Buy martech from inside the company and you probably know less than the analyst or the agency you are paying. A good analyst has read more contracts than you have. A good implementation partner has run a hundred builds to your three.
You still hold one advantage nobody in that room can take from you. Your payday is not at signature.
The research firm sells its subscription at evaluation time. The partner's commission runs for a published number of years and then stops. The vendor's own investor disclosures say the account has to be larger next year than this one. You are the one still there in year four, when the renewal quote arrives, when the integration hits a ceiling that turns out to be a subscription tier, when a feature you built on gets removed mid-term, and when the company you signed with is owned by someone you never evaluated.
Nobody in that room is lying to you. Each seat is answering the question its compensation rewards. Your advantage is not better information. It is a longer time horizon and a copy of the contract.
Every company named in this paper appears because of a public document — its own, or one filed about it — and for no other reason. SEC filings, pricing pages, partner FAQs, standard terms of service, an acquirer's announcement, a community ledger. All of it is free to read before you buy.
Who is in the room, and what each seat is paid for
The map below is my synthesis. Every incentive in it is one the party disclosed about itself, in a document I link. It alleges nothing about anyone's motives.
| Seat | Compensation event | Where they disclose it | The question this seat is rewarded for asking |
|---|---|---|---|
| Research firm | Subscription research sold to buyers and to vendors; report reprints licensed to vendors for marketing | Gartner FY2025 10-K; Gartner Flex Reprints guidelines; Forrester FY2025 10-K | Which vendors belong in the category, and how should the category be framed? |
| Implementation partner | Commission at a published rate for a published number of years, plus services fees | HubSpot partner commissions FAQ | Which platform, and can we win the build? |
| Vendor | Subscription revenue that has to expand net of churn to hit plan | Braze FY2026 results | How do we land, and how do we grow the account? |
| You | Salary. You carry the renewal, the integration, the migration, and the blame | Nothing to disclose. That is the point | What does year three cost, and what does leaving cost? |
What the research firms say about themselves
Both dominant research firms file annual reports, and the plain description of the business in a 10-K is more useful to a buyer than any commentary about analyst bias.
Gartner's FY2025 Form 10-K states that within the Insights segment, "Global Technology Sales ('GTS') sells products and services to users and providers of technology, while Global Business Sales ('GBS') sells products and services to all other functional leaders." The same filing attributes the firm's competitive advantage to independence ("Our independent operating model and research analysis generates unbiased insights") and carries a risk factor warning that if its viewpoints "are considered to be wrong, lack independence, or are not substantiated by appropriate research, our reputation will suffer." (Gartner FY2025 10-K, filed 2026-02-12)
Gartner runs a Flex Reprints program under which "reprint clients" license Gartner reports for external marketing use. The published guidelines list where a licensed reprint may appear, and the list runs from landing pages and paid search ads to billboards, booth signage and call scripts. (Gartner Flex Reprints Guidelines, accessed 2026-09-02)
Forrester's FY2025 Form 10-K states that "Our Hi-Tech groups focus on North American technology vendors," and that Forrester Consulting "includes consulting projects, content marketing, and advisory services." The same filing states that "For more than 40 years, Forrester has been providing objective, independent and data-driven research insights," adhering to "rigorous, unbiased research methodologies." (Forrester FY2025 10-K)
Read those filings together and one structure comes out of them. The same firm sells subscription research to the buyer, sells subscription research to the vendor, and licenses the resulting report back to the vendor as marketing collateral. All three are disclosed. Neither firm discloses what share of its revenue comes from technology providers, so nobody outside can put a proportion on it, and I am not going to invent one. The verified claim is categorical. They sell to both sides, and they say so.
Read a category report as a category report, then, and use it for what it is good at. A research firm is paid to answer "who belongs in this category" well. You need "what does this cost me in year three," which nobody upstream of the renewal has a reason to work out.
The numbers you are handed, and the one you already have
The most-cited statistic about martech and CRM project failure has no traceable primary source. I went looking for one before writing this and found articles citing articles, the definition of failure shifting between them, and the trail running back roughly a quarter-century. It is not in this paper. A number that survives on repetition rather than provenance is the artifact this piece argues against, and quoting it would refute the argument in the act of making it.
Sourced figures in the same neighborhood do exist. Gartner's martech topic page, reporting the 2025 Gartner Marketing Technology Survey, states that "Only 49% of tools are actively used, and just 15% of organizations qualify as high performers." (Gartner, Maximize ROI With Marketing Technology, accessed 2026-09-02)
That utilization figure comes from one self-reported survey question: "Thinking about the totality of the capabilities made available by marketing technology, what percentage of those capabilities are being utilised by your company today?" Scott Brinker's objection to it is worth carrying: "How does one determine what percentage of their capabilities are being used? Is there any quantitative analysis being done here at all by respondents, or is it just gut feel?" (MarTech Square, April 5, 2026, independent commentary quoting the survey instrument)
Quote the 49% if you want, but quote it for what it is: a senior marketer's estimate of their own company, produced by asking them. You have an admin console and a seat-level usage report, which makes your number better than the survey's for the only stack you are responsible for.
Worth knowing who the headline numbers describe, too. Gartner's 2026 CMO Spend Survey drew on 401 CMOs and marketing leaders across North America, the UK and Europe, "with the vast majority of respondents reporting annual revenue of over $1 billion." (Gartner, May 11, 2026) Those are billion-dollar marketing organizations. The contract clauses in the rest of this paper describe everyone.
The partner's clock runs three years
HubSpot publishes what it pays its Solutions Partners. For deals sold on or after April 1, 2023, partners earn "20% commission" for "three years" from the sold date, applying to new customers, cross-sells, and upsells from Starter plans. A separate referral program pays 20% for one year on eligible Professional or Enterprise deals of $3,000 or more, restricted to top-tier partners holding CRM Implementation Accreditation. (HubSpot Partner Commissions FAQ, accessed 2026-09-02)
HubSpot prints that on its own marketing site, and it settles how long the recommending party's economics and yours overlap: three years. On a five-year platform decision, two of those years run with one participant in the recommendation unpaid.
I am citing HubSpot because HubSpot publishes it. I did not verify any other vendor's partner economics, the structures differ, and nothing here should be read as typical.
Ask any recommending party what they are paid, by whom, at what rate, and for how long. Ask in writing. For at least one major platform the answer is already public, which makes the question a competence test rather than an insult. A straight answer costs a good partner nothing. A refusal is itself the finding.
The contract answers what the demo cannot
Standard terms are published and free, and you can read them before the first call. Almost nothing in them comes up during one.
HubSpot's Customer Terms of Service, last updated April 14, 2026, provide as follows. (HubSpot Customer Terms of Service, accessed 2026-09-02)
| What it governs | The published clause |
|---|---|
| Renewal | "your subscription will automatically renew for the shorter of the same duration as your prior term or one year" |
| Renewal pricing | "Upon renewal, we may increase your Subscription Fees up to our then-current list price set out in our Product and Services Catalog," with at least thirty days' notice. No cap is stated. |
| Cancelling | The customer must "turn off the auto-renewal setting in the Account & Billing section of your HubSpot account prior to the end of your Current Term" |
| Shrinking | "You are not permitted to downgrade your Subscription Service during your Current Term." Downgrades take effect at the next renewal on a new order form. |
| Money back | "All payment obligations are non-cancelable and all amounts paid are non-refundable, except as specifically provided for in this Agreement." |
| The product changing | "We may modify the Subscription Service during the Subscription Term, including by adding or removing features, functions, Limits, or Add-Ons." Modifications may not "materially degrade the overall functionality"; where they do and go unremedied, the terms state termination with a pro-rated refund as the customer's sole remedy. |
| The counterparty changing | "We may assign this Agreement to any HubSpot Affiliate or in the event of merger, reorganization, sale of all or substantially all of our assets, change of control or operation of law." |
Those seven clauses describe the shape of one published subscription. On their face: price rises to list on thirty days' notice, footprint does not shrink mid-term, fees are non-refundable, features can be added or removed during the term, and the agreement can be assigned on a change of control. None of it is unusual or hidden, and none of it is in the deck.
The best hour you will spend on a martech purchase is the one reading the standard terms before the first demo. Read them first and they change what you ask for; read them after the redlines and they only tell you what you agreed to.
On a multi-year platform the no-mid-term-downgrade clause deserves more of the negotiation than the discount does. A seat count or contact tier you have not yet grown into is a one-way door for the length of the term, and headcount and list size both move down as well as up.
I am quoting what one vendor's published clauses say. What any of it means for your agreement is a different question: interpretation, enforceability, and the state-law rules on auto-renewal all belong with your own counsel, on your own paper.
The sticker price is not the price
HubSpot's Marketing Hub pricing page lists Professional at "Starts at $800/mo" with three seats and 2,000 marketing contacts, and Enterprise at "Starts at $3,600/mo" with five seats and 10,000 contacts. Both tiers carry a footnote: cost shown "does not include the required, one-time Professional Onboarding for a fee of $3,000," or "Enterprise Onboarding for a fee of $7,000." Credits run "$9.00 per 1,000 credits when you pay annually." (HubSpot Marketing Hub pricing, accessed 2026-09-02)
A required fee the pricing page excludes from the price is a published floor under implementation cost, before an internal team or a partner works a single hour. It is the only implementation number in this paper: there is no defensible public figure for what a real build costs, and I am not going to construct one from a proxy.
The same vendor publishes API rate limits by subscription tier for privately distributed apps: Free and Starter at 100 requests per 10 seconds and 250,000 a day; Professional at 190 per 10 seconds and 625,000 a day; Enterprise at 190 per 10 seconds and 1,000,000 a day. An API Limit Increase pack raises the burst to 250 per 10 seconds and adds a million calls a day "on top of your base subscription," with a maximum of two increases available. (HubSpot API usage guidelines and limits, accessed 2026-09-02)
API rate limits read as a technical specification and behave as a pricing tier. If the integration you are buying the platform for runs against a ceiling that scales with tier and caps at two add-on packs, your data architecture has a subscription tier. Price it during evaluation instead of discovering it during a backfill.
The SSO Wall of Shame tracks vendors that price single sign-on into a higher tier, on the stated position that "SSO is a core security requirement for any company with more than five employees," while allowing that increases under roughly 10% may reflect real maintenance cost. One dated entry, recorded February 2026, lists Railway moving from a $20 base to $2,000 for SSO. (sso.tax, accessed 2026-09-02. A community ledger, not vendor documentation; check any entry against that vendor's live pricing page before you rely on it.)
Three costs never make the business case and all three are published before purchase: mandatory onboarding, the security and governance features that live one tier up, and the integration ceiling. Nobody hides them. Almost nobody goes looking, because the evaluation gets scoped as a feature comparison instead of a total-cost model.
Four AI prices, four different units
Three vendors sell broadly comparable AI capability in units that cannot be compared with each other, and all three publish their prices.
Salesforce publishes three consumption models for Agentforce: Flex Credits at "$500 USD/Per 100k Credits," Conversations at "$2 USD/Per conversation," and an Agentforce User License at "$5 USD/User/Month" that requires Flex Credits. A standard Agentforce action consumes 20 Flex Credits and a Voice action consumes 30. (Salesforce Agentforce pricing, accessed 2026-09-02)
Intercom's Fin publishes outcome pricing: "$0.99 each" for Resolutions, Procedure handoffs and Disqualifications, "$9.99 each" for Qualifications, with a "50 outcomes per month minimum." A Resolution is defined on the page as: "No further help is requested after Fin's last answer." (fin.ai/pricing, accessed 2026-09-02)
HubSpot prices agent work in credits at $9.00 per 1,000 paid annually, with Customer Agent at "50 credits per conversation resolved" and Data Agent at "10 credits per smart properties run." (HubSpot Marketing Hub pricing, accessed 2026-09-02)
Fifty credits at $9.00 per thousand is $0.45 per resolved conversation. One Agentforce action at 20 Flex Credits is $0.10, and the same vendor's conversation model is $2.00. Fin's resolution is $0.99. Four published prices, for four units their own vendors do not define identically.
"Conversation," "action," "resolution" and "credit" are four different things, and the definition is where the money is. Intercom deserves credit for printing its own on the pricing page: a resolution is counted when no further help is requested after Fin's last answer. A customer who gives up and closes the tab has, by the billing definition, been resolved. Nobody is being tricked. It is a demonstration of what consumption pricing turns on, which is that the unit definition is a commercial term wearing technical clothes, and the only party who can check it against a real month of traffic is the one running the instance.
Buyers are adopting these models and then fighting them. Reporting on Gartner's 2026 CMO Spend Survey says 56% of respondents increased their martech allocation toward consumption-based pricing over the past year against 9% who decreased it, and that 50% of organizations using consumption-based solutions are "continually renegotiating contracts to avoid unexpected usage and cost spikes." (Chief Marketer, June 24, 2026)
Consumption pricing is sold as fairness, and it is fair, in one direction. It moves forecasting risk from the party holding a decade of usage data across thousands of accounts to the party holding none. In an agentic-pricing evaluation the rate is the smaller question. Instrument the unit before signature, and insist the contract define a billable event in words you can measure in your own logs.
The vendor you renew with may not be the one you bought
On May 17, 2026, Publicis Groupe announced an agreement to acquire LiveRamp for an enterprise value of $2.167 billion at $38.50 per share in cash, a 29.8% premium to LiveRamp's May 15 close, unanimously approved by both boards and expected to close before year-end 2026 subject to regulatory and shareholder approval. (Publicis Groupe, May 17, 2026) The deal was announced, not closed.
An agency holding company acquiring identity and data-collaboration infrastructure is what an assignment clause looks like from the outside. A buyer who chose a platform partly on independence grounds can end up with a different counterparty, and an assignment clause of the kind quoted above leaves them no vote in it.
Vendor growth models are disclosed too. Braze reported a dollar-based net retention rate of 109% across all customers and 110% for customers at $500,000 or more in ARR, for the trailing twelve months ended January 31, 2026. Braze defines the metric as current-period ARR from a cohort divided by that cohort's ARR twelve months prior, including expansion and net of contraction and attrition, excluding new customers. (Braze FY2026 results, March 24, 2026)
Net retention above 100% is a disclosed operating model rather than a scandal, and the useful way to read it is as the vendor's own published forecast of your next invoice. At renewal, what a vendor told its investors the account would be worth in twelve months carries more weight than the discount you won at signature.
Vendor failure is the risk an evaluation can price from public data, and it concentrates at the small end. Of the 1,367 products removed from the 2026 marketing technology landscape, nearly 80% had 50 or fewer employees and 45.5% sat in the $1 million to $10 million revenue band. (chiefmartec, May 5, 2026) The other path has the opposite shape and no comparable dataset behind it: the vendor succeeds, gets acquired, and the two clauses quoted earlier in this paper decide what happens next. Assignment needs no consent from you. There is no mid-term exit. A model that prices only the collapse has priced the half of the risk with the tidier numbers.
Who can answer which question
The artifact worth building is a short set of questions sorted by who can answer them. The internet has enough feature-comparison grids.
| Answerable in a demo | Answerable only in the contract | Answerable only in production |
|---|---|---|
| Does it do the thing | What can the price do at renewal, and is there a cap | What a real month of usage meters |
| How the interface feels | Can we shrink mid-term | Where the integration hits its ceiling |
| What the roadmap says | What happens on change of control | What breaks when the vendor changes the product |
| Which integrations exist | Can features be removed during the term | What an export contains |
| Reference customers | What is refundable | How long a real migration takes |
Everything in column two is free to check before signature, on the vendor's own legal pages.
Column three is the honest limit, because those answers live in the running instance. Gartner's own guidance on evaluating martech vendors is to "Skip traditional RFPs in favor of competitive proofs of concept (POCs) and real-world use-case testing." (Gartner, accessed 2026-09-02) A proof of concept on real data at real volume is the closest pre-purchase substitute for the year-four view, and the only part of an evaluation that produces evidence instead of assertions.
What to do with the advantage
The vantage buys four moves, and it is worth nothing if you spend the evaluation doing what everyone else in the room is doing.
Read the standard terms before the first demo, while they can still change what you ask for. Ask every recommending party what they are paid, by whom, at what rate, and for how long. Build the total-cost model out of the published floor items that never reach a business case: required onboarding, the tier where security and governance live, and the API ceiling your integration will run into. And in anything priced by consumption, get the billable unit defined in the contract in terms you can measure in your own logs, before the first surprising invoice rather than after it.
None of it requires inside information. It requires the one person in the room who will still be there in year four.
This paper describes what published contract terms and pricing pages say. It does not interpret what any agreement means, and it is not legal or financial advice. Terms, prices, and consumption units listed here move; each is dated to the day I read it, and you should verify anything you rely on against the vendor's own live document. Questions about your own contract belong with your own counsel.
The opinions here are mine and do not represent any employer or client, past or present. Every company named appears because of a public document — its own, or one filed about it. None was contacted, paid, or shown this paper before publication.
Drafted with AI assistance; researched, edited, and verified by Jennifer Gallo LeBlanc.
Sources
Vendor and firm documents, accessed 2026-09-02 unless dated otherwise.
- Gartner, Inc., Form 10-K, FY2025 (filed 2026-02-12). GTS sells to "users and providers of technology"; the independence statement and risk factor. https://www.sec.gov/Archives/edgar/data/749251/000074925126000112/it-20251231.htm
- Gartner, Flex Reprints Guidelines. The reprint program and the enumerated marketing asset types. https://www.gartner.com/en/about/policies/flex-reprints
- Forrester Research, Inc., Form 10-K, FY2025. Hi-Tech groups; Forrester Consulting includes "content marketing"; the objectivity statement. https://www.sec.gov/Archives/edgar/data/1023313/000119312526105114/forr-20251231.htm
- Gartner, Maximize ROI With Marketing Technology (Martech). 2025 survey: 49% of tools actively used, 15% high performers; the POC-over-RFP recommendation. https://www.gartner.com/en/marketing/topics/marketing-technology
- Gartner press release, May 11, 2026. 2026 CMO Spend Survey: 401 respondents, vast majority above $1B revenue. https://www.gartner.com/en/newsroom/press-releases/2026-05-11-gartner-2026-cmo-spend-survey-finds-cmos-allocate-15-point-3-percent-of-marketing-budgets-to-ai-but-only-30-percent-are-ready-to-scale-ai-capabilities
- HubSpot, Understanding Your Partner Commissions. 20% for three years; the referral program's terms. https://www.hubspot.com/partner-commissions-faq
- HubSpot, Customer Terms of Service, last updated April 14, 2026. Renewal, downgrade, refund, modification and assignment clauses. https://legal.hubspot.com/terms-of-service
- HubSpot, Marketing Hub pricing. Tier pricing, seat and contact allowances, required onboarding fees, credit rate, agent credit costs. https://www.hubspot.com/pricing/marketing
- HubSpot, API usage guidelines and limits. Per-tier rate limits; the API Limit Increase pack and its two-purchase maximum. https://developers.hubspot.com/docs/developer-tooling/platform/usage-guidelines
- Salesforce, Agentforce pricing. Flex Credits, per-conversation pricing, per-action credit consumption. https://www.salesforce.com/agentforce/pricing/
- Intercom / Fin, pricing. Outcome pricing, the 50-outcome minimum, and the published Resolution definition. https://fin.ai/pricing/
- Braze, Inc., fiscal year and fourth quarter 2026 results, March 24, 2026. Dollar-based net retention and the metric's definition. https://investors.braze.com/news/news-details/2026/Braze-Reports-Fiscal-Year-and-Fourth-Quarter-2026-Results/default.aspx
- Publicis Groupe press release, May 17, 2026. The announced LiveRamp acquisition, terms, premium, and closing conditions. https://www.globenewswire.com/news-release/2026/05/17/3296252/0/en/publicis-to-acquire-liveramp-to-accelerate-data-co-creation-for-smarter-agents.html
- chiefmartec / MartechTribe, 2026 Marketing Technology Landscape, May 5, 2026. The profile of the products removed from the landscape. https://chiefmartec.com/2026/05/2026-marketing-technology-landscape-supergraphic-peak-martech-achieved-maybe/
Secondary sources, used as marked and never as authority for a primary figure.
- Chief Marketer, June 24, 2026. Trade reporting of a client-gated Gartner report: consumption-pricing adoption and renegotiation behavior. https://www.chiefmarketer.com/gartner-cmo-spend-survey-budgets-reflect-increase-in-consumption-based-martech-paid-media-spend/
- MarTech Square, April 5, 2026. The verbatim utilization survey question and Scott Brinker's objection to it. https://martechsquare.substack.com/p/martech-utilisation-is-a-vanity-metric
Community-maintained, entries individually dated.
- sso.tax, The SSO Wall of Shame. The SSO-tax position and dated vendor entries. https://sso.tax/