Andrei Haret

In B2B, the gap decides whether growing spend is safe: a strong ROAS on a nine-month deal means nothing if cash runs out at month four.

• Before touching a campaign, the method is to run three numbers: monthly ad spend budget, target CAC and average deal size, then calculate the implied payback window.
• When payback exceeds six months and there is no recurring revenue model, scaling paid media simply accelerates the cash problem, they argue.
• Fix the math first.

A client celebrates a $120 cost per lead against an average contract value of $600 and a three-month implementation cycle, meaning every lead generated loses money. Haret treats this as a business math problem no campaign optimisation can repair.

Read original · Original 2

The account is not broken; the definition of success is.

• Haret’s fix is to attach the Google click ID to the lead in the CRM, have sales mark the stage that actually matters (qualified or opportunity), and import that stage back into the ad account as the conversion bidding optimises toward.
• Cost per lead goes up while cost per qualified lead goes down.
• That trade is the whole point, they argue, so tell the client in advance: a rising CPL reads as failure to anyone who doesn’t know the plan.
• Meta uses the same approach with different plumbing.

The obstacle is not the technical import but whether the sales team updates the CRM consistently. Where the sales process cannot produce reliable stage data, Haret says, don’t build the loop yet; fix that first.

Read original

The algorithm was optimising toward noise.

• The costs: pausing winners because they look expensive, scaling losers because they look cheap, and losing months of learning that cannot be recovered.
• The suggested fix is specific: start with Enhanced Conversions in Google Ads to match hashed user data to actual pipeline, then layer in offline conversion uploads tied to the CRM so campaigns optimise toward deals rather than form fills.

Haret reports that this single change shifted CPL perception by 15 to 20 percent in accounts that thought they had a traffic problem. The real issue, they say, was measurement, not traffic.

Read original

Founders who spend weeks reconciling them are doing the wrong job while their bid strategy starves.

• The practical recommendation is to pick Google Ads as the single bidding signal, import offline conversions from the CRM, and use GA4 and CRM totals for context rather than bidding.
• One source drives the machine; the others show whether it’s working.

For newcomers, Haret’s point is that perfect attribution isn’t required; what matters is putting the right signal in the right place. Whichever conversion source currently feeds smart bidding determines what the account buys.

Read original · Original 2

The judgment calls stay with people who know the account: what to test next, which segments to push harder, when to push back on a client’s campaign goals, and whether a CPL that looks bad is actually fine.

• A CPL that looks bad in isolation may be acceptable given a 90-day sales cycle, and an audience that appears to underperform may be the warmest segment, just slower to convert.
• The risk of over-automating, Haret argues, isn’t moving too fast but making faster decisions that are still wrong.
• They also describe automated processes as delivering about 95 percent quality on average, with the last 5 percent falling exactly where client reports, legal, finance and communications live.
• Every process keeps a human as the final checkpoint.

Haret describes building an architecture of scheduled tasks and automated workflows across operations, sales, marketing and recruiting, then concluding that it’s like hiring junior employees with no middle management: fine at five, a bottleneck at twenty.

In a later post, Haret describes deliberately downgrading their AI subscription after noticing they were using roughly half the system they had built.

• Adding new tools felt productive without producing value, and switching costs are real, so they made a rule to finish the last thing before chasing the new one.
• The shift is from building more automations to consolidating fewer of them.

Read original · Original 2 · Original 3 · Original 4

Given a campaign with decent CTR and in-range CPL where sales says lead quality is bad, the average answer stays inside the account (adjust targeting, test new creative, tighten the audience).

• The operator asks what sales is actually seeing, what the form-fill path looks like, and whether the right conversion event is tracked.
• The same preference shapes hiring: Haret screens for mathematical thinking and process orientation rather than certifications or platform speed.
• Running ads is learnable in six to twelve months, they argue, while reasoning about CAC, LTV and payback as business decisions takes years.

For two years, Haret says, they’ve used this interview question: a client’s CPL dropped 30 percent and MQL quality also dropped, so what do you do?

• A media buyer celebrates the CPL drop; an operator asks why quality fell and what the downstream CAC impact is.
• In the final round, candidates get a simplified account scenario.
• Haret listens for whether they frame it as a performance problem or a business problem.

Read original · Original 2

Bad client fit costs more than the contract is worth: opportunity cost, mental overhead and slow damage to team morale when a client is never satisfied.

• Their current filter has three checks in the first twenty minutes: can the client define success in concrete terms; do they own their funnel, with an early insistence that the landing page converts fine treated as a yellow flag; and can they approve a creative or budget change within 48 hours?
• The filter, Haret says, protects the team, not revenue.

The distinction is between “more leads” as an answer and “40 qualified leads at under $150 CPL” as a real definition of success.

• The question is: what does a good outcome look like in six months?
• Existing clients are said to pass all three checks.

Read original

The economics argument underneath every other post is clear: the metric most teams report is not the one that determines whether they can afford to keep spending.

Andrei argues that ROAS describes revenue-to-spend at a point in time, while CAC payback measures how long money stays out before a customer returns it.

• Before touching a campaign at MarketinGO, they run three numbers: monthly ad spend budget, target CAC, and average deal size (ACV), then calculate the implied payback window.
• Their example: a client celebrates a $120 CPL despite an ACV of $600 and a three-month implementation cycle, meaning each lead loses money.
• They say that when CAC payback exceeds six months and there is no recurring revenue model, scaling paid media accelerates the cash problem; fix the math first.

For brands planning a campaign

For a B2B SaaS brief, request CAC, ACV, and current payback before proposing spend levels.

• Make the payback window the gate for any scaling recommendation.
• Content could show how to calculate the implied payback window from those three inputs.

It clearly describes the CRM feedback loop, including the operational reason these projects fail.

The argument: smart bidding is a feedback machine.

• If the only signal it receives is a form submission, a bot, a student, a competitor, and a real buyer all look identical.
• The account is not broken; the definition of success is.
• Haret's fix is to attach the Google click ID to the lead in the CRM, have sales mark the stage that matters (qualified or opportunity), and import that stage back into the ad account as the conversion bidding optimizes toward.
• They spell out the trade-off: cost per lead goes up while cost per qualified lead goes down.
• Clients need that conversation before anything changes, because a rising CPL on a report looks like failure.
• They also name the obstacle: if the sales process does not produce reliable stage data, fix that before building the loop.
• Meta gets the same treatment with different plumbing.

For brands planning a campaign

This is the anchor piece for a lead-quality brief. Pair it with a practical mapping exercise: which CRM stage represents a qualified lead, who updates it, and how the import is wired.

It frames the mismatch as expected, not broken.

Andrei explains that the three sources measure different things: GA4 is last-click, deduped, and session-based; Google Ads uses its own attribution window and includes modeled conversions; the CRM records what actually closed and became revenue.

• The totals will never agree, so founders who spend weeks reconciling them are doing the wrong job while their bid strategy starves.
• Haret recommends choosing Google Ads as the single bidding signal, importing offline conversions from the CRM, and using GA4 and CRM totals for context rather than bidding.
• One source drives the machine; the others check whether it is working.

For brands planning a campaign

Useful for an explainer on attribution parity. For a brand, the practical question is which system can influence bidding and which systems are reporting-only.

It works as an audit checklist: the numbers can look clean while the conversion setup feeds the algorithm noise.

Auditing dozens of B2B accounts where CPL looked clean and budgets were scaling, Andrei found roughly half the leads were phantom: form submissions firing twice, thank-you page visits counted as conversions, no Enhanced Conversions, and no offline upload.

• The costs they identify: pausing winners because they look expensive, scaling losers because they look cheap, and losing months of learning that cannot be recovered.
• Haret's fix is specific: start with Enhanced Conversions in Google Ads to match hashed user data to actual pipeline, then layer in offline conversion uploads tied to the CRM so campaigns optimize toward deals instead of form fills.
• They report that this single change shifted CPL perception by 15-20% in accounts that believed they had a traffic problem.

For brands planning a campaign

Good source for a pre-campaign tracking audit deliverable: a short list of checks (double firing, thank-you page counting, Enhanced Conversions, offline uploads) a brand can complete before increasing budget.

The reasoning pattern—start outside the ad account—underpins the measurement and CRM posts.

The question Andrei says they use in interviews: a campaign has run six weeks, CTR is decent, CPL is within range, but the sales team says lead quality is bad.

• What do you look at first?
• The average answer stays inside the account: adjust targeting, test new creative, tighten the audience.
• The operator's answer starts outside it: what is the sales team actually seeing, what does the form-fill path look like, and are we tracking the right conversion event?
• Haret says they have used the question for two years and the pattern holds.

For brands planning a campaign

Useful as a diagnostic frame for content and for assessing a partner: ask how they would find the cause of a lead-quality complaint before touching bids or creative.

It sets clear limits on automation, useful context before buying into AI-run campaign promises.

Andrei says their team uses AI daily to spot anomalies faster, generate headline variants, and turn raw data into readable summaries—and that those tasks work well.

• People who know the account still make the judgment calls: a CPL that looks bad in isolation can be acceptable given a client's 90-day sales cycle, and an audience that seems to underperform can be the warmest segment, just slower to convert.
• Haret argues the risk of over-automating is not moving too fast, but making faster decisions that are still wrong.

For brands planning a campaign

A counter-position to AI-automation pitches: content could specify which decisions a tool may make and which require account history and client context.

The account is not broken; the definition of success is.

• The fix is to attach the Google click ID to the lead in the CRM, have sales mark the stage that matters (qualified or opportunity), and import that stage back into the ad account as the conversion bidding optimizes toward.
• Cost per lead goes up and cost per qualified lead goes down; Haret treats that trade as the point.
• Agree on the plan with the client before making changes, so a rising CPL on a report does not look like failure to anyone unaware of the plan.

Attach the click ID to the lead in the CRM, have sales mark the qualified or opportunity stage, then import that stage as the bidding conversion. Haret says they apply this in most B2B accounts and describe a matching loop for Meta.

Haret says the hard part is not the technical import but getting the sales team to update the CRM consistently. If the sales process cannot produce reliable stage data, fix that before building the loop.

For brands planning a campaign

Directly supports a campaign on building a CRM-to-ad-platform feedback loop and framing the resulting CPL increase internally before the change ships. A brand can use this for content on building a CRM-to-ad-platform feedback loop and on how to frame the resulting CPL increase internally before the change ships.

Examples: Source 1 · Source 2

In B2B, that difference determines whether scaling is safe.

• A strong ROAS on a nine-month deal means nothing if cash runs out in month four.
• Before touching a campaign, calculate three numbers: monthly ad spend budget, target CAC, and average deal size.
• Use them to calculate the implied payback window.
• When payback exceeds six months and there is no recurring revenue model, scaling paid media simply accelerates the cash problem.

Run three numbers before any campaign work — monthly ad spend budget, target CAC, and average deal size (ACV) — then calculate the implied payback window.

Haret says most new clients do not know these numbers, which is why paid media often feels broken.

For brands planning a campaign

Gives a brand a numbers-first angle on when to scale or pull back paid media, tying campaign decisions to finance rather than platform metrics.

Examples: Source 1 · Source 2

A CPL that looks bad in isolation may be acceptable given a client's 90-day sales cycle, and an audience that appears to underperform may be the warmest segment, just slower to convert.

• AI sees the number, not its context.
• When a campaign has decent CTR and an in-range CPL but sales says lead quality is bad, Haret's advice is to start outside the ad account: examine what sales is seeing, how the form-fill path works, and whether the right conversion event is tracked.

Work backward from a broken result and look outside the ad account before adjusting targeting, creative, or audiences.

For brands planning a campaign

Useful for content on lead-quality metrics and on why a rising CPL can accompany a better outcome, which is the internal conversation around a campaign aimed at explaining a rising CPL that the posts suggest.

Examples: Source 1 · Source 2 · Source 3

The algorithm was optimizing toward noise.

• The costs: pausing winners because they look expensive, scaling losers because they look cheap, and losing months of learning that cannot be recovered.
• Haret also argues that GA4, Google Ads, and the CRM will never agree because they measure different things.
• The practical answer is to choose one source as the bidding signal and use the others for context.

Start with Enhanced Conversions in Google Ads, match hashed user data to the actual pipeline, then add offline conversion uploads tied to the CRM. Audit the conversion setup before changing bids or budgets.

For brands planning a campaign

Offers a step-by-step audit angle and explains why reconciliation is the wrong job for content on paid-media lead quality.

Examples: Source 1 · Source 2

Judgment stays with the people who know the account: what to test next, which segments to push, when to push back on a client's campaign requirements, and whether a bad-looking CPL is actually fine given the account's sales cycle.

• The risk of over-automating is not moving too fast but making faster decisions that are still wrong.
• In a related post, Haret describes building scheduled tasks and automated workflows, and concludes that AI delivers roughly 95% quality on average.
• The last 5% covers client reports, legal, finance, and communications, so every workflow keeps a human check before anything goes out.

Use AI for anomaly detection, headline variants, and summaries; keep a named human as the final checkpoint on client-facing outputs.

Haret says no tool they have used can yet self-correct and make judgment calls without supervision. They also describe downgrading their own AI plan after realizing they were using about half of the system they had built.

For brands planning a campaign

Gives a brand a position on AI in paid-media operations: accelerate execution, with human checkpoints on anything client-facing.

Examples: Source 1 · Source 2 · Source 3 · Source 4

Their current filter has three checks in the first 20 minutes: can the client define success in concrete terms ("40 qualified leads at under $150 CPL" rather than "more leads"); do they own their funnel, with early insistence that the landing page converts fine treated as a yellow flag; and can they approve a creative or budget change within 48 hours? Haret says the filter protects the team, not revenue, and that paid media cannot save a broken funnel.

For brands planning a campaign

Relevant to content on intake and qualification for B2B paid-media engagements, including what to screen before accepting a lead with budget.

Examples: Source 1

Their argument: running ads can be learned in 6 to 12 months, but reasoning about CAC, LTV and payback windows as business decisions takes years.

• Their filter question: a client's CPL dropped 30% and MQL quality also dropped — what do you do?
• A media buyer celebrates the CPL drop; an operator asks why quality dropped and what the downstream CAC impact is.
• In final-round interviews, they present a simplified account scenario and listen for whether the candidate frames it as a performance problem or a business problem.

For brands planning a campaign

Useful for briefs on team building, agency evaluation or the operator-versus-executor distinction; a secondary angle for the lead-quality brief.

Examples: Source 1 · Source 2

Ask them to clarify their relationship to the business.

View example

Raza describes turning off Meta's one-click auto-fill form, adding qualification gates, and requiring manual phone entry.

• CPL, Raza argues, measures how lazy the form is; cost per sale measures whether the business survives.
• Haret makes the same distinction from the CRM and bidding-signal side: a form-fill signal makes bots, students, competitors, and buyers look identical, so the account needs a qualified-stage conversion.

Raza's evidence comes mostly from first-person agency narratives and Meta Ads campaign examples, including a connection milestone post. Haret's is more diagnostic and platform-agnostic, centred on Google Ads and CRM feedback loops, CAC payback math, and hiring filters.

Read original · Original 2

Gelber writes about cost per opportunity, financial alignment with finance, and how marketing leaders who model sales efficiency upfront can unlock more budget. Haret writes about CAC payback, qualified-stage conversion imports, and why a rising CPL can be the right trade if cost per qualified lead falls.

Gelber's posts explore the agency-client relationship: account stability, reporting clarity, initiative, ownership, and marketing-finance conversations. Haret focuses more on measurement plumbing and unit economics, while also addressing paid media hiring and client qualification.

Read original · Original 2

Cox's recruiting posts argue that hiring managers confuse a good employee with someone like themselves, and that resumes screen out 'whispering talents'; Haret screens paid media hires for operator thinking rather than certifications or platform speed. Cox's 'don't pave the cowpath' post and Haret's AI posts share the view that automation should not replace judgment.

Cox focuses on talent acquisition and recruiting, including a TEDx talk on hiring and AI in TA. Haret focuses on B2B paid media measurement, applying the same judgment-versus-automation argument to account decisions and campaign scaling.

Read original · Original 2

In one post, he walks through the change: turning off Meta's one-click auto-fill form, adding two mandatory gates on monthly budget and buying timeline, and requiring prospects to type their phone numbers manually.

• He reports a blunt result: cost per lead rose from $3 to $16 and lead volume dropped about 70%, after which prospects answered calls, recognised the brand and had budget ready.
• A second post explains the reasoning: cheap leads ruin businesses faster than expensive ones, a bad offer will break a good ad, and a short diagnostic of a founder's conversion funnel builds more trust than a long generic deck.
• For a SaaS team struggling with junk volume from lead forms, this is a directly applicable account of the trade-off: accept a higher cost per lead to buy intent.

The reasoning stops at the ad account and the form.

• There is no unit-economics layer: no CAC payback window, average deal size or deal-cycle arithmetic, and no rule for when spend should scale.
• There is also no CRM feedback loop — no offline conversion import, click-ID-to-CRM matching, or Google Ads equivalent to the Meta fix.
• The method raises intent at the point of capture and reads cost per sale; it does not use business math to show that more spend is safe.
• Buyers who need to justify budget with payback numbers or fix conversion tracking first will not find that here.
• The comparison set has no follower, engagement or commercial-availability data; ask for those figures if reach or bookable format will determine your choice.

Examples: Source 1 · Source 2

His material reframes the buyer's decision: an external partner's value often lies in mental bandwidth, not campaign execution alone.

• The clearest example compares two clients.
• One went from three to eighteen sales opportunities a month on the same $10,000 budget, a sixfold drop in cost per opportunity, yet could not scale because finance held strict spending limits.
• The other aligned with finance before launch on the cost-per-opportunity numbers needed to hit revenue goals, then won budget approval the following month.
• Gelber reduces the agency standard to four things — stability, clarity, initiative and ownership — and describes a VP of Marketing at a PE-backed SaaS company who moved from exhausting weekly calls and untrustworthy reporting to consistent account management and reporting that reduced friction.
• For a SaaS company likely to seek more spend after a successful pilot, that framing is closer to the internal decision the buyer will have to make.

The trade-off is specificity.

• Where Andrei Haret gives a tracking audit list and CRM import steps, Scott Gelber offers principles and a client conversation.
• These posts do not explain how to detect phantom conversions, define the qualified-opportunity stage or attach a click ID to a lead.
• Cost per opportunity is related to CAC payback but is not the same calculation, and the payback-window rule for deciding whether scaling will accelerate a cash problem is absent.
• Gelber's posts are agency-voice commentary on client experience: they explain why a partnership succeeds or fails rather than give readers something to implement.
• Choose this material when budget alignment and account management are the bottleneck; choose Haret when measurement and lead quality inside the account are the bottleneck.

Examples: Source 1 · Source 2