forms.
Print and inflight. With the team, we place the brand in international magazines read by high-net-worth buyers, including issues carried on Emirates flights. A reader on a long flight sees the spread, and we follow the reply through the same intake as every other lead. Feeds Websites, WhatsApp and Inbound calls.
Brand system. The brand system is mine: how each brand in the group presents itself and what its authority rests on. Copy is drafted with AI in several languages, to the direction I set. Shapes Paid media, Websites and Landing pages. Set by Positioning & messaging.
Content, SEO & GEO. Content and SEO are a colleague’s craft. My part is measuring what that content earns, in search and in AI answer engines (GEO and AIO). Blog pieces that AI engines were already citing showed up in the readings, and those readings now decide which questions the content answers next. Feeds Websites. Fed by AI search visibility.
Media & listing partners. I run our placement deals with international listing platforms and media vendors, and negotiate the terms. Each platform is judged on the qualified leads it sends rather than on its traffic. Feeds Inbound calls and WhatsApp. Sits beside Referral partners.
Referral partners. Marketing recruits partner agencies for markets direct acquisition doesn’t reach, through LinkedIn and business-targeted campaigns. I set the terms on those deals. An agency that already sells to buyers in a market our ads can’t reach sends its client to us, and a colleague fronts that relationship day to day. Feeds Unified lead intake. Ends with the Partner channel team.

Capture and qualify

Landing pages. Pages for individual projects, with short forms in the buyer’s own language. An ad landing page and an event form both post into the same intake, each tagged to where the lead came from. Fed by Paid media, and by Expos and events. Feeds Unified lead intake. Tagged through Client-side tagging.
Websites. Websites and landing pages: a multi-brand group site in five languages, sub-brand sites, and landing pages for individual projects, built and shipped myself with AI writing the code. Feeds Unified lead intake, through forms and the Voice agent. Tagged through Client-side tagging.
Voice agent. I designed and shipped a voice agent on a real-time voice model for a company website, with AI writing the code. It quotes live prices from the CRM and takes a lead by voice. It can’t save a lead until it has read the phone number back and heard a yes, because one misheard digit means a buyer nobody calls. Lives on Websites. Feeds Unified lead intake. Run by AI agents.
WhatsApp. WhatsApp is where many buyers start. AI agents I designed ask the first questions before a person joins the thread. WhatsApp voice calls ring the same phone menu as a normal call, so a call from the chat app isn’t lost. Fed by Paid social, Media & listing partners, and Print. Feeds Unified lead intake. Run by AI agents.
Inbound calls. I chose the company’s hosted PBX, specified it, and wrote the phone menu that greets callers in their own language. A first-time caller becomes a new lead automatically, and the call lands on that lead’s timeline before anyone routes it. Fed by Media & listing partners, and Print. Feeds Unified lead intake. Writes every call to the CRM.
Unified lead intake. Every channel arrives through one path: ad forms, web and event forms, WhatsApp, calls and manual entry. Someone who fills a form and then calls the next day is one lead, matched on phone or email. Fed by every entry point. Feeds Source attribution. Writes to the CRM.
Source attribution. Each lead gets its source locked at intake, then filled in overnight. Google click IDs are resolved back to the campaign, ad group, ad and keyword every night, and AI sorts call transcripts and chats into lead or not, keeping unclear cases as leads. Fed by Unified lead intake. Feeds Lead scoring.
Lead scoring. AI agents ask the qualifying questions, and the lead score says how much of it is actually known. When few signals are known, the score says so, and a thin lead stays in play until someone asks the missing questions. Fed by Source attribution and AI agents. Feeds Lead routing, or Nurture when a lead isn’t ready.
Nurture. Leads that aren’t ready are flagged for nurture in the CRM, and leads marked lost move to a nurture board automatically. A separate WhatsApp agent is set aside for warm re-engagement. Fed by Lead scoring and Lead routing. Linked back to WhatsApp, and to Lead routing when a lead warms up.

Routing to teams

Lead routing. Lead routing is the hub. A quality score decides whether a lead goes straight to sales or to lead qualification first, then it goes to someone who speaks its language. A lead nobody touches moves to another rep on a short timer, inside business hours, unless the first rep actually called. Fed by Lead scoring. Feeds every team, and Nurture for lost leads.
Speed to lead. A first-contact rule measured in minutes, with the clock starting when routing hands a lead over. If a new lead waits past its limit, the system tags the breach and alerts a colleague, during working hours only. Times the handoff from Lead routing to Sales and Lead qualification.
Sales playbook. I co-wrote the sales protocol both teams work to: how fast a new lead gets its first call, and the follow-up cadence and qualifying questions after it. Sets the rules that Speed to lead times, for Sales and Lead qualification.
Lead qualification. The call-centre team I manage phones fresh leads first, qualifies them and passes the ready ones to sales. After each call, an AI summary arrives as a suggestion the rep can apply or ignore. Fed by Lead routing: the quality score sends a lead here, to a rep who speaks its language. Writes outcomes to the CRM.
Sales. Sales-ready leads go straight to a rep who speaks the buyer’s language, with the full history attached. The sale counts when the deposit lands, because that is when the business gets paid, and that outcome goes back to the ads. Fed by Lead routing, by quality score, then to a rep who speaks the lead’s language. Feeds the loop: CRM outcomes, Integrations, Offline conversions, Paid media.
Customer care. Existing buyers’ questions go to the post-sale team, not back into the new-sales queue. Fed by Lead routing.
Partner channel. Agencies and brokers who want to sell for us are split from buyers at intake and handled on partner terms. An agency’s client is handled with that agency, so the partner sees its referral land and gets credit for it. Fed by Lead routing, from Referral partners, on terms I set in the partner deals.
Account growth. Existing owners looking to resell go to the team that handles them, so new buyers and current owners never share a queue. Fed by Lead routing.
Suppliers & HR. Not everyone who writes in is buying: supplier pitches go to purchasing and job seekers to HR, so no rep spends a call on them. Fed by Lead routing.

MarTech and data

Tracking audits. Before I add a tag, I audit what is already firing and retire what doesn’t earn its place. One audit caught a Tag Manager trigger rule inflating pixel volume three to four times. Other finds are quieter, like an old container nothing loads any more. Checks Client-side tagging, Google Analytics and the Conversions API.
Consent gating. Tags that need consent wait for it. A tag that can’t yet be gated properly comes off the site until it can. Runs in Tag Manager.
Client-side tagging. I run the Google Tag Manager web container on the sites: which tags fire, on which trigger, and only with consent. A trigger rule that matched far more pages than intended is the kind of thing I look for first, because it inflates every number downstream. Fed by Websites and Landing pages. Feeds Server-side tagging.
Server-side tagging. I built a server-side Tag Manager container, with AI help, so events reach analytics and Meta from our server as well as the browser. Fed by Client-side tagging. Feeds Google Analytics and the Conversions API.
Google Analytics. Server-side Google Analytics, built by me with AI help, where site behaviour is read before it reaches a report. Duplicate streams counting the same visit get merged before anyone trusts the traffic line. Fed by Server-side tagging. Feeds Reporting and the CRM.
Meta pixel and CAPI. The pixel in the browser and the Conversions API from the server, pointed at the right pixel and checked live. A purchase event fires only when a sale is confirmed, so Meta never learns from a lead that didn’t buy. Fed by Server-side tagging. Feeds Paid social.
Offline conversions. Qualified leads and sales go back to the ad platforms: to Google Ads through Zapier, and to Meta as offline conversions matched on hashed contact details. Bidding learns from the leads sales actually qualified, instead of from whoever filled a form. Fed by CRM outcomes through Integrations. Feeds Paid search and Paid social. This is the loop. A call I made here
CRM. I designed and built it with AI, around one question: which campaign produced this sale. Reporting runs on it, and it replaced the licensed CRM the company used before. First touch is locked at intake, so a later edit can’t rewrite where a sale came from, and a conversion the ad platform didn’t report shows as missing, never as zero. Fed by the whole map. Feeds Reporting, Lead routing and Integrations. How the CRM works
Lifecycle stages. Each lead moves through named stages from new lead to sale, so a rep and an AI agent read its status the same way. The AI agents may set only their allowed stages, so a confident wrong answer has nowhere to send a buyer. Kept in the CRM pipeline and on the WhatsApp platform. Read by Lead routing and Reporting.
Integrations. Zapier and scheduled API feeds move data between the ad platforms, analytics, the phone system and the CRM without anyone copying it. If a scheduled feed stops running, it shows up on an admin page instead of going quietly stale. Carries CRM outcomes to Offline conversions, and links the ad platforms, analytics and the phone system to the CRM.
AI agents. I designed the production AI agents on WhatsApp, a router and a qualifier among them, and the voice agent on the website. I wrote their prompts and the handoff rules between them. Runs WhatsApp, the Voice agent and Lead scoring. Writes to the CRM.
AI with guardrails. AI reads and drafts, and written rules decide what it may change. AI writes the weekly report narrative and may tidy its wording, but it can’t change a number. Call summaries are proposals the rep applies or ignores. Inside Reporting, brand copy and the after-call notes in Lead qualification.
AI search visibility. On a schedule, the CRM asks ChatGPT, Claude, Gemini, Perplexity and other AI assistants the questions buyers ask, and records whether the brand is named and which sites are cited. A failed search is recorded apart from “not cited”, and an empty sample is marked as no data, so the trend can’t invent bad news. Feeds Content, SEO & GEO, and Reporting.
Reporting. Each month has a plan, a report, plan against actual and the trend, with cost per lead, per qualified lead and per sale on screen. Call reports by language and by department, worked out from the route each call took through the phone menu, since no field recorded either. Fed by the CRM and Google Analytics. Feeds North-star metrics and Business cases.
Business cases. When something needs a decision, I write a short business case for leadership: what, why and when. A direct-versus-agency channel analysis, written so the choice between the two could be made on commercial grounds. Fed by Reporting. Feeds Budget and channel mix.
Result validation. When a number or an analysis comes back, I check whether the result is real before anyone acts on it. A search-data slice came back sorted by clicks, which hid exactly the queries the tool was built to find. I corrected the sample before trusting the output. Fed by Reporting. Feeds North-star metrics.

Who I work with

Leadership. Spend is approved above me, so a budget change I want starts as a short business case to the people who sign. Performance goes up the same way, in monthly reports. Between the Strategy band and the channels the budget pays for.
Copy, SEO and design. The function is small and split by craft. SEO and copy sit with a colleague, and designers and developers join when the work needs them. Paid acquisition, measurement, the systems and the brand direction are mine. We wrote the sales protocol together. On the handoff from Content and the Brand system into Websites.
Vendors. I run the negotiations with the expo organiser and our listing and media partners. Print placements we arrange with the team, and the events I plan with the expo organiser, all feed the same intake as the ads. On the handoffs from Media & listing partners, Expos and events, and Print into the capture stage.

Ventures and consulting

Drasco Educational Services. The study-abroad company I co-founded, where I owned the whole funnel. Its web app is the EdTech side: students sign up there before they book a consultation. I grew the web app to 3,000+ users in six months and turned them into 150+ enrolments, and the company was profitable from day one. It’s mostly quiet now. Its own company, not wired into the engine. More on Drasco
Drasco Educational Services: Paid acquisition. Every bit of acquisition spend was ours, over three to four years of paid media. The web app grew to 3,000+ users in six months and turned them into 150+ enrolments. Fills the web app, where Webinars and the consultations take over.
Drasco Educational Services: Revenue 5.6× CAC. From the first campaign I weighed what an enrolment cost to win against what it brought in. Revenue came to about 5.6x acquisition cost, and the company was profitable from day one.
Drasco Educational Services: University and agency partners. I built the partner channel two ways. Direct outreach to universities from a list I built myself, and paid campaigns that recruited education agencies as referral partners. A second way in for students, beside paid media. A call I made here
Drasco Educational Services: Webinars. I ran webinars that turned a student’s interest into a booked consultation. Between paid media and the consultation.
Drasco Educational Services: Funnel operations. I set up the path from a sign-up on the web app to a consultation, and from the consultation to an enrolment, and ran the operations behind it.
Drasco Educational Services: Team leadership. I led the team in an office, and later managed it remotely from Cyprus.
TAGA. I consult for TAGA, a family business in outdoor advertising, on its marketing, and I handled its move from traditional to digital. I built its website in English and Arabic, made it readable to AI engines, and wrote an outdoor media proposal for an international prospect. Its own business, not wired into the engine. More on TAGA
TAGA: Going digital. I handled the family business's move from traditional to digital, in English and Arabic, working with its leadership.
TAGA: Website. I built and shipped taga.com.sa in English and Arabic, from a site generator made for it with AI writing the code. More on TAGA
TAGA: AI-readable site. The site ships an llms.txt and a clean structure, so AI engines can read who the company is without guessing. Built for the same AI engines that AI search visibility tracks, on a smaller site.
TAGA: Media proposal. For TAGA I write the presentations and proposals, among them an outdoor media proposal for an international prospect. The proposal set out outdoor and digital billboard packages for the prospect’s campaign.
DTC E-Commerce. I help a direct-to-consumer brand run its digital marketing on Shopify, starting with its launch. The numbers decide what a sale can afford before any ad runs, and every campaign points at one hero product. Its own business, not wired into the engine.
DTC E-Commerce: Break-even CPA. Before the first ad, I worked out what a sale can afford to pay for itself. An item whose margin can’t carry its own ads sells only as an add-on to the hero product.
DTC E-Commerce: One hero product, with bundle logic. One campaign behind one hero product, so a small budget isn’t spread thin. Everything else in the store rides along as a bundle or an add-on.
DTC E-Commerce: Rules-based campaigns. Campaigns follow written rules, from the point an ad is switched off to how fast budget may follow a winner. The kill rules are written before launch, so nobody argues them after a bad week.

Where a lead goes

From an expo or event

You stopped at our stand at an event abroad.
Your details went in on the event form.
The event form joins the same intake path as every lead from an ad.
Your source is locked in: this lead came from an event.
A few known answers, and the score says how much is known.
Routing reads what you asked about and picks the team.
You reach sales, with someone who speaks your language. On the CRM timeline, start to finish.

From paid search

You searched, and clicked an ad.
A page in your language, with a short form.
Your form joins the same intake path as every other lead.
The campaign, ad and keyword behind your click are attached overnight.
Your answers set the score, thin or full.
Routed to sales, to someone who speaks your language.
The first-contact clock starts. If nobody reaches you, another rep gets you.
You reach sales.
Your outcome goes back to the ad platforms, so they learn who buys.

From paid social

From a listing platform

You found us on an international listing platform and called.
The phone menu greets you in your language, and your call lands on your record before anyone routes it.
Your call joins the same intake path as every other lead.
Tagged to the listing platform, so it is judged on qualified leads like you.
Your call is scored like every other lead.
Routing reads what you asked about and picks the team.
You reach sales. Next time you call, the rep sees your name before picking up.

From a referral partner

An agency you already work with sent you.
You join the same intake path, tagged to the partner.
The partner stays attached to your record.
Scored, with the partner still attached.
Routed to the partner channel team.
Handled with the colleague who looks after partner agencies, and credited to the partner.

Close details

How each part works

At a multi-brand group selling high-ticket products, in Cyprus, I’ve built or run:

Acquisition
- Paid search and social, several markets
- Physical events abroad
- International listing platforms
- Vendor negotiations
- Print and inflight, with the team
- Partner-agency recruitment through LinkedIn and business-targeted campaigns
Web
- Group site in five languages
- A site for each sub-brand
- Project landing pages
Systems
- Server-side tracking and the Conversions API
- Offline-conversion feed so bidding learns from qualified leads
- The CRM and its reporting, designed and built using AI
- Phone system on a hosted PBX, linked to the CRM
AI
- Production AI agents on the WhatsApp platform
- Voice agent on a real-time voice model
- Scheduled checks of how AI answer engines describe the brand
People
- Sales protocol, co-written
- Managing the call-centre team

Routing to teams

What happens in the minutes after a lead arrives?

A first-contact rule measured in minutes, and a structured follow-up cadence.

Read moreClose details

The fix was a first-contact rule measured in minutes and a structured follow-up cadence, written into a sales protocol I co-authored. Sales is a separate team I don’t run.

A quality score sends each lead to the right team, then to someone who speaks the lead’s language, and an untouched lead moves to another rep on a short timer unless the first one actually called.

I chose our hosted PBX, specified it, wrote the phone menu that greets callers in their own language, and linked it to the CRM. Every call lands on the lead’s timeline, and outbound calls show the number customers already know.

Underneath all of it

What gets checked before anything new is added?

I audit what’s already installed and retire what doesn’t earn its place.

Read moreClose details

Before I add anything, I audit what’s already installed and retire what doesn’t earn its place. One audit caught a tag rule that inflated pixel volume three to four times.

Consent gates the tags, and the offline-conversion feed sends qualified leads from the CRM back to Google Ads, so bidding learns from qualified pipeline instead of form fills. Reporting runs monthly, plan against actual, with cost per lead, per qualified lead and per sale on screen.

The CRM

Which campaign produced this sale?

I designed the CRM around that question, and built it using AI.

Read moreClose details

I designed the CRM and its reporting around that question, and built it using AI. It replaced the licensed CRM the company used before.

The rules it keeps:

First touch is captured at intake and locked, so a later edit can’t rewrite where a sale came from.
One intake path for every channel, deduplicated on phone or email.
The lead score shows how many of its signals are actually known, so a thin lead reads as thin rather than bad.
A conversion the ad platform didn’t report is labelled missing, never filled in as zero.
A deposit closes the sale, because that is when the business gets paid.

The phone link is deliberately one-way: calls flow into the record, and the phone system stays the phone.

I defined the problem, designed the system and directed the AI that wrote the code. The architecture and the release calls are mine.

A voice agent that mishears one digit loses the buyer.

The server refuses to save a lead until the agent has read the number back.

Read moreClose details

Nobody sees the error. The lead just never gets a call. When I built a voice agent on a real-time voice model for a company website, I made the server refuse to save a lead until the agent has read the number back and heard a yes.

Most of my AI work has that shape: AI reads and drafts, and fixed rules decide what it may change.

On the WhatsApp platform I designed production AI agents, a router and a qualifier among them, each limited to the lifecycle stages it’s allowed to set.
An AI model reads call transcripts and proposes a summary; the rep applies it or doesn’t. Lead temperature and close reasons stay human decisions.
AI writes the weekly report narrative and may tidy its wording. It is never allowed to change a number.
Copy across languages is drafted with AI and decided by me. The machine gets the first draft and never the last word.

The other half of AI is being in its answers.

On a schedule, our CRM asks ChatGPT, Claude, Perplexity and other AI assistants the questions buyers actually ask and records whether the brand is named and which sites are cited; Copilot comes in through Microsoft Clarity. A failed search is recorded separately from “not cited”, and an empty sample reads as no data, never as zero.

I practise it on my own work. taga.com.sa ships an llms.txt, and so does this site, with structured data that says who I am.

People

Who signs, and who does the work?

Spend is approved above me, so every budget change starts as a case I make.

Read moreClose details

Spend is approved above me, so every budget change I want starts as a case I make to the people who sign. The marketing function is small and split by craft: SEO and copy sit with a colleague, and paid acquisition, measurement, the systems and the brand direction are mine. We wrote the sales protocol together.

Outside the company there are the vendors, the listing platforms and the expo organiser, whose negotiations I run, and the partner agencies, whose terms I set.

At Drasco I led the team in an office, then from Cyprus for a year.

On a small team I run on the four Ds: delegate it, delay it, delete it, or do it myself. Deciding which, every day, is most of the job.

A mechanical engineer who ended up running marketing.

2019Co-founded Drasco Educational Services.
2019BSc Mechanical Engineering, Eastern Mediterranean University.
2020MSc Engineering Business Management, University of Warwick (WMG).
2022MBA, Cyprus International University.
2026Released martech-verify, open source.
ongoingConsulting for TAGA, a family business in outdoor advertising, and for a DTC e-commerce brand on its launch.
nowMarketing at a multi-brand group selling high-ticket products, in Cyprus.

Arabic and English.

A number that reads like an answer is worse than a gap.

Every call, with the reasoning behind it, and the one I'd make differently

Write to me.

mohamed.w.bondok at gmail.com

Based in Cyprus