The Planes That Didn't Come Back: Allocating a Small Business Marketing Budget

Law Smith

Founder & President, Tocobaga

Law Smith is the Founder and President of Tocobaga, a Tampa-based ROI-focused marketing agency and SMB advisory. He has strategically advised 1,000+ small and medium businesses, executed 600+ integrated marketing campaigns, and written 30+ business plans across industries ranging from professional services to e-commerce to B2B technology. Tocobaga is a Google Partner and Squarespace Gold Partner headquartered in Ybor City.

Sometime during the Second World War, a federal court moved a hearing to the front of the docket so a Hungarian-born mathematician could keep reading what he had written. The mathematician was Abraham Wald. He had been born in Kolozsvár in 1902 to a religious Jewish family that homeschooled him because he would not attend classes on Saturdays, had been chased out of Vienna by the Anschluss, and had landed at Columbia University in 1938, where he would go on to found a field called sequential analysis.* He was also, on paper, an enemy alien. The reports he was producing for the United States military were classified, and someone had suggested that the pages be taken away from him. The court solved the problem the fastest way it could. It made him a citizen.*

The work in question came out of a room on Morningside Heights called the Statistical Research Group, assembled in the summer of 1942 with a staff that included Milton Friedman and George Stigler, and run on a simple rule: the statisticians answered the questions they were asked, and military decisions were made by the military.† One of the questions was about bombers. Planes came back from Europe full of holes, and the military wanted to know which parts of a plane, when hit, brought it down. Armor is heavy. Where to put it was the question underneath the question.

The version of what happened next that gets told in business schools and marketing decks goes like this: the officers wanted to armor the places with the most bullet holes, and Wald told them to armor the places with the fewest, because the planes hit there were the planes that never came home. It is a great story. It is also, according to the American Mathematical Society's review of the eight memoranda Wald actually wrote, a "plausible reconstruction." The memoranda never recommend where to put armor at all. What they do is harder and more useful. They take the only data anyone had, the holes in the planes that returned, and build a method to estimate what happened to the planes that did not.†

That is the whole job of a marketing budget, and almost nobody does it. An owner sitting down in October to decide what next year's marketing money should do has exactly Wald's data set: a pile of surviving evidence. The campaign that produced the client everyone remembers. The Facebook post that got shared. The trade show where the big handshake happened. The dollars that quietly bled out into channels nobody tracked, the leads that were never called back, the offer with no margin that got scaled anyway: those are the planes that did not come back, and they are not in the spreadsheet because nothing that disappears is ever in the spreadsheet.

Tocobaga, the lead generation agency and small business advisory that Law Smith runs out of Ybor City in Tampa and writes about on SolvingHow.com, has advised 1,000+ small and medium businesses, and some version of this conversation happens in most of them. The owner is not dumb. The owner is overwhelmed, and overwhelmed people spread armor evenly over the whole plane. A little Google Ads, a little Instagram, a boosted post, a magazine ad because the rep was nice, a sponsorship because the kids play there, a new website because the old one felt old. The plane gets heavy. It protects nothing in particular. The fix is a decision order, not a pie chart: one primary growth engine, then the owned assets that compound, then controlled experiments, with unit economics guardrails so the business never scales an offer that loses money on every sale. The benchmarks come with caveats, the examples come with arithmetic, and the cuts, when cash gets tight, come in a specific order.

How much should a small business spend on marketing?

Between 5 and 10 percent of revenue is the honest range for most established small businesses, and the number matters less than where it goes. The widely quoted benchmarks come from companies that look nothing like a two-partner law firm or a four-truck plumbing company, so they are a starting point, not a prescription.

Start with the two surveys everyone cites. Gartner's 2025 CMO Spend Survey put marketing budgets at 7.7 percent of company revenue, flat with 2024, based on 402 marketing leaders surveyed in early 2025, the vast majority of them at companies with annual revenue above $1 billion.‡ Read that last clause twice. A billion-dollar company spending 7.7 percent has a $77 million marketing department with a procurement team and a media agency. The same survey found that 59 percent of those marketing leaders called the budget insufficient to execute the year's strategy, which is a useful reminder that nobody at any size feels funded.‡ The spread inside the average tells the real story: half of the CMOs reported budgets of 6 percent of revenue or less, the budget-conscious group sat at 4 percent or below, and the big spenders ran above 10.5 percent.§

The CMO Survey, run out of Duke University's Fuqua School of Business, reported higher figures in the edition it published in March 2025: marketing at 9.4 percent of revenues and 11.4 percent of the overall company budget, from 281 U.S. marketing leaders surveyed in late January and early February 2025.‖ For B2B services companies, the group that includes law firms, consultancies and most professional services, the figure was 9.0 percent of revenues.‖ The Small Business Administration's own guidance, in a 2019 post on its blog, cited a 2018 industry estimate of 7.9 percent of revenues across businesses, with B2B services at 6.9 percent, B2B product companies at 6.3 percent, consumer product companies at 9.6 percent and consumer services at 11.8 percent.¶

So the range is real, and the caveats are bigger than the range. First, these are averages of companies that survived long enough to answer a survey, which is Wald's problem in a cardigan. The businesses that spent 12 percent on the wrong things are not filling out questionnaires from Duke. Second, "percent of revenue" is a lagging habit. It funds next year's growth based on last year's sales, which means a business that shrinks cuts marketing exactly when it needs leads most, and a business that grows funds marketing it may not need. Third, the percentage says nothing about what the money is for. A percent of revenue is armor distributed by weight. It tells an owner how many pounds to bolt on. It does not say where.

The practical use of a benchmark is as a sanity check at the end, not a starting point. If the unit economics say the business can afford to spend $9,000 a month to acquire customers and that works out to 11 percent of revenue, the benchmark says the plan is aggressive and the margin had better be real. If the same math says $2,000 a month and that is 2 percent of revenue, the benchmark says the business is probably underfunding the one thing that compounds. The rule that holds up across 600+ campaigns is to build the budget from the customer up and only then look at the percentage, and to treat any owner who starts with the percentage as a person about to spread the armor evenly.

What should a marketing budget be based on if not a percentage of revenue?

A marketing budget should be based on what one new customer is worth and what the business can afford to pay to get one. Four numbers do the work: gross margin, customer lifetime value, allowable acquisition cost and payback period, and an owner who knows those four can set a budget in an afternoon.

Gross margin first, because everything downstream is a fraction of it. Marketing is not paid for with revenue. It is paid for with the gross profit left after the cost of delivering the thing: the technician's hours, the product cost and shipping, the associate's billable time. A business with a 60 percent gross margin can afford twice the acquisition cost of a business with a 30 percent margin at the same price point, which is why the e-commerce store and the law firm should never be reading the same benchmark.

Lifetime value second, and the honest version of it: the gross profit a customer generates over a realistic horizon, not the romantic one. A residential HVAC customer who does one repair and never calls again has a lifetime value of one repair. The one on a maintenance plan who replaces a system in year four is a different customer with a different number, and the budget should know which one the campaign is bringing in. For a two-partner law firm, the horizon includes the referral, since a satisfied estate planning client who sends a sibling has roughly doubled what the firm can afford to spend acquiring the first one.

Allowable acquisition cost third. This is the ceiling: the most the business is willing to pay for one new customer, derived from lifetime gross profit rather than from what the ad platform charges. A common guardrail is to keep lifetime gross profit at three times acquisition cost or better, so that the business keeps two-thirds of the value it created after paying to create it. Three to one is a convention, not a law. A business with fast repeat purchases and strong cash can run tighter, and a business selling one-time projects with no referrals should run wider.

Payback period fourth, and this is the one overwhelmed owners skip, which is a shame because it is the one that kills them. Payback is how long it takes the gross profit from a new customer to return the money spent acquiring that customer. A 12-month payback on a software subscription is normal and survivable with a credit line. A 12-month payback at a restaurant with 16 days of cash in the bank is a slow-motion bankruptcy, and the JPMorgan Chase Institute's 2016 study of 597,000 small businesses found restaurants holding exactly that: a median of 16 cash buffer days, against 27 days for the median small business overall.** Lifetime value tells an owner whether the channel is profitable. Payback tells the owner whether the business will be alive to collect.

Put the four together and the budget writes itself. Worked example, with the arithmetic shown. A residential HVAC and plumbing company in Brandon, east of Tampa, does $2.4 million a year. Average ticket is $640. The average new customer does 2.3 jobs over three years, so three-year revenue per customer is $1,472. Gross margin after technicians, parts and truck costs is 48 percent, so three-year gross profit per customer is about $707, and first-year gross profit from the first job is about $307. At a three-to-one guardrail, allowable acquisition cost is $707 divided by three, or roughly $236 per new customer. The company spends $4,500 a month on Google Ads and Local Services Ads, which produces 54 leads, a cost per lead of $83.33, and 19 new customers, a cost per customer of $236.84. The channel is sitting right on the guardrail, and the first job alone ($307 in gross profit) pays back the $237 in acquisition cost inside the first month. Verdict: fund it, and fund the next dollar carefully, because the next $1,500 has to produce at least seven more customers ($1,500 divided by $236 is 6.4) to stay inside the line. If it produces four, the marginal dollars are losing money even though the average still looks fine, and the average is what the agency report will show.

That last sentence is the whole discipline in one line. Averages hide the marginal plane. Wald's memoranda were careful about assumptions in a way the legend is not; he worked out bounds on the answers rather than pretending to exact ones, and a budget should be held the same way, with the marginal customer and the marginal dollar in view.†

Why does spreading a marketing budget across many channels fail?

Spreading a small budget across many channels fails because no single channel gets enough money to produce data, and without data nothing can be improved. The owner ends up with eight small experiments, none of them conclusive, and a conviction that "nothing works," when the truth is that nothing was tested.

The mechanics are concrete, not philosophical. Google's automated bidding, the thing that makes a modern search campaign work, learns from conversions. Practitioner guidance published in Search Engine Land puts the working floor for the Target CPA strategy at about 30 conversions in 30 days and the floor for Target ROAS at 50 or more, and warns against changing targets before the campaign has collected enough conversions to learn from.†† Thirty conversions a month at an $85 cost per lead is roughly $2,550 of spend in one campaign before the machine has anything to say. An owner who splits $3,000 across Google, Meta, LinkedIn and a podcast sponsorship has guaranteed that the only channel with a learning algorithm will never learn, and that the other three will be judged on a sample size of a dozen clicks.

There is a second cost, and it is the owner's attention. Every channel needs creative, landing pages, tracking, weekly review and someone who understands its particular dialect. A business owner who is also the sales team and the HR department has maybe three hours a week for marketing. Three hours across six channels is 30 minutes each, which is enough time to notice a problem and not enough to fix one. Three hours on one channel is a real operator.

Michael Lewis wrote Moneyball in 2003 about a baseball team that could not afford to spread its money. The 2002 Oakland Athletics had $44 million for player salaries in a season when the New York Yankees' payroll exceeded $125 million.‡‡ They had just lost Jason Giambi to those Yankees, Johnny Damon to the Red Sox and Jason Isringhausen to the Cardinals.§§ Billy Beane's answer was not to buy a little of everything. It was to find the one undervalued thing, on-base percentage, and buy as much of it as the budget allowed, while treating the statistics the rest of the league paid for, stolen bases, runs batted in and batting average, as "relics of a 19th-century view of the game."‡‡ That team won 103 games and ran off a 20-game winning streak, an American League record at the time, from August 13 to September 4.§§ They also lost in the first round of the playoffs, three games to two, which is the part of the story that keeps it honest: concentration raises the floor and the ceiling, and it still does not guarantee October.§§

Concentration has a failure mode, too, and it deserves plain language. One channel can get expensive overnight, a platform can change its rules, and a business with 100 percent of its leads from Google Ads has a landlord, not a channel. The answer is not to spread thin from day one. It is to win one channel, then use the profit from it to fund the owned assets that make the business less dependent on any channel, then run a second experiment with real money. Order matters more than mix.

What is the right order to fund marketing for a small business?

The right order is one primary growth engine first, owned assets second, and controlled experiments third, with everything else at zero until the first two are working. This is a priority filter, not a pie chart, and it exists so an overwhelmed owner can say no to the magazine rep without a meeting.

Tier one is the engine: the single channel most likely to produce a qualified lead this quarter for this kind of business. For a local service company it is almost always Google in some form, because the customer with a flooded kitchen searches, calls and hires within the hour. For an e-commerce store it is usually a paid social or paid search channel plus the product page it lands on. For a law firm or a consultancy it is often referral infrastructure plus search, because the buyer asks a friend first and Googles second. The engine gets the majority of the money, usually 50 to 70 percent of the budget, until it hits the unit economics guardrail or runs out of demand, whichever comes first.

Tier two is the owned assets: the email list, the website, the Google Business Profile, the reviews, the content that answers the questions buyers actually ask. These compound. An email list of 8,000 past customers and prospects costs almost nothing to mail and does not get more expensive when a competitor bids on the same keyword. Litmus's 2025 State of Email survey of nearly 500 marketing professionals found that 35 percent of marketing leaders reported getting $10 to $36 back for every dollar spent on email and another 30 percent reported $36 to $50, with 21 percent admitting they do not measure it at all.‖‖ Treat those numbers as self-reported and generous, because they are. Even cut in half, email is the on-base percentage of small business marketing: the undervalued stat that wins games while the expensive stats get the headlines. Owned assets get 20 to 35 percent of the budget, more in a mature business that already has an engine humming.

Tier three is controlled experiments: one new channel at a time, with a fixed budget, a fixed window and a kill number written down before the first dollar goes out. Ten to 15 percent of the budget, never more, and never two experiments at once, because two experiments at once is how the planes end up with armor everywhere.

The ordered system, the one an owner can run this month:

• Step one: write down gross margin, three-year gross profit per customer, allowable acquisition cost at three to one, and the current cash buffer in days. If any of the four is unknown, the budget waits until it is known.

• Step two: pick the one engine by asking where the last 20 customers actually came from and where the next 20 are most likely to search, ask or browse. Pick one. Not the one that is fun.

• Step three: fund the engine to the point where it produces at least 30 leads a month, or to the guardrail, whichever comes first, and give it 90 days before judging it.

• Step four: set up tracking that can attribute each new customer to a source: a call tracking number per channel, a required lead source question at intake asking how the customer found the business, and a monthly reconciliation of leads to signed customers in a spreadsheet or the CRM.

• Step five: put the owned assets on a maintenance budget: one email a month to the whole list at minimum, a complete and current Google Business Profile, a reviews request built into the close of every job, and a website that loads fast and shows a phone number above the fold.

• Step six: choose one experiment per quarter, write the kill number (the cost per lead above which it dies) before launch, and run it for the full window without touching it.

• Step seven: every month, reconcile spend to customers by channel, compute marginal cost per customer on the engine, and move money from anything losing to the engine or the list.

• Step eight: every quarter, re-run the four numbers from step one, because the margin moved, the ticket moved, and the budget should move with them.

The table below sets the three business types side by side: engine first, owned assets second, experiments last, with the guardrail that decides when the engine gets more fuel.

The budget priority filter by business type: engine first, owned assets second, experiments third, and the guardrail that decides when the engine gets more money.

Business type	Primary growth engine (50 to 70 percent)	Owned assets (20 to 35 percent)	Controlled experiments (10 to 15 percent)	Guardrail that decides more fuel
Local service (HVAC, plumbing, roofing, dental)	Google Ads and Local Services Ads landing on a page built for a phone call, with the Google Business Profile converting the map pack	Customer email list with a seasonal offer, a reviews system on every closed job, real photos on the profile, a website the business owns	One per quarter: Nextdoor, direct mail to a zip code, a trade partner referral program, with a kill number set in advance	Marginal cost per new customer under three-year gross profit divided by three; payback inside the first job
E-commerce and product (Shopify store, DTC brand)	Paid social or paid search plus the product page it lands on, funded only when first-order contribution margin is survivable	Email and SMS list for the repeat purchase, product photography, the storefront itself	One per quarter: a marketplace, a creator partnership, a bundle test, run once with a kill number	Return on ad spend above first-order breakeven; payback period shorter than the cash buffer in days
Law firm and professional services (consultancy, CPA, lobbying)	Referral system plus search capture (Google Ads, Local Services Ads where available) for the moment the buyer looks	Owned website with substantive answer pages, a past-client and referral-source newsletter, Google Business Profile and reviews	One per quarter: a referral partner program, a speaking slot, a narrow LinkedIn test, with a kill number	Cost per signed client under lifetime gross profit (including expected referrals) divided by three; partner hours available to serve the work

How should a local service business allocate its marketing budget?

A local service business should put most of its budget into capturing search demand (Google Ads, Local Services Ads and the Google Business Profile that converts it), fund reviews and email as the owned layer, and run one experiment at a time in channels like Nextdoor, direct mail or a trade partner program. The customer with a problem searches with intent, so intent capture is the engine.

Take the Brandon HVAC company again, now as the eleven-location platform a private equity group might assemble out of it. Each location spends $4,500 a month on Google Ads and Local Services Ads, which is $49,500 a month across the platform, producing at the same rates roughly 594 leads and 209 new customers a month at about $237 each. The arithmetic scales, but two things change. First, at eleven locations the marginal cost per customer differs by market, so the budget should move toward the locations with headroom and away from the ones that have exhausted local search demand, which is a monthly decision, not an annual one. Second, the owned layer becomes a real asset: a combined customer file of tens of thousands of households, each with a known equipment age, is the maintenance plan and replacement pipeline that a buyer pays for at the exit. Tocobaga has watched more than one platform spend heavily on acquisition while the customer file sat in a shoebox, and the pattern that keeps surfacing across 1,000+ advisory conversations is that the list is the cheapest lead source the business owns and the last one anybody funds.

For a single-location service business, a $6,000 monthly budget tends to sort like this. Engine: $3,800 across Google Ads and Local Services Ads, with branded search protected and the landing page built for a phone call rather than a brochure. Owned: $1,400 covering a monthly customer email with a seasonal offer, a reviews system that asks every closed job, Google Business Profile upkeep with real photos, and the hosting and upkeep of a website the business actually owns. Experiment: $800 on one channel for the quarter, with a kill number written down. Everything else, including the youth league banner, the chamber directory listing and the radio spot, goes to zero until the engine is at the guardrail. The banner can come back when the business is turning away work.

How should an e-commerce or product business allocate its marketing budget?

An e-commerce business should build the budget on contribution margin per order, treat paid social and paid search as the engine only when first-order economics are survivable, and push the repeat purchase into email and SMS where the cost per order approaches zero. Product businesses fail on payback more often than on profitability, because the ad platform gets paid today and the repeat order arrives in month four.

Worked example. A Shopify store selling premium coffee equipment has an average order value of $80. Product cost runs 45 percent of the price, payment processing takes 3 percent, and shipping and packaging cost a flat $13.60 per order, so contribution on an $80 order is $28, or 35 percent. Breakeven return on ad spend for a first order is $80 divided by $28, or 2.86. The store's Meta campaigns run at a 2.2 return. Spend $10,000, get $22,000 in revenue, which is 275 orders, which is $7,700 in contribution, which is a first-order loss of $2,300. Over the next twelve months those 275 customers place about 0.9 additional orders each, call it 248 orders and $19,840 in revenue, which adds $6,944 in contribution. Twelve-month contribution is $14,644 against $10,000 of spend, a 1.46 to 1 return, with payback landing somewhere in the third or fourth month depending on how fast the repeats come in. Profitable, slowly.

Now scale it the way an excited founder scales it, to $30,000 a month. The store is floating a $6,900 first-order loss every month while it waits for repeats, which is a hole of roughly $20,000 by the end of the first quarter, which is more than the entire cash reserve of the median small business in the JPMorgan Chase Institute's data, where half of all small businesses held less than one month of outflows in the bank.** The offer is profitable on paper and lethal in practice, and the only reason an owner cannot see it is that the dashboard shows return on ad spend and not payback. Those are the missing planes.

The fix is not to cut the engine. It is to change the plane. Raise the order value so the flat shipping cost stops eating the margin: at a $110 bundle with the same product cost ratio and the same $13.60 to ship, contribution per order rises to $43.60, and the first-order loss on $10,000 of spend shrinks from $2,300 to about $1,280 at the same 2.2 return. Get the return to 2.6 with a better product page and the first order breaks even, give or take a few hundred dollars, and everything after it is profit. Then move the repeat purchase off the ad platform. A list of 12,000 customers that converts 2 percent per send at an $80 order produces 240 orders and $19,200 in revenue per email, at a platform cost measured in hundreds of dollars, which is why the email ROI figures in the Litmus survey, generous as self-reported figures always are, point in a believable direction.‖‖ Budget split for a store at this stage: 55 percent to the paid engine, 30 percent to the owned layer (email, SMS, product photography and the site), 15 percent to one experiment, which for most stores should be the marketplace or the creator partnership the founder keeps talking about, run once with a kill number instead of forever on a hunch.

How should a law firm or professional services firm allocate its marketing budget?

A law firm or professional services firm should fund a referral system and search capture as the engine, invest heavily in the owned assets that establish expertise (the website, the content that answers client questions, the email list of past clients and referral sources), and experiment cautiously with paid social, because the buyer takes longer to decide and checks more sources before calling. Trust is the product, so the budget buys proof.

A two-partner estate planning and probate firm in South Tampa, worked through. Average matter is $3,400. After attorney and paralegal time directly on the file, the firm keeps about 55 percent, or $1,870 per matter. One in four clients refers another client within two years, so the expected value of a new client is the first matter plus a quarter of a second one: $3,400 plus $850, or $4,250 in revenue and about $2,338 in gross profit. At a three-to-one guardrail, allowable acquisition cost is about $779 per signed client. The firm spends $3,000 a month on Google Ads and Local Services Ads and gets 24 consultations and 8 signed clients, a cost per signed client of $375, less than half the ceiling, with payback inside the first invoice. That channel should get more money until the cost per signed client approaches the ceiling or the partners run out of hours, and in a two-partner firm the hours run out first, which is its own budget signal: the next marketing dollar is an associate, not an ad.

The owned layer carries more weight here than in any other business type, and the B2B services average in The CMO Survey, 9.0 percent of revenues against 6.7 percent of the overall company budget, reflects a category that spends on expertise rather than impressions.‖ For the firm above, a $6,000 monthly budget sorts into $3,000 on the search engine, $2,100 on owned assets (a quarterly client newsletter that reminds 1,100 past clients the firm exists, one substantive article a month that answers a real question like what probate costs in Florida, Google Business Profile and reviews, and a website the firm owns outright rather than rents from a legal directory), and $900 on one experiment per quarter, most often a referral partner program with the financial advisors and CPAs who already see the client first. LinkedIn, for a firm like this, is a trust layer, not a lead engine, and gets content, not budget.

The same shape holds for a management consultancy or a lobbying shop with longer cycles and bigger tickets: fewer leads, more proof, and an owned list of past clients and referral sources that gets mailed with something useful every month whether or not there is anything to sell. A consultancy with 60 past clients and 200 warm contacts does not need a Meta campaign. It needs the 260 people to remember, on the Tuesday the problem shows up, whose name to say.

How does an owner know when to scale a channel and when to stop?

Scale a channel when the marginal cost per customer is still below the allowable acquisition cost and the payback period is shorter than the cash buffer, and stop when either of those flips. The average cost per customer is the wrong number to watch, because the average stays pretty while the last dollars spent turn ugly.

The guardrails, in the order an owner should check them each month. One: contribution margin after all delivery costs is positive on the offer being advertised. If it is not, no amount of marketing fixes it, and scaling it only makes the hole bigger faster. Two: the cost of the last batch of customers, not the lifetime average, is under the allowable acquisition cost. Three: payback period is shorter than the cash buffer in days, with room to spare, because the JPMorgan Chase Institute found that a quarter of small businesses held fewer than 13 days of cash, and a 90-day payback at 13 days of cash is a loan the owner did not know was taken out.** Four: the leads being bought are the leads being closed. A channel producing 60 leads and two customers is not a lead problem, it is a sales process problem, and more budget will produce more unclosed leads.

The reason this matters is in the Bureau of Labor Statistics survival tables. Of the business establishments born in 2013, 79.6 percent were still operating a year later, 50.6 percent after five years and 34.7 percent after ten.¶¶ The ones that die in years two through five are rarely the ones with no customers. They are often the ones that found a way to acquire customers at a loss and did it faster, which is the specific thing a marketing budget without guardrails is designed to do. Growth is not the goal. Profitable growth with payback inside the cash buffer is the goal, and a budget that cannot say which of the two it is producing is spreading armor by feel.

What should a small business cut first when cash is tight?

Cut experiments first, then unmeasured spend, then the owned-asset upkeep that is not producing, and cut the engine last, in small steps, while watching marginal cost per customer. The instinct to cut marketing to zero in a downturn is the single most expensive reflex in small business, and the evidence against it is old and clear.

Harvard Business Review's study of 4,700 public companies through the recessions of 1980 to 1982, 1990 to 1991 and 2000 to 2002 found that 17 percent did not survive and only about 9 percent came out stronger, outperforming rivals by at least 10 percent in sales and profit growth.*** The companies that cut hardest did not win. The firms that focused on defensive cost cutting had a 21 percent probability of pulling ahead after the recession, the ones that relied mainly on layoffs had 11 percent, and the "progressive" companies that combined selective cuts with continued investment in markets and assets had 37 percent.*** Those are big companies with balance sheets, and the lesson scales down, because a two-partner firm that goes dark for six months is invisible on the day the market turns and spends the following year rebuilding what it had.

The cut order, specifically. First, the quarter's experiment, which was always designed to be killable. Second, anything that cannot be tied to a customer: the sponsorship, the print ad, the directory, the agency retainer that reports impressions. Third, the owned-asset spend that is maintenance rather than production: the website refresh can wait, the newsletter cannot, because the list is the cheapest demand the business has and mailing it costs almost nothing. Fourth, and only fourth, the engine, trimmed from the top by cutting the worst-performing campaigns, locations or keywords first and watching the marginal cost per customer as the budget comes down. A business that cuts the engine 20 percent and loses only 8 percent of its customers has found fat. A business that cuts 20 percent and loses 25 percent has cut muscle and should put the money back.

There is one more category, and it is the one owners defend hardest: the spend that worked last time. France spent most of the 1930s and around 3 billion francs, a figure even the reference books flag as uncertain, building the Maginot Line along its eastern border, a fortification designed with great care for the war France had just fought and named for André Maginot, the minister of war who championed it.††† In May 1940 the German army went around it through the Ardennes, and the campaign took about six weeks, from the attack on May 10 to the armistice on June 22.††† Every business has a Maginot Line: the trade show that built the company in 2011, the phone book money that became a directory listing that became a "digital presence package," the radio buy the founder swears by. When cash is tight, that spend goes before the engine, however good the story, because the story is from the last war.

How often should a small business review its marketing budget?

Monthly for reconciliation, quarterly for reallocation and annually for the reset, which is a lighter lift than it sounds once the tracking exists. The annual budget most owners build in December is a forecast made with the least information of the year, and the businesses that treat it as a monthly instrument rather than an annual document are the ones whose budgets stop lying.

The monthly review is 45 minutes with one sheet: spend by channel, leads by channel, customers by channel, cost per customer by channel, and the marginal cost on the engine. Leads that came in with no source get a source, by calling them if necessary, because unsourced leads are planes that landed at another airfield and never got counted. The quarterly review moves money: experiments get killed or promoted to the owned or engine tier, the engine gets more fuel if it is under the guardrail, and the four unit economics numbers get recomputed, since the average ticket and the margin drift every quarter whether anyone is watching. The annual reset asks the bigger question of whether the engine is still the right engine, which for a business that has grown from one location to four, or from a solo practice to a firm with two associates, it often is not.

One caution that the 2025 Gartner data makes vivid: paid media now takes 30.6 percent of the average marketing budget, and paid search alone is 13.9 percent of digital spend.§ At the same time, 39 percent of the surveyed marketing leaders said they planned to cut agency budgets, and 22 percent said generative AI had let them lean less on outside agencies for creative and strategy.‡ Translation for a small business: the platforms are taking a bigger share of every dollar, the fee layer on top of them is under pressure, and an owner paying an agency a percentage of ad spend is paying a toll that rises exactly when the platform gets more expensive. That arrangement rewards the agency for spending, not for customers, which is why Tocobaga charges a flat fee instead, never a percentage of media, with contracts that do not auto-renew and with every account, asset and deliverable owned by the client from the first day and carrying no agency branding. Tocobaga is also not the right fit for every business and will say so on the first call, which is a cheaper way to find out than a twelve-month retainer.

What mistakes do small business owners make with marketing budgets?

The most common mistakes are budgeting from last year's revenue instead of this year's unit economics, spreading the money across too many channels to learn anything, judging channels on averages instead of marginal cost, and defending legacy spend because it worked once. Underneath all four is the same error Wald was hired to correct: deciding based on the planes that came back.

The owner who says Facebook did not work usually ran $400 over three weeks with a boosted post and no landing page, which is not a test of Facebook, it is a test of $400. The owner who says Google Ads is too expensive is usually looking at cost per click, a number with no customer in it, instead of cost per signed customer against allowable acquisition cost. The owner who says referrals are the only thing that works is right about the past and has no system for making the next referral happen, so the engine is a weather pattern. And the owner who says there is no budget for marketing often has one, spread so thin across a website host, a directory, a sponsorship and a part-time social media person that it is doing nothing in particular, like armor bolted evenly across a plane that cannot climb.

A simpler mistake hides under those: not counting. A rule Tocobaga holds every client to is that a budget without a lead source on every new customer is not a budget; it is a wish with a credit card. The tracking setup is not sophisticated. A phone number per channel, one required question at intake, a monthly reconciliation of signed customers to sources. The hard part is doing it in month seven when the business is busy, which is also when the data starts to be worth something.

An owner who wants a second set of eyes on the budget, the unit economics and the order of operations can book a fit call with Tocobaga at (813) 534-1261 or through SolvingHow.com; the call is free, and if the business needs something Tocobaga does not do, the answer will be a name and a number, not a proposal.

Abraham Wald never saw combat. He never touched the armor. He sat in an office at Columbia with a table of hits on planes that had survived and spent his war refusing to assume that the survivors were the whole story, and when the government decided that an enemy alien should not read his own memoranda, a court made him a citizen so the work could continue.* In December 1950, traveling in India on a lecture tour at the invitation of the Indian government, he and his wife died when their Air India DC-3 crashed in the Nilgiri Mountains; their two children were home in the United States.* The method outlived him, and it is a small method, which is why it works. Look at the surviving data. Ask what it cannot see. Then decide where the armor goes. A marketing budget is that question with a dollar sign on it, and the owner who asks it every month will fly a lighter plane, carry more fuel, and come back.

How much should a small business spend on marketing?

Most established small businesses land between 5 and 10 percent of revenue. Gartner's 2025 survey of mostly billion-dollar companies put the average at 7.7 percent, The CMO Survey found 9.4 percent, and SBA guidance cites a 7.9 percent industry estimate. The better method builds the budget from unit economics first: gross margin, customer lifetime value, allowable acquisition cost and payback period. The percentage then serves as a sanity check rather than the starting point.

What is an allowable customer acquisition cost?

Allowable customer acquisition cost is the most a business can pay to win one new customer while staying profitable. A common guardrail divides lifetime gross profit per customer by three, so the business keeps two-thirds of the value it creates. A customer worth $707 in three-year gross profit supports about $236 in acquisition cost. That ceiling applies to every channel and tells an owner when to add budget and when to stop.

Is it better to focus on one marketing channel or several?

One channel tested hard beats several channels tested lightly. Practitioners generally recommend about 30 conversions a month before automated bidding can optimize, and an owner's attention divides the same way money does. A small business should fund one primary growth engine until it reaches its acquisition cost ceiling, then build owned assets like an email list and website, then run one controlled experiment at a time with a kill number written down in advance.

What is a marketing payback period and why does it matter for a small business?

Payback period is how long the gross profit from a new customer takes to return the money spent acquiring that customer. It matters because the median small business holds about 27 days of cash, according to JPMorgan Chase Institute research, and a quarter hold fewer than 13. A channel can be profitable over twelve months and still drain the account in the first three. Payback should be shorter than the cash buffer, with room to spare.

Should a small business cut marketing when cash is tight?

Not to zero. Harvard Business Review's study of 4,700 companies across three recessions found that firms combining selective cuts with continued investment had a 37 percent chance of pulling ahead afterward, against 21 percent for firms that only cut costs. Cut experiments first, then unmeasured spend like sponsorships and directories, then non-producing upkeep, and trim the primary engine last in small steps while watching marginal cost per customer.

How does a local service business decide its marketing budget?

Start with the value of one customer: average ticket, repeat jobs over three years, and gross margin after technicians, parts and trucks. Divide three-year gross profit by three to get the allowable acquisition cost. Fund Google Ads, Local Services Ads and the Google Business Profile as the engine until cost per new customer reaches that ceiling. Then fund reviews, a customer email list and the website, and run one experiment per quarter with a preset kill number.

Sources

*

Wikipedia. "Abraham Wald." Biographical entry covering his 1902 birth in Kolozsvar, emigration to the United States in 1938, Columbia University faculty appointment, enemy alien status and naturalization, the Statistical Research Group, and his death in the December 13, 1950 Air India DC-3 crash in the Nilgiri Mountains of India. https://en.wikipedia.org/wiki/Abraham_Wald

†

Casselman, Bill. "The Legend of Abraham Wald." Feature Column, American Mathematical Society, June 2016. Reviews the eight Statistical Research Group memoranda on aircraft survivability and describes the popular armor story as a "plausible reconstruction." https://www.ams.org/samplings/feature-column/fc-2016-06

‡

Gartner. "Gartner 2025 CMO Spend Survey Reveals Marketing Budgets Have Flatlined at 7.7% of Overall Company Revenue." Press release, May 12, 2025. Survey of 402 CMOs and marketing leaders in North America, the UK and Europe, most at companies with annual revenue above $1 billion, fielded February through March 2025; 59 percent report insufficient budget; paid media at 30.6 percent of budgets; 39 percent plan agency cuts; 22 percent cite GenAI reducing agency reliance. https://www.gartner.com/en/newsroom/press-releases/2025-05-12-gartner-2025-cmo-spend-survey-reveals-marketing-budgets-have-flatlined-at-seven-percent-of-overall-company-revenue

§

Campaign. "Marketing budgets hold at 7.7% in 2025: Gartner CMO survey." May 2025. Reports the Gartner 2025 CMO Spend Survey distribution: half of CMOs at 6 percent of revenue or less, budget-conscious CMOs at 4 percent or less, big spenders above 10.5 percent, paid media at 30.6 percent of budgets and paid search at 13.9 percent of digital spend. https://www.campaignlive.com/article/marketing-budgets-hold-77-2025-gartner-cmo-survey/1920581

‖

The CMO Survey (Duke University Fuqua School of Business, Deloitte and the American Marketing Association). "Highlights and Insights Report," March 2025. 281 U.S. marketing leaders surveyed January 21 to February 12, 2025; marketing at 9.4 percent of revenues and 11.4 percent of overall budget; B2B services at 9.0 percent of revenues and 6.7 percent of overall budget. https://cmosurvey.org/wp-content/uploads/2025/03/The_CMO_Survey-Highlights_and_Insights_Report-2025.pdf

¶

Lesonsky, Rieva. "How to Get the Most From Your Marketing Budget." U.S. Small Business Administration blog, July 9, 2019. Cites a 2018 industry estimate of average marketing spending at 7.9 percent of revenues, with B2B product at 6.3 percent, B2B services at 6.9 percent, B2C product at 9.6 percent and B2C services at 11.8 percent. https://www.sba.gov/blog/how-get-most-your-marketing-budget

**

JPMorgan Chase Institute. "Cash Is King: Flows, Balances, and Buffer Days. Evidence from 600,000 Small Businesses." September 2016. Analysis of 597,000 small businesses and 470 million transactions, February to October 2015; median of 27 cash buffer days; about half hold less than one month; a quarter hold fewer than 13 days; restaurants at a median of 16 days. https://www.jpmorganchase.com/institute/all-topics/business-growth-and-entrepreneurship/report-cash-flows-balances-and-buffer-days

††

Saskin Gales, Jyll. "Everything you need to know about Target CPA bidding." Search Engine Land, February 12, 2025. Practitioner guidance recommending at least 30 conversions in 30 days before using Target CPA, 50 or more for Target ROAS, and collecting 20 to 30 conversions at a target before adjusting it. https://searchengineland.com/target-cpa-bidding-451823

‡‡

Wikipedia. "Moneyball (book)." Entry on Michael Lewis, Moneyball: The Art of Winning an Unfair Game (2003): the 2002 Athletics competed on a $44 million player salary budget while the Yankees' payroll exceeded $125 million, and the book's argument that on-base and slugging percentage outperform the traditional statistics it calls "relics of a 19th-century view of the game." https://en.wikipedia.org/wiki/Moneyball_(book)

§§

Wikipedia. "2002 Oakland Athletics season." Record of 103 wins and 59 losses, the American League record 20-game winning streak from August 13 to September 4, 2002, the offseason losses of Jason Giambi, Johnny Damon and Jason Isringhausen, and the 3 to 2 Division Series loss to Minnesota. https://en.wikipedia.org/wiki/2002_Oakland_Athletics_season

‖‖

Litmus. "The ROI of Email Marketing." July 16, 2025. Litmus 2025 State of Email survey of nearly 500 marketing professionals worldwide; 35 percent of marketing leaders report $10 to $36 back per dollar spent on email, 30 percent report $36 to $50, and 21 percent do not measure email ROI. https://www.litmus.com/blog/infographic-the-roi-of-email-marketing

¶¶

U.S. Bureau of Labor Statistics. "34.7 percent of business establishments born in 2013 were still operating in 2023." The Economics Daily, January 12, 2024. Business Employment Dynamics data: 79.6 percent of establishments born in 2013 survived one year, 50.6 percent survived five years and 34.7 percent survived ten. https://www.bls.gov/opub/ted/2024/34-7-percent-of-business-establishments-born-in-2013-were-still-operating-in-2023.htm

***

Gulati, Ranjay, Nitin Nohria and Franz Wohlgezogen. "Roaring Out of Recession." Harvard Business Review, March 2010. Study of 4,700 public companies through the recessions of 1980 to 1982, 1990 to 1991 and 2000 to 2002: 17 percent did not survive, about 9 percent flourished, and firms combining selective cuts with investment had a 37 percent probability of breakaway performance versus 21 percent for cost-cutting-focused firms and 11 percent for firms relying mainly on layoffs. https://hbr.org/2010/03/roaring-out-of-recession

†††

Wikipedia. "Maginot Line." Built from 1929 into the late 1930s at a cost the entry gives as around 3 billion French francs while flagging the figure for clarification; named for Minister of War Andre Maginot; bypassed through the Ardennes when the German attack began on May 10, 1940, with the armistice signed June 22, 1940. https://en.wikipedia.org/wiki/Maginot_Line