AI and marketing for IT services companies: what a CMO needs to know in 2026
In 2026, AI changes marketing for IT services companies in three ways. Buyers use AI assistants to research and shortlist vendors. Your clients expect AI productivity to show up in price and outcomes. And your own team can produce far more analysis and content for the same budget. The CMOs who gain are the ones who fix their data and proof first, get named in AI answers, and keep a human in the loop wherever trust matters.
Take the CMO briefing to your leadership team.
A designed briefing for marketing heads: the numbers in one page, a 30-term glossary, a use-case and risk matrix, a 90-day plan, 15 questions to ask any AI vendor, an AI use policy starter and board talking points.
- One-page summary and the numbers, with sources
- Glossary, risk register and use-case matrix
- 90-day plan, vendor questions and a policy starter
Eight things a CMO should take from this guide.
- Buyers have moved. About half of B2B software buyers now start research with an AI chatbot more often than with Google, and most use one somewhere in the journey. Your first impression is increasingly an AI answer.
- Search is not dead, but clicks are thinner. Google still sends the large majority of referral traffic, yet more searches end without a click. Rankings alone no longer tell you how visible you are.
- Your clients are repricing you. Tech services buyers expect productivity gains and outcome-based pricing. Marketing has to prove outcomes, not describe capacity.
- Proof beats claims. Case studies with numbers, named experts and third-party reviews are what AI systems quote and what buyers verify.
- The upside is real. Used well, AI shortens research, drafting, localisation and reporting, and frees experts to do the thinking only they can do.
- The downside is real too. Wrong or generic content, leaked client data, failed agent projects, and new transparency rules all carry cost.
- Foundations decide results. Gartner finds CMOs buy AI tools faster than they build data, process and governance. The 30% who are ready spend more and gain more.
- Start small, measure, scale. A 90-day plan with three to five use cases, a policy and a scorecard beats a large programme with no owner.
Everything on this page is sourced and dated. The full list is at the end. Where a figure comes from a vendor or a forecast rather than a survey, we say so.
What is actually happening in AI and search.
Three shifts matter to a marketing head. The first is usage: AI assistants have become mass-market. At Google I/O in May 2026, Google reported more than 2.5 billion monthly users of AI Overviews, more than 1 billion monthly users of AI Mode, and more than 900 million monthly users of the Gemini app. Other assistants, including ChatGPT, Claude, Perplexity and Microsoft Copilot, add to that.
The second is behaviour. Buyers use these tools to research vendors, compare options and prepare for sales calls. The third is economics: AI is changing how technology services are priced and bought, which changes what your own marketing has to say.
| What | Figure | Source and date |
|---|---|---|
| B2B software buyers who start research with an AI chatbot more often than with Google | 51% | G2 AI search report, survey of 1,076 decision makers, March 2026 |
| B2B software buyers who rely on AI chatbots somewhere in research | 71%, up from about 60% seven months earlier | G2, March 2026 |
| Buyers who chose a different vendor than planned because of an AI chatbot | 69% | G2, March 2026 |
| B2B buyers who used generative AI in a recent purchase, mainly to gather vendor and product information | 45% | Gartner, 645 buyers, fielded Aug to Sep 2025, published May 2026 |
| B2B buyers who prefer to validate AI-generated insights with a sales rep | 69% | Gartner |
| B2B buyers who prefer a rep-free experience | 67% | Gartner, March 2026 |
| US Google searches that ended without a click | 68%, up from about 60% in 2024 | SparkToro, via Techwyse, Jan to Apr 2026, Similarweb panel |
| Lower click-through for the top result when an AI Overview appears | 58% lower, in a December 2025 sample | Ahrefs, 300,000 keywords, published Feb 2026 |
| Share of B2B website referral traffic from AI search | Under 2%, but about 35 times its March 2025 level | Octane11, 400 million sessions, Sep 2025 to Jun 2026 |
| Closed-won B2B deals with an AI search touch | About 1 in 11 | Octane11, same dataset |
Read these together. AI search drives a small share of visits today, but it influences a large share of decisions, because buyers use it early to decide who to look at. A vendor that is absent from the answer is absent from the shortlist, and the visit that never happened will not appear in your analytics. Gartner forecast in 2024 that traditional search engine volume would fall by 25% by 2026 as chatbots take queries. Google's own figures show AI features growing inside search rather than replacing it, so plan for both.
What this means for IT services buyers in practice.
IT services is a considered purchase: committees, long cycles and real risk. AI does not remove that. It changes the early and middle stages.
Research starts with a question, not a keyword
Buyers ask an assistant for "partners who can migrate SAP to S/4HANA for a mid-sized manufacturer" and get a short list with reasons. Pages that answer that question plainly are the ones quoted.
The shortlist forms inside the answer
G2 found AI chatbots are the top source shaping software shortlists, ahead of review sites, and that a third of buyers bought from a vendor they had never heard of before the chatbot named it. For services firms, that is an opening for mid-sized players, and a risk for incumbents who assumed brand was enough.
Buyers verify, then validate with people
Semrush data reported by eMarketer shows the most common workflow among US B2B professionals who use AI is to start with AI and then check the answer in traditional search. Gartner finds 69% still turn to sales reps to validate what AI told them. Your site, reviews and sales team must agree with each other.
Self-service is the default
Two thirds of buyers prefer a rep-free experience for much of the journey. Pricing models, proof, architecture notes and comparison pages that answer questions without a call carry more weight than before.
Buyer data in this guide comes mainly from software and B2B purchasing surveys. IT services purchases are larger and slower, so treat the percentages as direction, not as exact figures for your market. The shape of the behaviour is consistent.
The vocabulary, in plain language.
Your board, your agency and your vendors will use these terms. Here is what each means, and why a CMO should care. A longer glossary is in the briefing PDF.
| Term | What it means | Why it matters to you |
|---|---|---|
| Generative AI | Software that creates text, images, code or audio from a prompt, learned from large amounts of data. | The engine behind most tools your team and your buyers now use. |
| Large language model (LLM) | The model behind assistants such as ChatGPT, Claude and Gemini. It predicts useful text from patterns. | It can sound certain and be wrong. Facts about your firm must be easy for it to find. |
| AI Overviews and AI Mode | Google features that answer a query directly with AI, with links to sources. AI Mode is a fuller conversational search. | They take clicks from the top of results. Being cited in them is a new goal. |
| GEO, AEO, LLMO | Generative engine optimisation, answer engine optimisation, LLM optimisation. Names for making your content easy to cite in AI answers. | Overlaps heavily with good SEO, with more weight on clarity, evidence and third-party mentions. See our GEO guide. |
| Zero-click search | A search that ends on the results page without a visit to any site. | Falling clicks do not always mean falling interest. Measure visibility as well as visits. |
| Agent and agentic AI | AI that takes multi-step actions toward a goal, using tools, rather than only answering. | Promising, and heavily hyped. Gartner expects over 40% of agentic AI projects to be cancelled by the end of 2027. |
| Agent washing | Vendors relabelling ordinary chatbots or automation as agents. | Gartner estimates only around 130 of the thousands of vendors claiming agentic products offer real agentic features. Ask for a demo on your data. |
| RAG (retrieval-augmented generation) | A method where the model looks up your own documents before it answers. | The standard way to make an internal assistant accurate about your firm and its work. |
| Hallucination | A confident answer that is false or invented. | The main reason for human review of anything customer-facing. |
| Prompt and context window | The instruction you give a model, and how much text it can consider at once. | Prompt quality and the material you supply decide output quality more than the tool brand. |
| MCP (Model Context Protocol) | An open standard for connecting AI assistants to tools and data sources. | Lets agents work inside your CRM, analytics and content systems. Raises access and security questions. |
| Grounding and citations | Tying an AI answer to named sources. | Citations are how AI answers send visibility to you. Be the source. |
| AI slop | Low-effort, generic AI-made content published at scale. | Buyers and search engines discount it. Google's spam policy on scaled content abuse covers it. |
| Shadow AI | Staff using unapproved AI tools with company or client data. | A confidentiality and compliance risk. Fix it with approved tools and a clear policy. |
| Human in the loop | A person reviews or approves AI output before it is used. | The control that makes AI acceptable for client-facing work. |
| Content provenance | Labels and metadata showing whether content is AI-generated or edited. | Starting to matter under the EU AI Act transparency rules. |
| llms.txt | A proposed plain-text file that points AI systems at key pages. | Support from the major AI platforms is unconfirmed. Low cost, low proof of effect. |
| AI SDR | AI that researches prospects and writes or sends outreach. | Can lift volume, and can wreck deliverability and brand. See our email deliverability guide. |
Your clients are using AI on you. That changes what marketing must say.
Most technology services firms face a second effect that other industries do not: AI changes the product they sell. Bain's January 2026 survey of 280 technology services decision makers found that clients expect IT budgets to stay roughly flat while services spend rises modestly, that about three quarters expect 5% to 10% of technology spend to go to AI and machine learning, and that they expect measurable productivity gains in areas such as developer productivity, testing and contact centre automation. Bain expects that to push clients toward lower delivery costs, faster timelines and outcome-based pricing, and advises vendors to revise their pricing before clients demand it.
The financial results of the large Indian services firms for FY26 show the pressure. As reported by The Register, TCS revenue was down 0.5%, Wipro up 4% and HCLTech up 11.2%, and executives at several firms used the words "deflation" and "degrowth" when describing AI's effect on pricing. HCLTech's CEO guided to a possible 3% to 5% revenue dip from AI deflation. Analyst estimates reported by Outlook Business put the productivity benefits shared with clients on application services at 10% to 15%, though firms do not disclose deal terms, so treat that as an estimate.
What this means for marketing:
- Lead with outcomes. Replace capacity language (headcount, years of experience, delivery centres) with measured results: cycle time cut, cost per transaction, defects avoided, revenue protected.
- Show how you use AI in delivery. Name the practices: where AI writes tests, accelerates migration, or runs support, how it is reviewed, and what it changed for a named client. Vague "AI-powered" claims are discounted.
- Explain your commercial model. If you offer outcome-based or hybrid pricing, say so on the page. Buyers will ask for it anyway.
- Make proof self-serve. Case studies with numbers, architecture notes, and reference-ready clients, easy to find without a call. See our lead-to-SAL ratio guide for measuring what marketing contributes.
- Sharpen your niche. Gartner-style category lists, industry pages and platform specialisms matter more when a buyer's AI assistant asks "who is strongest at X for Y". Breadth without evidence loses.
- Tell the talent story. The hardest skills to source, per Bain, are cybersecurity, AI and ML engineering and data science. Employer brand is part of how buyers judge capacity.
Where AI genuinely helps marketing at an IT services company.
The strongest uses do not replace expertise. They remove the slow, repetitive work around it. Here are the ones we see paying off, with the check that keeps each safe.
| Use case | What you gain | Human check |
|---|---|---|
| Research and synthesis: analyst reports, competitor sites, call transcripts, support themes | Days of reading become hours. More decisions grounded in evidence. | Verify every figure against the source before it is used. |
| First drafts of articles, case studies and sales notes from expert interviews | Experts talk for 30 minutes instead of writing for six hours. | Expert edits for accuracy. Editor edits for voice. No unreviewed publication. |
| Repurposing: a guide into posts, emails, a webinar outline, a one-page brief | One idea reaches many channels without starting from zero. | Check that claims stay accurate in the short form. |
| SEO and GEO operations: page audits, schema, internal links, FAQ drafting, entity consistency | Large sites are fixed faster. Structured, citeable pages. | Review for accuracy. Never publish scaled pages created only for rankings. |
| Localisation and multi-language pages | New markets reached faster, as in our Coinmen Korean-language SEO work. | Native-speaker review for terminology and tone. |
| Account-based personalisation: tailored landing pages, outreach angles, briefs for sellers | Relevance at a scale a small team cannot hand-write. | Keep volume within deliverability limits. Honest, specific, never invented facts. |
| Sales enablement: call summaries, proposal first drafts, RFP answer libraries | Sellers spend time on buyers, not on documents. | Review for client confidentiality and pricing accuracy. |
| Analytics: anomaly alerts, attribution checks, forecast scenarios, monthly report drafting | Faster insight, fewer missed breaks in the funnel. | Analysts confirm the numbers. The model explains, it does not own the report. |
| Campaign operations: ad variants, audience tests, creative resizing | More tests per month at the same cost. | Brand and compliance review for every live variant. |
The test for any use case: if a wrong answer would embarrass a client or breach a contract, a person must review it. If it only costs a little time, automate more freely.
The negative side, and how to contain it.
Most AI marketing failures are predictable. These are the ones to plan for before they happen.
Wrong or generic content
Models invent facts, and unreviewed content sounds like everyone else. Buyers notice, and Google's spam policies, clarified in May 2026 to cover AI Overviews and AI Mode, target scaled content made to manipulate rankings. Fix: expert input, named authors, review, and fewer, better pages.
Trust and authenticity
Gartner found 51% of buyers think they are more likely to meet misleading information from generative AI, against 49% for sales reps. Fix: show your sources, your people and your dated results.
Confidentiality and shadow AI
Staff pasting client details into public tools can breach contracts. Fix: approved tools, data rules, and training. Never put client names or confidential detail into an unapproved system.
Agent washing and failed projects
Gartner expects over 40% of agentic AI projects to be cancelled by end of 2027, and few vendors offer real agent capability. Fix: pilot on your data, with success measures agreed in advance.
Misrepresentation by AI answers
An assistant may describe your firm wrongly, or cite an outdated source. Fix: monthly checks, a clear and consistent public profile, and corrections through the sources the model cites.
Measurement gaps
AI answers and zero-click searches hide influence from analytics. Fix: add "how did you hear about us" fields, track branded search, and test AI answers each month.
Outreach and deliverability
AI makes mass personalised email cheap, and mailbox providers punish spam complaints and generic bulk. Fix: smaller, better targeted sends and strict authentication. See our deliverability guide.
Regulation and disclosure
EU AI Act transparency obligations apply from 2 August 2026, covering deepfake disclosure and chatbot notices, with some machine-readable marking duties for existing systems delayed to 2 December 2026 under the AI omnibus. Draft guidance is still evolving. Fix: ask counsel, log where you use AI, and label synthetic media and chatbots. This guide is not legal advice.
Tool sprawl and cost
Gartner finds CMOs spend an average of 15.3% of marketing budget on AI, while only 30% say their AI capability is mature. Fix: consolidate tools, name owners, and cut anything without a measured result.
How to appear in the answers your buyers read.
Getting named by an assistant depends on the same things that make a firm credible to a human buyer, written so a machine can quote them. Studies of which sources AI platforms cite, summarised by Contently, found LinkedIn the most-cited domain for professional queries across six AI platforms in a Profound analysis of 1.4 million citations, and found that brands present on several review sites were cited more often than brands with none. These are third-party studies with their own methods, but the direction is consistent: AI systems lean on sources your buyers already trust.
Answer the buyer's question first
Put a two-sentence answer at the top of each service, industry and platform page. Name the service, the buyer and the proof.
Make your facts consistent everywhere
Same name, services, locations, founding date and tagline on your site, LinkedIn, directories and review profiles. Inconsistency produces wrong answers.
Publish evidence, not adjectives
Case studies with figures and the client's context, dated and named where permitted. See our case studies for the pattern.
Earn third-party presence
Review sites, analyst and award listings, expert commentary in trade press, and an active LinkedIn page with substantive posts. These are the pages assistants cite.
Keep the technical basics open
Let reputable AI crawlers read your pages, use clean headings and structured data, keep pages fast, and avoid hiding key content behind scripts or forms.
Test monthly
Ask the assistants your buyers use the questions your buyers ask. Record whether you are named, how you are described and which sources are cited, then fix gaps.
For the full method, read our GEO guide for technology companies and our SEO guide for IT services companies.
Data, process, people and governance come before tools.
Gartner's 2026 CMO Spend Survey of 401 marketing leaders found that 70% call becoming an AI leader a critical goal this year, yet 70% say their internal processes are not mature enough to implement and scale AI, and only 30% report mature AI readiness. AI-ready organisations put 21.3% of their budgets toward AI, against 15.3% on average. Gartner's analyst warns that CMOs invest in tools faster than they build the data foundations. Marketing budgets are 7.8% of revenue on average, so there is little new money. Most gains have to come from redirecting effort.
- Data. One clean source for customers, accounts, content and results. AI that reads messy data produces confident nonsense.
- Process. Written workflows for content, approvals, campaign launch and reporting. AI speeds up a process. It does not repair a missing one.
- People. An owner for AI in marketing, a small group of trained champions, and time set aside to learn. Hire for editorial judgement and analytics, not for prompt tricks.
- Policy. A one-page AI use policy: approved tools, data that may never be entered, review rules, disclosure rules, and who signs off. A starter is in the briefing.
- Vendors. Contracts that cover data use, model training on your data, confidentiality, security review and exit terms. Ask for evidence, not demos on their data.
- Measurement. A monthly scorecard tying AI use to leads sales accepts, cost per asset and cycle time.
What to do in the next three months.
Baseline and rules
Audit how AI is already used in the team, including shadow tools. Write the AI use policy. Pick three to five use cases from the table above that are low risk and high effort today. Record baselines: time per asset, leads sales accepts, and what AI assistants currently say about your firm.
Pilot and fix the foundation
Run each use case with named owners and review steps. Fix the public facts and proof the assistants rely on: service pages, case studies, LinkedIn, review profiles. Start the monthly AI answer test.
Decide and scale
Compare results with baselines. Stop what did not work, keep what did, and write the second wave. Report to the board in one page: results, risks, cost and next quarter. Fold the learning into the marketing plan.
If you have no marketing plan yet to anchor this, start with our B2B digital marketing plan, then add the AI use cases to it.
What to measure, so AI is judged by results.
| Measure | Why it matters | How |
|---|---|---|
| Share of AI answers naming you | Early signal of shortlist presence | Monthly test of 20 to 30 buyer questions across the main assistants |
| Lead-to-SAL ratio | The share of leads sales accepts, the number that proves value | CRM, by source. See our guide. |
| Branded and direct search trend | AI exposure often shows up as people searching your name later | Search Console and analytics |
| "How did you hear about us" mentions of AI | Catches influence analytics cannot see | Free-text field on forms, tagged monthly |
| Cost and time per asset | Shows productivity gains honestly | Hours and spend per article, case study, campaign |
| Error and correction rate | Keeps quality visible | Count of factual corrections after review |
Report fewer measures, consistently. A one-page scorecard each month beats a dashboard no one reads.
Choosing AI tools and agencies.
Start from the use case, not the tool. Most CMOs end up with a mix: a few well-chosen tools for production and analytics, internal know-how for the editorial and data work, and a specialist partner where the skill is scarce, such as GEO and technical SEO. Before you sign anything, ask these six questions:
- Show it on my data. A demo on the vendor's examples proves nothing.
- What happens to our data? Storage, training on your inputs, retention, and who can see it.
- How is quality reviewed? Who checks output, what the error rate is, how mistakes are corrected.
- What is genuinely automated? Ask which steps are agents and which are scripts or manual work. This is where agent washing hides.
- How will we know it worked? Measures and baselines agreed before the start.
- How do we leave? Export, contract exit and what you keep.
The briefing PDF has 15 questions and the red flags to watch for.
Where the numbers come from.
Figures are from the sources below, as published on the dates shown. Survey results describe the samples named and may not match your market. Forecasts are forecasts.
- Google, I/O 2026 keynote summary (May 2026): AI Overviews, AI Mode and Gemini monthly users.
- G2, AI search insight report (March 2026 survey of 1,076 B2B decision makers).
- Gartner, buyers validate AI insights with sales reps (May 2026; 645 buyers, fielded Aug to Sep 2025).
- Gartner, 67% of B2B buyers prefer a rep-free experience (March 2026).
- Gartner, 2026 CMO Spend Survey (May 2026; 401 CMOs, fielded Jan to Mar 2026).
- Gartner on agentic AI project cancellations, via MarTech (forecast first published June 2025).
- Gartner, search engine volume forecast (February 2024; a forecast, not a measurement).
- eMarketer, B2B buyers start with AI, then check its work (September 2026, citing Semrush data from March to April 2026).
- SparkToro zero-click analysis, via Techwyse (June 2026; Jan to Apr 2026 Similarweb data).
- Ahrefs, AI Overviews and click-through (February 2026; December 2025 data).
- Octane11, State of B2B AI Search Q2 2026 (July 2026; 400 million sessions).
- Bain, tech services buyer survey (April 2026; January 2026 survey of 280 decision makers).
- The Register, Indian IT FY26 results (April 2026).
- Outlook Business, outcome-based deals in IT services (analyst estimates cited from Jefferies).
- Contently, top sources LLMs cite (April 2026; summarising Profound, SEMrush and SE Ranking studies).
- PPC Land, Google spam policies and AI Overviews (Google documentation update of 15 May 2026).
- Baker McKenzie, EU AI transparency from 2 August 2026 (2026).
Take the CMO briefing to your leadership team.
A designed briefing for marketing heads: the numbers in one page, a 30-term glossary, a use-case and risk matrix, a 90-day plan, 15 questions to ask any AI vendor, an AI use policy starter and board talking points.
- One-page summary and the numbers, with sources
- Glossary, risk register and use-case matrix
- 90-day plan, vendor questions and a policy starter
Algoworks
Want a second pair of eyes on this?
Book a pipeline review. We will look at how your funnel works today and show you what we would change first.
- A 30-minute callYou describe your funnel and where it leaks.
- A look at your funnel dataRead-only access to analytics and your nurture platform.
- First findingsWhat we would change first, and what it should move.